<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Sidharth Bangera</title>
    <description>The latest articles on DEV Community by Sidharth Bangera (@sidharth_bangera).</description>
    <link>https://dev.to/sidharth_bangera</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4167066%2F2504cd58-3050-4161-95b8-d7e57c7c1685.jpg</url>
      <title>DEV Community: Sidharth Bangera</title>
      <link>https://dev.to/sidharth_bangera</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/sidharth_bangera"/>
    <language>en</language>
    <item>
      <title>Running a DynamoDB GUI Without a Mouse: Testing Keyboard and Screen-Reader Accessibility</title>
      <dc:creator>Sidharth Bangera</dc:creator>
      <pubDate>Thu, 12 Nov 2026 13:31:15 +0000</pubDate>
      <link>https://dev.to/sidharth_bangera/running-a-dynamodb-gui-without-a-mouse-testing-keyboard-and-screen-reader-accessibility-4g12</link>
      <guid>https://dev.to/sidharth_bangera/running-a-dynamodb-gui-without-a-mouse-testing-keyboard-and-screen-reader-accessibility-4g12</guid>
      <description>&lt;h2&gt;
  
  
  Introduction
&lt;/h2&gt;

&lt;p&gt;A database GUI can make working with DynamoDB easier, but can you still complete everyday operations without using a mouse?&lt;/p&gt;

&lt;p&gt;In this walkthrough, I tested &lt;a href="https://tables.serverlesscreed.com/" rel="noopener noreferrer"&gt;&lt;strong&gt;Tables by ServerlessCreed&lt;/strong&gt;&lt;/a&gt;, a desktop DynamoDB GUI, using keyboard navigation and Windows Narrator.&lt;/p&gt;

&lt;p&gt;The goal was to open a table, execute a Scan, inspect and edit an item, and export results without relying on the mouse. I also tested platform shortcuts, focus behavior, screen-reader announcements, and recovery from a network failure.&lt;/p&gt;

&lt;h2&gt;
  
  
  Prerequisites
&lt;/h2&gt;

&lt;p&gt;To reproduce this workflow, you need:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;a href="https://tables.serverlesscreed.com/downloads" rel="noopener noreferrer"&gt;Tables by ServerlessCreed&lt;/a&gt; installed.&lt;/li&gt;
&lt;li&gt;Access to an approved, non-production DynamoDB environment.&lt;/li&gt;
&lt;li&gt;A configured AWS connection with suitable read/write permissions.&lt;/li&gt;
&lt;li&gt;A keyboard and Windows Narrator or another screen reader.&lt;/li&gt;
&lt;li&gt;A disposable test item for editing operations.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;For this test, I used:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Component&lt;/th&gt;
&lt;th&gt;Configuration&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Tables version&lt;/td&gt;
&lt;td&gt;v3.3.42&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Operating system&lt;/td&gt;
&lt;td&gt;Windows 11&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;AWS Region&lt;/td&gt;
&lt;td&gt;&lt;code&gt;us-east-1&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Screen reader&lt;/td&gt;
&lt;td&gt;Windows Narrator&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Test tables&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;Composite_Scan_500&lt;/code&gt;, &lt;code&gt;DemoTable&lt;/code&gt;, &lt;code&gt;ABC-EFG&lt;/code&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Testing approach&lt;/td&gt;
&lt;td&gt;Keyboard navigation, screen-reader inspection, scaling, and error recovery&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  1. Testing platform keyboard shortcuts
&lt;/h2&gt;

&lt;p&gt;I started by opening the Keyboard section under Preferences &amp;amp; Application Lifecycle.&lt;/p&gt;

&lt;p&gt;Tables documented five Windows shortcuts:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Shortcut&lt;/th&gt;
&lt;th&gt;Action&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;Ctrl+K&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Open command palette&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;Ctrl+B&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Toggle table sidebar&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;Ctrl+I&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Toggle item inspector&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;Ctrl+Enter&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Run active query&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;Esc&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Close active surface&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;All five worked during the initial testing.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F6y6xaob5e3h111jl5bld.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F6y6xaob5e3h111jl5bld.png" alt="Tables Keyboard preferences showing five documented platform shortcuts and their scopes." width="800" height="301"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;Figure 1 — The five documented platform shortcuts in Tables.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;I then opened the PartiQL editor and entered:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight sql"&gt;&lt;code&gt;&lt;span class="k"&gt;SELECT&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt; &lt;span class="k"&gt;FROM&lt;/span&gt; &lt;span class="s2"&gt;"Composite_Scan_500"&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;With the editor focused, pressing &lt;code&gt;Ctrl+K&lt;/code&gt; opened the command palette. Pressing &lt;code&gt;Esc&lt;/code&gt; closed it, while &lt;code&gt;Ctrl+B&lt;/code&gt; toggled the sidebar.&lt;/p&gt;

&lt;p&gt;The SQL statement remained unchanged throughout these interactions.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fivka0342pylspp3h4iet.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fivka0342pylspp3h4iet.png" alt=" " width="800" height="264"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fbfq6yeqzhvltt8slifhe.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fbfq6yeqzhvltt8slifhe.png" alt="Tables command palette opened over the PartiQL Workbench." width="800" height="329"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;Figure 2 — The command palette opens without altering the SQL statement.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;This confirmed that the tested global shortcuts did not interfere with editor contents. However, a deliberate reassignment conflict was not tested because no shortcut-remapping option was available in the inspected settings.&lt;/p&gt;

&lt;h2&gt;
  
  
  2. Navigating with Windows Narrator
&lt;/h2&gt;

&lt;p&gt;Next, I enabled Windows Narrator using &lt;code&gt;Win + Ctrl + Enter&lt;/code&gt; and navigated through Tables using &lt;code&gt;Tab&lt;/code&gt;, &lt;code&gt;Shift+Tab&lt;/code&gt;, and &lt;code&gt;Enter&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;The Settings menu was reachable without a mouse.&lt;/p&gt;

&lt;p&gt;Narrator identified it as an &lt;strong&gt;Open Settings Menu button&lt;/strong&gt; and announced its menu items, including:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Preferences &amp;amp; Lifecycle, menu item 1 of 3.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fpu0cec7lb7n5lbebdkka.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fpu0cec7lb7n5lbebdkka.png" alt="Settings menu opened with a visible keyboard focus outline." width="800" height="307"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;Figure 3 — Keyboard focus on the Settings menu while Windows Narrator identifies the available controls.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Pressing &lt;code&gt;Esc&lt;/code&gt; closed the menu successfully. However, focus restoration was occasionally unclear, with the visible focus indicator appearing around the Settings tooltip or an empty rectangular area.&lt;/p&gt;

&lt;p&gt;I then navigated to &lt;code&gt;DemoTable&lt;/code&gt; and opened it using &lt;code&gt;Enter&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;Narrator announced the table name, active status, and item count.&lt;/p&gt;

&lt;p&gt;This demonstrated that the main table-selection workflow was accessible through keyboard navigation.&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Running a Scan, editing an item, and exporting data
&lt;/h2&gt;

&lt;p&gt;With &lt;code&gt;DemoTable&lt;/code&gt; open, I navigated to the &lt;strong&gt;Run&lt;/strong&gt; button.&lt;/p&gt;

&lt;p&gt;Narrator correctly announced “Run button,” and pressing &lt;code&gt;Enter&lt;/code&gt; executed a read-only Scan.&lt;/p&gt;

&lt;p&gt;Tables returned the four existing records.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Flxem0fs7a9v42ah5kve9.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Flxem0fs7a9v42ah5kve9.png" alt="Run button focused in DemoTable, with four records displayed." width="800" height="256"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;Figure 4 — Executing a DynamoDB Scan through keyboard navigation.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Although the Scan succeeded, Narrator did not announce a distinct completion message or returned-item count.&lt;/p&gt;

&lt;h3&gt;
  
  
  Editing a disposable item
&lt;/h3&gt;

&lt;p&gt;Next, I opened the Create Item dialog using the keyboard.&lt;/p&gt;

&lt;p&gt;I created a disposable item with &lt;code&gt;userID = 106&lt;/code&gt; and &lt;code&gt;Status = Active&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;After verifying its creation, I opened the item editor and changed its primary key from &lt;code&gt;106&lt;/code&gt; to &lt;code&gt;150&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;Tables displayed a warning explaining that changing a primary key recreates the item.&lt;/p&gt;

&lt;p&gt;After saving, the updated record appeared under &lt;code&gt;userID = 150&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F68vurxweoli8tenhhs3d.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F68vurxweoli8tenhhs3d.png" alt=" " width="800" height="277"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fidd1r8oezyxrn8mm2d0n.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fidd1r8oezyxrn8mm2d0n.png" alt=" " width="800" height="356"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fa5g1uio03gh5b017qck0.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fa5g1uio03gh5b017qck0.png" alt="DemoTable showing the disposable record with userID 150 after editing." width="800" height="370"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Figure 5 — The disposable item after changing its primary key through the keyboard-accessible editor.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;During item inspection, I also encountered two limitations: I could not deselect a selected item using the attempted &lt;code&gt;Enter&lt;/code&gt; or &lt;code&gt;Space&lt;/code&gt; interactions, and &lt;code&gt;Esc&lt;/code&gt; did not close the persistent Item Inspector.&lt;/p&gt;

&lt;h3&gt;
  
  
  Exporting the results
&lt;/h3&gt;

&lt;p&gt;Finally, I navigated to the Export option using the keyboard, which opened the Data Transfer Center in Tables.&lt;/p&gt;

&lt;p&gt;I selected JSON export for the loaded results, navigated to &lt;strong&gt;Start export&lt;/strong&gt;, and completed the Windows Save dialog without using the mouse.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fvo97k5duuvap9gakyaif.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fvo97k5duuvap9gakyaif.png" alt="Tables Data Transfer Center showing JSON export options for six loaded records." width="800" height="374"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Figure 6 — Exporting loaded DynamoDB results to a local JSON file through keyboard navigation.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;The exported file was independently opened in VS Code and contained six records.&lt;/p&gt;

&lt;p&gt;Tables indicated that the export used already loaded results without requiring an additional DynamoDB read.&lt;/p&gt;

&lt;h2&gt;
  
  
  4. Empty states, errors, and recovery
&lt;/h2&gt;

&lt;p&gt;I also tested how Tables communicated an empty table and a network failure.&lt;/p&gt;

&lt;p&gt;Opening &lt;code&gt;ABC-EFG&lt;/code&gt;, an empty table, displayed a message explaining that no items were present.&lt;/p&gt;

&lt;p&gt;Narrator announced the table name and zero-item count, although the central empty-state description was not announced during the observed interaction.&lt;/p&gt;

&lt;h3&gt;
  
  
  Network failure
&lt;/h3&gt;

&lt;p&gt;I deliberately disconnected the internet and attempted a read-only Scan.&lt;/p&gt;

&lt;p&gt;Tables displayed a DNS lookup failure along with a network-error notification.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fbcz7vn4fnoai7tzk5m3x.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fbcz7vn4fnoai7tzk5m3x.png" alt="Tables displaying a red DNS lookup failure message after a read-only Scan." width="799" height="429"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Figure 7 — Tables displays a network-related error, which Windows Narrator successfully announces.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Unlike the successful Scan completion, the network error was announced by Narrator.&lt;/p&gt;

&lt;p&gt;After reconnecting, the error disappeared and the table returned to its normal empty state. No separate recovery announcement was observed.&lt;/p&gt;

&lt;p&gt;The failure was caused by the deliberately disconnected network, not an AWS permissions issue.&lt;/p&gt;

&lt;h2&gt;
  
  
  5. Accessibility Findings and Observations
&lt;/h2&gt;

&lt;p&gt;The testing demonstrated that Tables supports several essential keyboard-driven operations, including table navigation, query execution, item editing, and data export. Windows Narrator also identified key interface controls and announced network errors.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Area&lt;/th&gt;
&lt;th&gt;Observed result&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Platform shortcuts&lt;/td&gt;
&lt;td&gt;All five documented shortcuts worked as expected&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Keyboard navigation&lt;/td&gt;
&lt;td&gt;Application controls and tables were accessible using the keyboard&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Scan execution&lt;/td&gt;
&lt;td&gt;Successfully executed using Enter&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Item creation and editing&lt;/td&gt;
&lt;td&gt;Completed using keyboard navigation&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;JSON export&lt;/td&gt;
&lt;td&gt;Successfully exported and independently verified&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Screen-reader support&lt;/td&gt;
&lt;td&gt;Control names, table information, and network errors were announced&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Item selection&lt;/td&gt;
&lt;td&gt;The selected item could not be deselected using the attempted Enter or Space interactions&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Focus restoration&lt;/td&gt;
&lt;td&gt;Occasionally unclear after closing the Settings menu&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Status announcements&lt;/td&gt;
&lt;td&gt;Scan completion and successful network recovery were not explicitly announced&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Overall, the core workflow was successfully completed without relying on a mouse. The results also highlighted areas where keyboard interaction and screen-reader feedback could be more consistent.&lt;/p&gt;

&lt;p&gt;In particular, predictable selection behavior, clear focus indication, and meaningful announcements after an operation can help users navigate the application more confidently.&lt;/p&gt;

&lt;p&gt;These observations reflect the behavior encountered during testing and identify opportunities for further accessibility improvements.&lt;/p&gt;

&lt;h2&gt;
  
  
  Practical takeaways
&lt;/h2&gt;

&lt;p&gt;This testing demonstrated that a core DynamoDB workflow in Tables can be completed through keyboard navigation.&lt;/p&gt;

&lt;p&gt;I could open a table, execute a Scan, create and edit a disposable record, and export the results without relying on a mouse.&lt;/p&gt;

&lt;p&gt;The platform shortcuts also worked, including opening the command palette while preserving the contents of the PartiQL editor.&lt;/p&gt;

&lt;p&gt;However, accessibility involves more than activating controls. Predictable focus behavior, item deselection, dialog dismissal, and meaningful screen-reader announcements are equally important.&lt;/p&gt;

&lt;p&gt;The network failure test illustrated this distinction particularly well: Tables displayed the error, Narrator announced it, and the interface returned to a usable state after connectivity was restored.&lt;/p&gt;

&lt;p&gt;For me, the main takeaway is that &lt;strong&gt;a database GUI should make every step understandable, not just executable, without a mouse&lt;/strong&gt;.&lt;/p&gt;

&lt;p&gt;Costs and limits: No additional Tables-specific costs were measured during this test. The small number of DynamoDB reads and writes remained subject to standard AWS pricing, while the export of already loaded results required no additional DynamoDB reads. Testing was limited to Windows 11 and Tables v3.3.42.&lt;/p&gt;

&lt;p&gt;For further details, see &lt;a href="https://aws.amazon.com/dynamodb/pricing/" rel="noopener noreferrer"&gt;DynamoDB pricing&lt;/a&gt; and &lt;a href="https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/Scan.html" rel="noopener noreferrer"&gt;Scanning tables in DynamoDB&lt;/a&gt;.&lt;/p&gt;




&lt;p&gt;&lt;strong&gt;Application:&lt;/strong&gt; &lt;a href="https://tables.serverlesscreed.com/" rel="noopener noreferrer"&gt;Tables by ServerlessCreed&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Download:&lt;/strong&gt; &lt;a href="https://tables.serverlesscreed.com/downloads" rel="noopener noreferrer"&gt;Official Tables downloads&lt;/a&gt;&lt;/p&gt;

</description>
      <category>a11y</category>
      <category>aws</category>
      <category>database</category>
      <category>testing</category>
    </item>
    <item>
      <title>Securing DynamoDB Access for AI Agents: Testing MCP, OAuth, and Least-Privilege Controls in Tables</title>
      <dc:creator>Sidharth Bangera</dc:creator>
      <pubDate>Wed, 11 Nov 2026 17:12:02 +0000</pubDate>
      <link>https://dev.to/sidharth_bangera/securing-dynamodb-access-for-ai-agents-testing-mcp-oauth-and-least-privilege-controls-in-tables-3102</link>
      <guid>https://dev.to/sidharth_bangera/securing-dynamodb-access-for-ai-agents-testing-mcp-oauth-and-least-privilege-controls-in-tables-3102</guid>
      <description>&lt;h2&gt;
  
  
  Introduction
&lt;/h2&gt;

&lt;p&gt;When I connect an AI agent to a database, I want to control which AWS connection it can use, what operations it can perform, and whether an unauthorized request can reach the database.&lt;/p&gt;

&lt;p&gt;In this walkthrough, I use &lt;a href="https://tables.serverlesscreed.com/" rel="noopener noreferrer"&gt;&lt;strong&gt;Tables by ServerlessCreed&lt;/strong&gt;&lt;/a&gt;, a DynamoDB GUI with a built-in MCP server, to connect Codex CLI to a non-production DynamoDB environment and test the access controls around that connection.&lt;/p&gt;

&lt;p&gt;The workflow covers starting a local MCP server, exposing a single AWS profile, completing OAuth authorization, retrieving DynamoDB table metadata, and checking what happens when a request is made without authentication.&lt;/p&gt;

&lt;p&gt;I also verify the server's network binding independently using PowerShell. This matters because a successful connection alone does not establish whether the server is listening only on the local machine.&lt;/p&gt;

&lt;h2&gt;
  
  
  Prerequisites
&lt;/h2&gt;

&lt;p&gt;To reproduce the workflow, you need:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;a href="https://tables.serverlesscreed.com/" rel="noopener noreferrer"&gt;Tables by ServerlessCreed&lt;/a&gt; installed. The installer is available from the &lt;a href="https://tables.serverlesscreed.com/downloads" rel="noopener noreferrer"&gt;official download page&lt;/a&gt;.&lt;/li&gt;
&lt;li&gt;Access to an approved, non-production DynamoDB environment.&lt;/li&gt;
&lt;li&gt;At least one AWS connection configured in Tables, preferably with read-only permissions.&lt;/li&gt;
&lt;li&gt;An MCP-compatible AI client, such as Codex CLI.&lt;/li&gt;
&lt;li&gt;PowerShell or another tool for verifying local network connections.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;For this test, I used:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Component&lt;/th&gt;
&lt;th&gt;Configuration&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;AWS profile&lt;/td&gt;
&lt;td&gt;&lt;code&gt;read-only&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Region&lt;/td&gt;
&lt;td&gt;&lt;code&gt;us-east-1&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;MCP transport&lt;/td&gt;
&lt;td&gt;OAuth-protected Streamable HTTP&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;MCP endpoint&lt;/td&gt;
&lt;td&gt;&lt;code&gt;http://127.0.0.1:61041/mcp&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;MCP client&lt;/td&gt;
&lt;td&gt;Codex CLI &lt;code&gt;0.167.1&lt;/code&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Client model&lt;/td&gt;
&lt;td&gt;GPT-6-Luna&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;OAuth consent&lt;/td&gt;
&lt;td&gt;Read only&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;The endpoint port was assigned during the test and may differ in another session.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Starting the MCP server and verifying local access
&lt;/h2&gt;

&lt;p&gt;I opened the Workspace Agent panel in Tables, selected the &lt;strong&gt;MCP&lt;/strong&gt; tab, and enabled the built-in MCP server.&lt;/p&gt;

&lt;p&gt;Initially, no AWS profiles were exposed. This allowed the server to start without immediately making any configured AWS connection available to an external agent.&lt;/p&gt;

&lt;p&gt;After the server started, Tables displayed the endpoint:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;http://127.0.0.1:61041/mcp
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;I then exposed the &lt;code&gt;read-only&lt;/code&gt; AWS profile in &lt;code&gt;us-east-1&lt;/code&gt;, leaving the other connections unexposed.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fvahqn42dtehb25qvz70j.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fvahqn42dtehb25qvz70j.png" width="800" height="518"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7oy2c5ph10rpy9sguwhi.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7oy2c5ph10rpy9sguwhi.png" alt="Tables Workspace Agent showing the running MCP server on localhost with the read-only AWS profile exposed." width="702" height="963"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Figure 1 — Tables Workspace Agent showing the running MCP server, local endpoint, and exposed read-only AWS profile.&lt;/em&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  Verifying the loopback binding
&lt;/h3&gt;

&lt;p&gt;Next, I checked whether the MCP server was actually listening on the loopback interface.&lt;/p&gt;

&lt;p&gt;I used PowerShell to inspect the listening TCP connection:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight powershell"&gt;&lt;code&gt;&lt;span class="nv"&gt;$port&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="mi"&gt;61041&lt;/span&gt;

&lt;span class="nc"&gt;Get&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;NetTCPConnection&lt;/span&gt; &lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;State&lt;/span&gt; &lt;span class="nc"&gt;Listen&lt;/span&gt; &lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;LocalPort&lt;/span&gt; &lt;span class="nv"&gt;$port&lt;/span&gt; &lt;span class="o"&gt;|&lt;/span&gt;
    &lt;span class="nc"&gt;Select&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;Object&lt;/span&gt; &lt;span class="nc"&gt;LocalAddress&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nc"&gt;LocalPort&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nc"&gt;State&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nc"&gt;OwningProcess&lt;/span&gt;

&lt;span class="nc"&gt;Test&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;NetConnection&lt;/span&gt; &lt;span class="mf"&gt;127.0&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="mf"&gt;0.1&lt;/span&gt; &lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;Port&lt;/span&gt; &lt;span class="nv"&gt;$port&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The results showed:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Check&lt;/th&gt;
&lt;th&gt;Observed result&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Local address&lt;/td&gt;
&lt;td&gt;&lt;code&gt;127.0.0.1&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Port&lt;/td&gt;
&lt;td&gt;&lt;code&gt;61041&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Connection state&lt;/td&gt;
&lt;td&gt;Listen&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;TCP test&lt;/td&gt;
&lt;td&gt;&lt;code&gt;TcpTestSucceeded: True&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fdi5a93ntlu7edimm2wnm.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fdi5a93ntlu7edimm2wnm.png" alt="Windows PowerShell displaying a listening TCP connection on 127.0.0.1 port 61041 and a successful local connectivity test." width="799" height="286"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;PowerShell confirms that the MCP server is listening on 127.0.0.1:61041, with TcpTestSucceeded: True. This independently verifies the observed local binding rather than relying only on the endpoint displayed in Tables.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;This was the first important verification point. Rather than relying only on the endpoint shown in Tables, I confirmed that the observed listening socket was bound to the loopback address.&lt;/p&gt;

&lt;h2&gt;
  
  
  2. Exposing one AWS profile and authorizing Codex
&lt;/h2&gt;

&lt;p&gt;With the server running, I moved to the client connection.&lt;/p&gt;

&lt;p&gt;Tables generated a Codex CLI connection command for the exposed &lt;code&gt;read-only&lt;/code&gt; profile.&lt;/p&gt;

&lt;p&gt;I registered the profile-specific MCP endpoint using:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight powershell"&gt;&lt;code&gt;&lt;span class="n"&gt;codex&lt;/span&gt; &lt;span class="n"&gt;mcp&lt;/span&gt; &lt;span class="n"&gt;add&lt;/span&gt; &lt;span class="n"&gt;dynamoconsole&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="n"&gt;read&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="n"&gt;only&lt;/span&gt; &lt;span class="o"&gt;--&lt;/span&gt;&lt;span class="n"&gt;url&lt;/span&gt; &lt;span class="s2"&gt;"http://127.0.0.1:61041/mcp?profile=read-only"&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Codex detected OAuth support and opened the browser authorization flow.&lt;/p&gt;

&lt;p&gt;The authorization request was then displayed inside the Tables Agent panel.&lt;/p&gt;

&lt;p&gt;The request identified:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Client: Codex&lt;/li&gt;
&lt;li&gt;Requested AWS profile: &lt;code&gt;read-only&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;Region: &lt;code&gt;us-east-1&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;Advertised scope: &lt;code&gt;dynamoconsole:read&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;Selected permission: Read only&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;I kept the permission level set to &lt;strong&gt;Read only&lt;/strong&gt; rather than granting broader access.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fsrwtlu9mm1bv776afufh.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fsrwtlu9mm1bv776afufh.png" alt="Tables MCP authorization panel showing a Codex connection request, the read-only AWS profile, and Read only consent selected." width="794" height="59"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;Codex requests access to the exposed read-only AWS profile in us-east-1. Tables displays the OAuth approval request with Read only permission selected before authorization.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8t8f7xornw2x2q1fjxcb.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8t8f7xornw2x2q1fjxcb.png" alt="Tables MCP panel displaying Codex as an approved client with Read only access and an authorization-approved audit event." width="800" height="217"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;After approving the request, Tables lists Codex as an authorized client with Read only access. The local MCP audit records the authorization approval and displays a verified audit chain.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F57fzzrzaoh67d2ty8qki.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F57fzzrzaoh67d2ty8qki.png" alt="Tables MCP panel showing a pending Codex authorization request for the read-only AWS profile, with Read only consent selected and Approve connection and Deny options visible." width="713" height="1027"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Figure 3 — Tables showing Codex's OAuth request for the exposed read-only profile, with Read only consent selected.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;After approving the connection, the browser displayed an authentication-complete message.&lt;/p&gt;

&lt;p&gt;Tables also listed Codex under &lt;strong&gt;Approved clients&lt;/strong&gt;, showing the authorized profile and Read only permission.&lt;/p&gt;

&lt;p&gt;The local MCP audit recorded an &lt;code&gt;authorization-approved&lt;/code&gt; event and displayed a verified audit-chain indicator.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F99zkfdnk612gosvpeyyv.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F99zkfdnk612gosvpeyyv.png" alt=" " width="799" height="165"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F1xre2qnaqzzknedhxg66.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F1xre2qnaqzzknedhxg66.png" alt="Tables MCP panel showing Codex under Approved clients with Read only permission and an authorization-approved audit event." width="702" height="842"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Figure 4 — After OAuth approval, Tables lists Codex as an authorized MCP client with Read only access. The local audit records the authorization event and displays a verified audit chain.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;This step demonstrated that starting the MCP server and granting a client access were separate operations.&lt;/p&gt;

&lt;p&gt;The client needed explicit authorization before using the exposed AWS connection.&lt;/p&gt;
&lt;h2&gt;
  
  
  3. Listing DynamoDB tables through the MCP connection
&lt;/h2&gt;

&lt;p&gt;Once OAuth authorization was complete, I opened Codex and inspected the connected MCP tools.&lt;/p&gt;

&lt;p&gt;Codex reported that &lt;code&gt;dynamoconsole-read-only&lt;/code&gt; was connected and exposed &lt;strong&gt;eight tools&lt;/strong&gt;, including:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;code&gt;dynoconsole_table_list&lt;/code&gt;&lt;/li&gt;
&lt;li&gt;&lt;code&gt;dynoconsole_table_describe&lt;/code&gt;&lt;/li&gt;
&lt;li&gt;&lt;code&gt;dynoconsole_item_get&lt;/code&gt;&lt;/li&gt;
&lt;li&gt;&lt;code&gt;dynoconsole_query_prepare&lt;/code&gt;&lt;/li&gt;
&lt;li&gt;&lt;code&gt;dynoconsole_query_run&lt;/code&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;I began with table discovery rather than retrieving items or executing queries.&lt;/p&gt;

&lt;p&gt;I asked Codex to use only the authorized MCP connection to list DynamoDB tables in &lt;code&gt;us-east-1&lt;/code&gt;, without accessing item contents or modifying data.&lt;/p&gt;

&lt;p&gt;Codex invoked:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;dynoconsole_table_list
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The tool returned &lt;strong&gt;14 DynamoDB tables&lt;/strong&gt; through the approved &lt;code&gt;read-only&lt;/code&gt; profile.&lt;/p&gt;

&lt;p&gt;The response also reported:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;truncated: false
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;No item retrieval, query execution, or data modification was reported during this operation.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fllvia1b36f4qmefnep65.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fllvia1b36f4qmefnep65.png" alt="Codex CLI showing a successful DynamoDB table-list operation through the authorized read-only MCP connection, returning 14 tables in us-east-1." width="800" height="416"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Figure 5 — Codex successfully invokes dynoconsole_table_list through the authorized read-only MCP connection and discovers 14 DynamoDB tables in us-east-1. The response reports truncated: false, with no item retrieval or data modifications performed.&lt;/em&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  Inspecting table metadata
&lt;/h3&gt;

&lt;p&gt;I then used &lt;code&gt;dynoconsole_table_describe&lt;/code&gt; to inspect a non-production table named &lt;code&gt;DemoTable&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;The tool returned:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Attribute&lt;/th&gt;
&lt;th&gt;Result&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Table status&lt;/td&gt;
&lt;td&gt;ACTIVE&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Partition key&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;userID&lt;/code&gt; (String)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Sort key&lt;/td&gt;
&lt;td&gt;None&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Indexes&lt;/td&gt;
&lt;td&gt;None reported&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Billing mode&lt;/td&gt;
&lt;td&gt;PAY_PER_REQUEST&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Item count&lt;/td&gt;
&lt;td&gt;4&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Table size&lt;/td&gt;
&lt;td&gt;157 bytes&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;The metadata request completed without retrieving the individual items stored in the table.&lt;/p&gt;

&lt;p&gt;These two operations demonstrated that Codex could discover DynamoDB resources through the OAuth-authorized Tables MCP connection.&lt;/p&gt;

&lt;h2&gt;
  
  
  4. Testing profile isolation and unauthorized requests
&lt;/h2&gt;

&lt;p&gt;After verifying authorized access, I checked how the connection behaved when the requested access was unavailable.&lt;/p&gt;

&lt;h3&gt;
  
  
  Attempting to access an unexposed profile
&lt;/h3&gt;

&lt;p&gt;Although I had multiple AWS connections configured in Tables, only &lt;code&gt;read-only&lt;/code&gt; was exposed to the MCP client.&lt;/p&gt;

&lt;p&gt;I asked Codex to list tables using the unexposed &lt;code&gt;admin-profile&lt;/code&gt;, without switching to another AWS connection or using the AWS CLI.&lt;/p&gt;

&lt;p&gt;Codex reported that the connected MCP server provided tools only for the single approved profile.&lt;/p&gt;

&lt;p&gt;It did not find an administrative-profile-specific tool, and no admin-profile request was executed.&lt;/p&gt;

&lt;p&gt;This confirmed the profile restriction visible through the connected client. It was not a separate server-side bypass test.&lt;/p&gt;

&lt;h3&gt;
  
  
  Sending an unauthenticated MCP request
&lt;/h3&gt;

&lt;p&gt;Next, I tested the MCP endpoint directly, without supplying an OAuth token.&lt;/p&gt;

&lt;p&gt;Using PowerShell, I sent a JSON-RPC &lt;code&gt;tools/list&lt;/code&gt; request to the local server.&lt;/p&gt;

&lt;p&gt;The expected behavior was an authentication rejection rather than successful tool discovery.&lt;/p&gt;

&lt;p&gt;The server returned:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;HTTP Status: 401
Result: The remote server returned an error:
(401) Unauthorized.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fo2lk2irqeqndx4lw8izn.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fo2lk2irqeqndx4lw8izn.png" alt="Windows PowerShell showing a JSON-RPC tools/list request to the local Tables MCP server and an HTTP 401 Unauthorized response." width="800" height="320"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Figure 6 — An unauthenticated tools/list request was sent directly to the local MCP endpoint using PowerShell. Tables returned HTTP 401 Unauthorized, confirming that the tested endpoint rejected tool discovery without an OAuth token.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;This was the main negative security test.&lt;/p&gt;

&lt;p&gt;The endpoint was reachable locally, but the unauthenticated request was rejected instead of returning the available MCP tools.&lt;/p&gt;

&lt;h3&gt;
  
  
  Checking the read-only tool boundary
&lt;/h3&gt;

&lt;p&gt;I also inspected the capabilities available under the granted Read only consent.&lt;/p&gt;

&lt;p&gt;Codex reported that the connected server exposed no write-staging or write-execution tools.&lt;/p&gt;

&lt;p&gt;No write operation was attempted during this capability inspection.&lt;/p&gt;

&lt;p&gt;This verified the restricted tools visible to the authorized client, although it did not independently test a write-capable request against the server.&lt;/p&gt;

&lt;h2&gt;
  
  
  5. Verifying recovery after an unauthorized request
&lt;/h2&gt;

&lt;p&gt;Finally, I checked whether the rejected request affected an existing authorized connection.&lt;/p&gt;

&lt;p&gt;After the unauthenticated request returned HTTP 401, I returned to the same Codex session and requested DynamoDB table discovery through the approved MCP connection.&lt;/p&gt;

&lt;p&gt;Codex successfully invoked &lt;code&gt;dynoconsole_table_list&lt;/code&gt; again.&lt;/p&gt;

&lt;p&gt;The response returned:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Tables found: 14
Profile: read-only
Region: us-east-1
truncated: false
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fm40x89uxma89qbqmvmd1.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fm40x89uxma89qbqmvmd1.png" alt="Codex CLI showing successful authorized DynamoDB table discovery after an HTTP 401 rejection, with 14 tables returned." width="799" height="420"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Figure 7 — After the unauthenticated MCP request was rejected with HTTP 401, Codex successfully retrieved 14 DynamoDB tables through the existing authorized read-only connection without repeating OAuth authorization.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;This showed that the unsuccessful request did not prevent the previously authorized client from continuing its permitted read-only workflow.&lt;/p&gt;

&lt;p&gt;I did not need to reconfigure the AWS profile or repeat OAuth authorization to perform the recovery check.&lt;/p&gt;

&lt;h2&gt;
  
  
  Practical takeaways
&lt;/h2&gt;

&lt;p&gt;This test was mainly about whether an MCP client could access DynamoDB through an explicitly authorized connection while remaining within the tested access boundaries.&lt;/p&gt;

&lt;p&gt;The important results were:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Validation&lt;/th&gt;
&lt;th&gt;Observed result&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Local MCP server&lt;/td&gt;
&lt;td&gt;Started successfully&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Loopback binding&lt;/td&gt;
&lt;td&gt;Verified on &lt;code&gt;127.0.0.1&lt;/code&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Profile exposure&lt;/td&gt;
&lt;td&gt;One read-only profile exposed&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;OAuth consent&lt;/td&gt;
&lt;td&gt;Read only approved&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;MCP connectivity&lt;/td&gt;
&lt;td&gt;Connected with 8 tools&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Authorized table listing&lt;/td&gt;
&lt;td&gt;14 tables returned&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Table metadata inspection&lt;/td&gt;
&lt;td&gt;Successful&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Unexposed profile access&lt;/td&gt;
&lt;td&gt;No admin-specific tools available&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Unauthenticated request&lt;/td&gt;
&lt;td&gt;HTTP 401 Unauthorized&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Authorized recovery&lt;/td&gt;
&lt;td&gt;Successful&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;The testing demonstrated a working local MCP connection with selective profile exposure, OAuth-based client authorization, and rejection of an unauthenticated tool request.&lt;/p&gt;

&lt;p&gt;There were also limits to what this test established. I did not independently validate a server-side profile-isolation bypass attempt, excessive-consent rejection, write staging, or every possible MCP authorization path. Cleanup verification was not part of the completed evidence.&lt;/p&gt;

&lt;p&gt;For me, the useful part of this workflow is being able to connect an AI client to DynamoDB without automatically exposing every configured AWS profile or granting write-capable tools.&lt;/p&gt;

&lt;p&gt;The test also reinforced why successful client connectivity is only a starting point. Verifying the listening interface, inspecting the granted permissions, testing unauthorized requests, and confirming authorized recovery provide much stronger evidence of how the integration behaves.&lt;/p&gt;




&lt;p&gt;&lt;strong&gt;Application:&lt;/strong&gt; &lt;a href="https://tables.serverlesscreed.com/" rel="noopener noreferrer"&gt;Tables by ServerlessCreed&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Download:&lt;/strong&gt; &lt;a href="https://tables.serverlesscreed.com/downloads" rel="noopener noreferrer"&gt;Tables official download page&lt;/a&gt;&lt;/p&gt;

</description>
      <category>dynamodb</category>
      <category>aws</category>
      <category>mcp</category>
      <category>ai</category>
    </item>
    <item>
      <title>Simulating DynamoDB Costs in Tables: GSIs, Backups, Streams, and Replicas</title>
      <dc:creator>Sidharth Bangera</dc:creator>
      <pubDate>Sat, 10 Oct 2026 11:06:28 +0000</pubDate>
      <link>https://dev.to/sidharth_bangera/simulating-dynamodb-costs-in-tables-gsis-backups-streams-and-replicas-5h18</link>
      <guid>https://dev.to/sidharth_bangera/simulating-dynamodb-costs-in-tables-gsis-backups-streams-and-replicas-5h18</guid>
      <description>&lt;h2&gt;
  
  
  Introduction
&lt;/h2&gt;

&lt;p&gt;When I evaluate a DynamoDB design change, I want to understand how it affects cost before I change the AWS resources themselves.&lt;/p&gt;

&lt;p&gt;In this walkthrough, I use &lt;a href="https://tables.serverlesscreed.com/" rel="noopener noreferrer"&gt;&lt;strong&gt;Tables by ServerlessCreed&lt;/strong&gt;&lt;/a&gt;, a DynamoDB GUI with cost analysis and simulation tools, to compare a baseline workload with several what-if changes: two global secondary indexes (GSIs), point-in-time recovery (PITR), backup storage, DynamoDB Streams reads, and two replica Regions.&lt;/p&gt;

&lt;p&gt;The test uses a fixed synthetic workload so that each cost driver can be added separately and compared with the previous state. The dollar values shown here are &lt;strong&gt;simulator estimates&lt;/strong&gt;, not actual AWS charges.&lt;/p&gt;

&lt;h2&gt;
  
  
  Prerequisites
&lt;/h2&gt;

&lt;p&gt;To reproduce the workflow, you need:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;1. &lt;a href="https://tables.serverlesscreed.com/" rel="noopener noreferrer"&gt;Tables by ServerlessCreed&lt;/a&gt; installed. The installer is available from the &lt;a href="https://tables.serverlesscreed.com/downloads" rel="noopener noreferrer"&gt;official download page&lt;/a&gt;.&lt;/li&gt;
&lt;li&gt;Access to a non-production DynamoDB environment.&lt;/li&gt;
&lt;li&gt;A table available in Tables for cost analysis.&lt;/li&gt;
&lt;li&gt;A known Region and a consistent simulation period.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;For this test, I used:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Tables version:&lt;/strong&gt; &lt;code&gt;v3.3.37&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Table:&lt;/strong&gt; &lt;code&gt;Export_Source_Test&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Primary Region:&lt;/strong&gt; &lt;code&gt;us-east-1&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Simulation period:&lt;/strong&gt; 7 days&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Table items:&lt;/strong&gt; 100&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Execution:&lt;/strong&gt; local what-if simulation&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The simulation changed only estimator inputs; it did not modify those AWS resources.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Starting from the observed cost baseline
&lt;/h2&gt;

&lt;p&gt;I first opened &lt;strong&gt;Cost Estimation&lt;/strong&gt; using the table’s observed metrics.&lt;/p&gt;

&lt;p&gt;Reads, writes, and table storage were available, but Streams reads and backup/PITR inputs were incomplete. Tables marked those components as incomplete instead of silently treating them as zero.&lt;/p&gt;

&lt;p&gt;For the 7-day observed view, the current and forecast cost remained below &lt;code&gt;$0.01&lt;/code&gt;. I used this only as the observed starting point; the detailed comparisons below use a fixed synthetic baseline.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F3s8efsei358k94n09rl0.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F3s8efsei358k94n09rl0.png" alt="Tables Cost Optimizer for Export_Source_Test in us-east-1, showing a 7-day observed baseline with current spend below $0.01, forecast below $0.01, write amplification 1.0×, and no material savings found." width="800" height="402"&gt;&lt;/a&gt;&lt;em&gt;Observed cost baseline for Export_Source_Test. The 7-day Cost Optimizer view shows current and forecast spend below $0.01, write amplification of 1.0×, and no material savings identified before any simulation inputs were added.&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  2. Loading a fixed synthetic baseline
&lt;/h2&gt;

&lt;p&gt;To make the comparison repeatable, I loaded the built-in synthetic baseline.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Input&lt;/th&gt;
&lt;th&gt;Value&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Table storage&lt;/td&gt;
&lt;td&gt;10 GB&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Table read units&lt;/td&gt;
&lt;td&gt;6,048,000 over 7 days&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Table write units&lt;/td&gt;
&lt;td&gt;3,024,000 over 7 days&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Peak reads&lt;/td&gt;
&lt;td&gt;20/second&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Peak writes&lt;/td&gt;
&lt;td&gt;10/second&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Stream GetRecords calls&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Backup storage&lt;/td&gt;
&lt;td&gt;0 GB&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;PITR&lt;/td&gt;
&lt;td&gt;Off&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Simulated GSIs&lt;/td&gt;
&lt;td&gt;None&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Replica Regions&lt;/td&gt;
&lt;td&gt;None&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;With those inputs, Tables produced these monthly estimates:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Billing model&lt;/th&gt;
&lt;th&gt;Estimated monthly cost&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;On-Demand&lt;/td&gt;
&lt;td&gt;$13.84&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Provisioned&lt;/td&gt;
&lt;td&gt;$10.47&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Reserved 1 year&lt;/td&gt;
&lt;td&gt;$6.16&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Reserved 3 year&lt;/td&gt;
&lt;td&gt;$4.33&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;For the comparisons below, I used &lt;strong&gt;On-Demand&lt;/strong&gt; consistently so that each change was measured against the same billing model.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fv357ygv4ise2goia9tux.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fv357ygv4ise2goia9tux.png" alt="Tables Cost Estimation synthetic baseline showing 10 GB storage, 6,048,000 read units, 3,024,000 write units, 20 peak reads per second, 10 peak writes per second, zero Stream GetRecords calls, zero backup storage, PITR off, and no GSIs or replicas." width="800" height="613"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F51zy61a39zxgikcpbck5.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F51zy61a39zxgikcpbck5.png" alt="Tables Cost Estimation baseline pricing showing On-Demand $13.84, Provisioned $10.47, Reserved 1 year $6.16, and Reserved 3 year $4.33 per month." width="800" height="396"&gt;&lt;/a&gt;&lt;em&gt;Synthetic baseline inputs. I loaded a fixed 7-day workload with 10 GB of storage, 6,048,000 read units, 3,024,000 write units, no Streams activity, no backup storage, PITR disabled, and no simulated GSIs or replicas.&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Comparing two simulated GSIs
&lt;/h2&gt;

&lt;p&gt;I next added two simulated GSIs with the same write volume but very different read activity.&lt;/p&gt;

&lt;p&gt;For &lt;code&gt;SimulatedGSI1&lt;/code&gt;, I used:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Read units: 500,000 over 7 days
Write units: 3,024,000 over 7 days
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The On-Demand estimate increased from &lt;code&gt;$13.84&lt;/code&gt; to &lt;code&gt;$22.21/month&lt;/code&gt;, an increment of:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;+$8.37/month
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The breakdown showed approximately &lt;code&gt;$0.27/month&lt;/code&gt; for GSI reads and &lt;code&gt;$8.10/month&lt;/code&gt; for GSI writes.&lt;/p&gt;

&lt;p&gt;For &lt;code&gt;SimulatedGSI2&lt;/code&gt;, I used:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Read units: 5,000 over 7 days
Write units: 3,024,000 over 7 days
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;With both GSIs present, the estimate reached &lt;code&gt;$30.31/month&lt;/code&gt;. The second GSI added approximately:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;+$8.10/month
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fks457u3zynci8801151t.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fks457u3zynci8801151t.png" alt="Tables GSI simulation showing two GSIs with different read volumes and the same write volume." width="800" height="617"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F82ipcjcce76ivocl6w1a.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F82ipcjcce76ivocl6w1a.png" alt="Tables GSI simulation showing two GSIs with different read volumes and the same write volume." width="799" height="490"&gt;&lt;/a&gt;&lt;em&gt;GSI simulation inputs. SimulatedGSI1 uses 500,000 read units and SimulatedGSI2 uses 5,000 read units over the 7-day period, while both use 3,024,000 write units.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F35yi8iuqntah1kmg08zm.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F35yi8iuqntah1kmg08zm.png" alt="Tables cost breakdown showing both simulated GSIs and an On-Demand estimate of $30.31 per month, with GSI write costs dominating the index cost." width="800" height="570"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ff3z8iz28r2v02g0m77oo.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ff3z8iz28r2v02g0m77oo.png" alt="Tables cost breakdown showing both simulated GSIs and an On-Demand estimate of $30.31 per month, with GSI write costs dominating the index cost." width="800" height="572"&gt;&lt;/a&gt;&lt;em&gt;GSI cost result. With both simulated indexes added, the On-Demand estimate reaches $30.31/month. The similar write volume keeps the maintenance cost of the two GSIs close despite the large difference in read usage.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;This produced the clearest ROI comparison in the test.&lt;/p&gt;

&lt;p&gt;&lt;code&gt;SimulatedGSI2&lt;/code&gt; served &lt;strong&gt;100 times fewer simulated reads&lt;/strong&gt; than &lt;code&gt;SimulatedGSI1&lt;/code&gt;, but it carried the same write volume and almost the same incremental cost. For this workload, that makes the second GSI the weaker-ROI candidate.&lt;/p&gt;

&lt;p&gt;That does not mean a low-read production index should automatically be removed. It may still support an infrequent but important access pattern. The comparison instead gives a concrete reason to review whether its usage justifies the ongoing cost.&lt;/p&gt;

&lt;h2&gt;
  
  
  4. Adding PITR, backup storage, and Streams reads
&lt;/h2&gt;

&lt;p&gt;I then added three more cost drivers one at a time.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Change&lt;/th&gt;
&lt;th&gt;Estimate before&lt;/th&gt;
&lt;th&gt;Estimate after&lt;/th&gt;
&lt;th&gt;Increment&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Enable PITR&lt;/td&gt;
&lt;td&gt;$30.31&lt;/td&gt;
&lt;td&gt;$32.31&lt;/td&gt;
&lt;td&gt;+$2.00/month&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Add 20 GB backup storage&lt;/td&gt;
&lt;td&gt;$32.31&lt;/td&gt;
&lt;td&gt;$34.31&lt;/td&gt;
&lt;td&gt;+$2.00/month&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Add 1,000,000 Stream GetRecords calls over 7 days&lt;/td&gt;
&lt;td&gt;$34.31&lt;/td&gt;
&lt;td&gt;$35.17&lt;/td&gt;
&lt;td&gt;+$0.86/month&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fc92dkc4104lmor0dfm45.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fc92dkc4104lmor0dfm45.png" alt=" " width="800" height="446"&gt;&lt;/a&gt;&lt;em&gt;PITR enabled in the simulation. Point-in-time recovery was enabled while the existing GSI assumptions remained unchanged.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fkcdbtxoxldgurmbk0kh3.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fkcdbtxoxldgurmbk0kh3.png" alt="Tables cost breakdown showing PITR enabled, a Backups plus PITR cost of $2.00, and an On-Demand monthly total of $32.31." width="800" height="448"&gt;&lt;/a&gt;&lt;em&gt;PITR cost result. Enabling PITR increased the simulated On-Demand estimate from $30.31 to $32.31/month, an increment of $2.00/month.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;I then kept PITR enabled and added 20 GB of backup storage.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fj9tx72i0mmceizw61z7z.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fj9tx72i0mmceizw61z7z.png" alt=" " width="800" height="618"&gt;&lt;/a&gt;&lt;em&gt;Backup storage added. Backup storage was increased to 20 GB while PITR and the existing GSI assumptions remained unchanged.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fqi1aoezsyiyvkoickxnq.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fqi1aoezsyiyvkoickxnq.png" alt="Tables cost breakdown showing 20 GB of backup storage with PITR enabled, a combined Backups plus PITR cost of $4.00, and an On-Demand total of $34.31 per month." width="800" height="450"&gt;&lt;/a&gt;&lt;em&gt;Backup storage cost result. Adding 20 GB of backup storage increased the simulated On-Demand estimate from $32.31 to $34.31/month, another $2.00/month increase.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Finally, I added 1,000,000 Stream GetRecords calls over the same seven-day period. This increased the simulated On-Demand estimate from $34.31 to $35.17/month, an additional $0.86/month.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F6d40ajynxwf4hn10pkrc.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F6d40ajynxwf4hn10pkrc.png" alt="Tables Cost Estimation showing one million Stream GetRecords calls over seven days, with two simulated GSIs, PITR enabled, and 20 GB of backup storage." width="799" height="568"&gt;&lt;/a&gt;&lt;em&gt;Streams read activity added. I set the simulation to 1,000,000 DynamoDB Streams GetRecords calls over the 7-day period while keeping the existing GSIs, PITR, and backup assumptions unchanged.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F6dredb2q3bewpf9z4810.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F6dredb2q3bewpf9z4810.png" alt="Tables cost breakdown showing one million Stream GetRecords calls and an On-Demand monthly estimate of $35.17 after PITR and 20 GB of backup storage were already included." width="800" height="507"&gt;&lt;/a&gt;&lt;em&gt;Streams cost result. Adding 1,000,000 Stream GetRecords calls increased the simulated On-Demand estimate from $34.31 to $35.17/month, an increment of $0.86/month.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;AWS documents PITR and on-demand backup charges as storage-based. DynamoDB Streams reads are billed in Streams read request units, with each &lt;code&gt;GetRecords&lt;/code&gt; call counting as one unit, subject to the applicable AWS pricing and Free Tier conditions.&lt;/p&gt;

&lt;p&gt;The &lt;code&gt;$0.86&lt;/code&gt; value above is the &lt;strong&gt;Tables simulator estimate&lt;/strong&gt; for this seven-day Streams input.&lt;/p&gt;

&lt;p&gt;Before adding replicas, the simulated On-Demand total was:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;$35.17/month
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  5. Adding two replica Regions
&lt;/h2&gt;

&lt;p&gt;I then added two replica Regions:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Replica 1: us-east-2
Replica 2: us-west-1
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;After adding &lt;code&gt;us-east-2&lt;/code&gt;, the On-Demand estimate increased from &lt;code&gt;$35.17&lt;/code&gt; to &lt;code&gt;$61.97/month&lt;/code&gt;.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Increment: +$26.80/month
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The breakdown showed &lt;code&gt;$24.30/month&lt;/code&gt; for replicated writes and &lt;code&gt;$2.50/month&lt;/code&gt; for replicated storage.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fai48mb26x67e95jz7bm4.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fai48mb26x67e95jz7bm4.png" alt="Tables Cost Estimation showing us-east-2 added as the first replica Region with two simulated GSIs, PITR enabled, 20 GB of backup storage, and one million Stream GetRecords calls." width="799" height="602"&gt;&lt;/a&gt;&lt;em&gt;First replica Region added. I added us-east-2 as the first simulated replica while keeping the existing GSIs, PITR, backup storage, and Streams assumptions unchanged.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fhtx8ubts9czadyalhdj8.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fhtx8ubts9czadyalhdj8.png" alt="Tables cost breakdown after adding the us-east-2 replica, showing an On-Demand estimate of $61.97 per month and replica-related write and storage costs." width="800" height="565"&gt;&lt;/a&gt;&lt;em&gt;First replica cost result. Adding us-east-2 increased the simulated On-Demand estimate from $35.17 to $61.97/month, an increment of $26.80/month.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;After adding &lt;code&gt;us-west-1&lt;/code&gt;, the estimate increased again:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;$61.97/month -&amp;gt; $91.79/month
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The second replica added:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;+$29.82/month
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;With both replicas present, Tables showed &lt;code&gt;$51.32/month&lt;/code&gt; for replicated writes and &lt;code&gt;$5.30/month&lt;/code&gt; for replicated storage. The combined replica increase was therefore:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;+$56.62/month
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8wi5wuuiqv4500qxjynd.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8wi5wuuiqv4500qxjynd.png" alt="Tables Cost Estimation showing two replica Regions, us-east-2 and us-west-1, with the existing GSIs, PITR, backup storage, and Streams assumptions still configured." width="799" height="616"&gt;&lt;/a&gt;&lt;em&gt;Second replica Region added. I then added us-west-1 as a second simulated replica while keeping the rest of the workload unchanged.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8dqesqt0f8j7m8kin5h4.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8dqesqt0f8j7m8kin5h4.png" alt="Tables cost breakdown with replica Regions us-east-2 and us-west-1, showing a final On-Demand estimate of $91.79 per month with replicated write and storage costs." width="800" height="562"&gt;&lt;/a&gt;&lt;em&gt;Two-replica cost result. Adding us-west-1 increased the simulated On-Demand estimate from $61.97 to $91.79/month. The two replicas added $56.62/month in total compared with the pre-replica state.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;The replica increments differ because the selected Regions use different regional pricing.&lt;/p&gt;

&lt;p&gt;This behavior is consistent with AWS DynamoDB Global Tables billing, where replicated writes are charged in every Region containing a replica and each replica also carries storage cost.&lt;/p&gt;

&lt;p&gt;Tables also showed the one-year and three-year reserved options as &lt;strong&gt;Not offered&lt;/strong&gt; after the Global Table scenario was added. AWS documentation states that reserved capacity cannot be purchased for replicated write capacity units.&lt;/p&gt;

&lt;h2&gt;
  
  
  Practical takeaways
&lt;/h2&gt;

&lt;p&gt;The synthetic baseline started at &lt;code&gt;$13.84/month&lt;/code&gt; On-Demand. After adding two GSIs, PITR, backup storage, Streams reads, and two replica Regions, the final simulated total reached &lt;code&gt;$91.79/month&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;The component-level comparison was more useful than the final number alone.&lt;/p&gt;

&lt;p&gt;For this workload:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;GSI write activity contributed much more to index cost than the difference in read activity.&lt;/li&gt;
&lt;li&gt;The lower-read GSI had nearly the same maintenance cost as the higher-read GSI because both used the same simulated write volume.&lt;/li&gt;
&lt;li&gt;PITR, backup storage, and Streams reads produced smaller incremental increases.&lt;/li&gt;
&lt;li&gt;The two replica Regions were the largest incremental cost driver in the simulation.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;These results are specific to the assumptions used in this test. Production traffic, item sizes, table class, Free Tier eligibility, discounts, Region-specific pricing, and the business value of individual access patterns can all change the outcome.&lt;/p&gt;

&lt;p&gt;For me, the useful part of this workflow is being able to compare these design choices before changing the infrastructure. The simulation does not replace real usage data or an AWS bill, but it provides a consistent way to test assumptions and identify which cost drivers deserve closer review.&lt;/p&gt;

</description>
      <category>dynamodb</category>
      <category>aws</category>
      <category>finops</category>
      <category>cloud</category>
    </item>
    <item>
      <title>Running, Saving, and Reopening a Parameterized DynamoDB PartiQL SELECT in Tables</title>
      <dc:creator>Sidharth Bangera</dc:creator>
      <pubDate>Thu, 08 Oct 2026 18:02:07 +0000</pubDate>
      <link>https://dev.to/sidharth_bangera/running-saving-and-reopening-a-parameterized-dynamodb-partiql-select-in-tables-48d4</link>
      <guid>https://dev.to/sidharth_bangera/running-saving-and-reopening-a-parameterized-dynamodb-partiql-select-in-tables-48d4</guid>
      <description>&lt;h2&gt;
  
  
  Introduction
&lt;/h2&gt;

&lt;p&gt;When I reuse a database query, I want the query structure to stay consistent while the input value can change.&lt;/p&gt;

&lt;p&gt;In this walkthrough, I use &lt;a href="https://tables.serverlesscreed.com/" rel="noopener noreferrer"&gt;&lt;strong&gt;Tables by ServerlessCreed&lt;/strong&gt;&lt;/a&gt;, a DynamoDB GUI with a PartiQL Workbench, to run a parameterized DynamoDB PartiQL &lt;code&gt;SELECT&lt;/code&gt;, verify the returned items, save the statement, restart the application, reopen the saved query, and verify the result again.&lt;/p&gt;

&lt;p&gt;I also include a simple syntax-error recovery check. The test uses a deterministic fixture, so I know the expected result before execution and can verify the workflow against actual data rather than relying only on a success message.&lt;/p&gt;

&lt;h2&gt;
  
  
  Prerequisites
&lt;/h2&gt;

&lt;p&gt;To reproduce the workflow, you need:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;a href="https://tables.serverlesscreed.com/" rel="noopener noreferrer"&gt;Tables by ServerlessCreed&lt;/a&gt; installed. The installer is available from the &lt;a href="https://tables.serverlesscreed.com/downloads" rel="noopener noreferrer"&gt;official download page&lt;/a&gt;.&lt;/li&gt;
&lt;li&gt;Access to a non-production DynamoDB environment.&lt;/li&gt;
&lt;li&gt;A table with predictable test data.&lt;/li&gt;
&lt;li&gt;A known expected result that you can compare with the query output.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;For this test, I used a table named &lt;code&gt;table_bulk&lt;/code&gt; with 120 deterministic items. The fixture was prepared so that 30 items had:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Status = BULK_EDITED
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;In this table, &lt;code&gt;OrderId&lt;/code&gt; is the partition key and &lt;code&gt;Status&lt;/code&gt; is the sort key. Because &lt;code&gt;BULK_EDITED&lt;/code&gt; appears on multiple items, &lt;code&gt;Status&lt;/code&gt; was a useful field for testing a parameterized query that returned more than one record.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. The query I wanted to reuse
&lt;/h2&gt;

&lt;p&gt;Instead of putting the filter value directly into the SQL, I used a named parameter:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight sql"&gt;&lt;code&gt;&lt;span class="k"&gt;SELECT&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt;
&lt;span class="k"&gt;FROM&lt;/span&gt; &lt;span class="nv"&gt;"table_bulk"&lt;/span&gt;
&lt;span class="k"&gt;WHERE&lt;/span&gt; &lt;span class="nv"&gt;"Status"&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;{{&lt;/span&gt;&lt;span class="n"&gt;statusValue&lt;/span&gt;&lt;span class="p"&gt;}};&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The parameter value for this run was:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;statusValue = BULK_EDITED
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;This keeps the query structure reusable while letting the value be supplied separately when the statement is run.&lt;/p&gt;

&lt;p&gt;I opened the &lt;strong&gt;PartiQL Workbench&lt;/strong&gt;, selected &lt;code&gt;table_bulk&lt;/code&gt;, and entered the statement. Autocomplete suggestions were available again in the v3.3.37 build while I was typing.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ffmz0j2mqrxplebnmq0eg.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ffmz0j2mqrxplebnmq0eg.png" alt="Tables PartiQL Workbench showing table_bulk selected and the Status attribute suggested by autocomplete" width="781" height="252"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;After selecting &lt;code&gt;table_bulk&lt;/code&gt;, Tables shows the &lt;code&gt;Status&lt;/code&gt; attribute in autocomplete while I build the PartiQL SELECT.&lt;/em&gt;&lt;/p&gt;
&lt;h2&gt;
  
  
  2. Supplying the parameter and reviewing the query
&lt;/h2&gt;

&lt;p&gt;When I ran the parameterized statement, Tables opened a &lt;strong&gt;Query parameters&lt;/strong&gt; dialog and detected &lt;code&gt;statusValue&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;I entered:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;BULK_EDITED
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;and chose &lt;strong&gt;Load for review&lt;/strong&gt;.&lt;/p&gt;

&lt;p&gt;Tables then resolved the parameterized statement to:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight sql"&gt;&lt;code&gt;&lt;span class="k"&gt;SELECT&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt;
&lt;span class="k"&gt;FROM&lt;/span&gt; &lt;span class="nv"&gt;"table_bulk"&lt;/span&gt;
&lt;span class="k"&gt;WHERE&lt;/span&gt; &lt;span class="nv"&gt;"Status"&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s1"&gt;'BULK_EDITED'&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;This review step was useful because I could see the exact statement that would be executed before running it.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F4gtmh1l9a6shfyc6yae5.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F4gtmh1l9a6shfyc6yae5.png" alt="Tables PartiQL Workbench showing the parameterized SELECT before entering the query parameter." width="800" height="326"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fkd0b3o760rufv57bxxlg.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fkd0b3o760rufv57bxxlg.png" alt="Tables Query parameters dialog prompting for the statusValue input before loading the query for review." width="800" height="441"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fwqiyr9a3k4aidiypqqko.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fwqiyr9a3k4aidiypqqko.png" alt="Tables Query parameters dialog showing statusValue with the value BULK_EDITED." width="800" height="388"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;After running the parameterized statement, Tables prompts for statusValue; I entered BULK_EDITED and loaded the resolved query for review.&lt;/em&gt;&lt;/p&gt;
&lt;h2&gt;
  
  
  3. Verifying the first result
&lt;/h2&gt;

&lt;p&gt;I then ran the resolved query.&lt;/p&gt;

&lt;p&gt;The expected result was 30 matching items because the fixture had been prepared with exactly 30 items whose &lt;code&gt;Status&lt;/code&gt; value was &lt;code&gt;BULK_EDITED&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;Tables returned &lt;strong&gt;30 items&lt;/strong&gt;, which matched that expectation.&lt;/p&gt;

&lt;p&gt;This was the important verification point for the first run. I was not relying only on a generic success message; I had a known fixture and an expected result before execution.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fax9jcvugeqq6x9a74qrc.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fax9jcvugeqq6x9a74qrc.png" alt="Tables PartiQL results showing a successful query with 30 returned BULK_EDITED items." width="800" height="514"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;After running the resolved SELECT, Tables returns the expected 30 items from the deterministic fixture.&lt;/em&gt;&lt;/p&gt;
&lt;h2&gt;
  
  
  4. Saving the statement and reopening it after restart
&lt;/h2&gt;

&lt;p&gt;Next, I saved the &lt;strong&gt;parameterized template&lt;/strong&gt; rather than only the resolved literal query.&lt;/p&gt;

&lt;p&gt;The saved statement retained:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight sql"&gt;&lt;code&gt;&lt;span class="k"&gt;SELECT&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt;
&lt;span class="k"&gt;FROM&lt;/span&gt; &lt;span class="nv"&gt;"table_bulk"&lt;/span&gt;
&lt;span class="k"&gt;WHERE&lt;/span&gt; &lt;span class="nv"&gt;"Status"&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;{{&lt;/span&gt;&lt;span class="n"&gt;statusValue&lt;/span&gt;&lt;span class="p"&gt;}};&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;I then restarted Tables, returned to the PartiQL Workbench, and selected the saved statement from &lt;strong&gt;My statements&lt;/strong&gt;.&lt;/p&gt;

&lt;p&gt;The parameterized query itself was still there after restart, but the previous parameter value was not retained. Selecting the saved query opened the &lt;strong&gt;Query parameters&lt;/strong&gt; dialog again, so I entered &lt;code&gt;BULK_EDITED&lt;/code&gt; a second time.&lt;/p&gt;

&lt;p&gt;That behavior is important: Tables preserved the reusable query template, while the input value had to be supplied again when reopening it.&lt;/p&gt;

&lt;p&gt;After loading the value for review and rerunning the query, Tables again returned the expected 30 items.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Check&lt;/th&gt;
&lt;th&gt;First run&lt;/th&gt;
&lt;th&gt;Reopened run&lt;/th&gt;
&lt;th&gt;Expected keys/items&lt;/th&gt;
&lt;th&gt;Actual keys/items&lt;/th&gt;
&lt;th&gt;Difference&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Query template&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;{{statusValue}}&lt;/code&gt; present&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;{{statusValue}}&lt;/code&gt; present&lt;/td&gt;
&lt;td&gt;Same saved statement&lt;/td&gt;
&lt;td&gt;Same saved statement reopened&lt;/td&gt;
&lt;td&gt;None&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Parameter value&lt;/td&gt;
&lt;td&gt;Entered &lt;code&gt;BULK_EDITED&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;Re-entered &lt;code&gt;BULK_EDITED&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;Same input value&lt;/td&gt;
&lt;td&gt;Same input value&lt;/td&gt;
&lt;td&gt;None&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Result&lt;/td&gt;
&lt;td&gt;30 items&lt;/td&gt;
&lt;td&gt;30 items&lt;/td&gt;
&lt;td&gt;30 deterministic fixture items matching &lt;code&gt;Status = BULK_EDITED&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;Same 30-item result set observed after restart&lt;/td&gt;
&lt;td&gt;None observed&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;I did not export a separate key manifest for this run, so the comparison was based on the deterministic fixture and the matching 30-item result before and after restart.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8yqpub0mwr2rh0lypa2l.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8yqpub0mwr2rh0lypa2l.png" alt="Tables PartiQL Workbench showing the saved parameterized statement reopened with a Query parameters dialog for statusValue." width="799" height="481"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;After restarting Tables and reopening the saved statement, the parameterized query remains intact and Tables prompts me to enter statusValue again.&lt;/em&gt;&lt;/p&gt;
&lt;h2&gt;
  
  
  5. Recovering from a syntax error
&lt;/h2&gt;

&lt;p&gt;I also tested a simple recovery case by deliberately changing the statement to invalid syntax:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight sql"&gt;&lt;code&gt;&lt;span class="n"&gt;SELECCC&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt;
&lt;span class="k"&gt;FROM&lt;/span&gt; &lt;span class="nv"&gt;"table_bulk"&lt;/span&gt;
&lt;span class="k"&gt;WHERE&lt;/span&gt; &lt;span class="nv"&gt;"Status"&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s1"&gt;'BULK_EDITED'&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Tables rejected the statement and displayed:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Statement wasn't well formed, can't be processed: Expected data manipulation
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The malformed statement remained visible in the editor, so I could correct it directly instead of recreating it from scratch.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fn986ys3so8lg132p1xba.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fn986ys3so8lg132p1xba.png" alt="Tables PartiQL Workbench showing a malformed SELECCC statement and the resulting syntax error." width="800" height="511"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;After running the deliberately malformed SELECCC statement, Tables reports a syntax error while leaving the statement in the editor for correction.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;I corrected the query and ran it again. Tables reported a successful execution and returned &lt;strong&gt;30 items&lt;/strong&gt;, matching the expected fixture result.&lt;/p&gt;

&lt;p&gt;The query history also kept the failed statement and the successful corrected statement as separate entries during this retest.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fa05xxd5b33yw7213bagy.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fa05xxd5b33yw7213bagy.png" alt="Tables PartiQL Workbench showing the corrected SELECT and a successful 30-row result after syntax-error recovery." width="800" height="517"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;After correcting the statement, Tables executes the SELECT successfully and returns the expected 30 matching items.&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Conclusion
&lt;/h2&gt;

&lt;p&gt;This test confirmed the full workflow I wanted to verify in Tables: creating a parameterized DynamoDB PartiQL query, supplying the parameter value, checking the returned items, saving the query template, reopening it after restart, and running it again with the same expected result.&lt;/p&gt;

&lt;p&gt;The saved statement retained the &lt;code&gt;{{statusValue}}&lt;/code&gt; placeholder, while the parameter value had to be entered again after reopening. With the same &lt;code&gt;BULK_EDITED&lt;/code&gt; input and unchanged fixture, both runs returned the expected 30 items.&lt;/p&gt;

&lt;p&gt;The syntax-error check also showed that a malformed statement remained available for correction, and the corrected query returned the expected result afterward.&lt;/p&gt;

&lt;p&gt;For this test, the most useful part of the workflow was keeping the query reusable without hard-coding the input value into the saved template, while still verifying the result against a deterministic fixture.&lt;/p&gt;

</description>
      <category>aws</category>
      <category>dynamodb</category>
      <category>database</category>
      <category>tutorial</category>
    </item>
  </channel>
</rss>
