DEV Community: Karthik Sakthivel

Amazon Bedrock announces support for cost allocation tags on inference profiles

Karthik Sakthivel — Wed, 06 Nov 2024 17:09:45 +0000

What's new at AWS 📢

🔰 Amazon Bedrock now supports customers to allocate and track on-demand foundation model usage.

🔰 With this, customers can categorize their GenAI inference costs by department, team, or application using AWS cost allocation tags.

⚠️ What is Amazon Bedrock:
✔️ It is a fully managed service that offers a choice of high-performing foundation models from leading AI companies via a single API.
✔️ It also provides a broad set of capabilities such as security, privacy, and responsible AI capabilities built in.

🔰 These capabilities help customer to build tailored applications for multiple use cases across different industries.

🔰 Importantly it is helping organizations by ensuring customer trust and data governance.

🔰 You can leverage this feature by creating an application inference profile and tagging it.

⚠️ What is Inference profiles:
✔️ These profiles are a resource in Amazon Bedrock that define a model and one or more Regions
✔️ Inference profile can route model invocation requests.

⚠️ Types of inference profiles:
1️⃣ Cross region inference profiles
2️⃣ Application inference profiles

⚠️ When to use inference profiles:
✔️ Track usage metrics
✔️ Use tags to monitor costs
✔️ Cross-region inference

📌 Explore more about cross-region inference profiles:
https://aws.amazon.com/blogs/machine-learning/getting-started-with-cross-region-inference-in-amazon-bedrock/

AWS announces UDP support for AWS PrivateLink and dual-stack Network Load Balancers

Karthik Sakthivel — Tue, 05 Nov 2024 16:34:59 +0000

What's new at AWS 📢

✴️ AWS announces UDP support for AWSPrivateLink and dual-stack Network Load Balancers

✴️ Before, AWS PrivateLink only supported TCP, while NLB supported UDP only over IPv4.

✴️ After, this feature enables customers who use AWS PrivateLink and clients that use IPv6 to access UDP-based applications such as media-streaming, gaming, VoIP and other applications.

✴️ To use this capability, customer can now create a UDP listener on their existing or newly created dual-stack NLBs.

✴️ A dual-stack NLB supports both IPv4 and IPv6, and can process client requests that use either IP version.

✴️ What is AWS Network Load Balancer:
✔️ It is a high performance layer 4 load balancer that provides availability and scalability to applications while serving as the single point of contact for clients.

✴️ What is AWS PrivateLink:
✔️ It is a highly available and scalable service that enables to privately connect VPC to supported services such as;
✔️ AWS services, custom Services, Third-party SaaS services on the AWS Marketplace

✴️ Customer can attach the NLB to a VPC Endpoint Service, enabling clients to connect over AWS PrivateLink.

✴️ Customers that access services over AWS PrivateLink can now create VPC endpoint of type ‘interface’ to access UDP-based Endpoint Services, just like used to access TCP-based services today.

📌 Keynote for Pricing:
✔️ There are no additional charges for this capability.
✔️ Standard charges for AWS PrivateLink and Network Load Balancer apply.

📌 Explore more dual stack UDP NLB clients:
https://aws.amazon.com/blogs/networking-and-content-delivery/accelerate-ipv6-application-migration-with-aws-privatelink-and-dual-stack-network-load-balancers-udp-support/

Amazon EC2 Auto Scaling supports final validation time for instance refresh

Karthik Sakthivel — Mon, 04 Nov 2024 12:06:42 +0000

What's new at AWS 📢

🔯 With this feature customer can utilize this time to conduct necessary validation or testing.

🔯 To ensure the successful deployment of new EC2 instances to Auto Scaling group before an instance refresh is marked successful.

🔯 Importantly, with an instance refresh, customer can update EC2 instances in Auto Scaling groups either all at once or incrementally using instance refresh checkpoints.

🔯 These checkpoints enable to validate the health of replaced instances at different stages.

🔯 Overall, this new feature helps to monitor the following:
✅ Health of all updated instances
✅ Address any last-minute issues
✅ Ensure a successful deployment.

🔯 Addition to that, during this period, you can also take advantage of manual or automatic rollbacks as needed to revert any changes.

🔯 When to use an instance refresh:
✅ Deploying a new AMIor new user data script across ASG
✅ Migrating EC2 instances to new instance types for optimization.
✅ Switching ASG configuration from launch configuration to a launch template.

📌 Checkout for EC2 Auto Scaling instance refresh:
https://docs.aws.amazon.com/autoscaling/ec2/userguide/instance-refresh-overview.html

AWS India customers can now setup automated recurring payments on their cards

Karthik Sakthivel — Mon, 28 Oct 2024 17:33:36 +0000

What's new at AWS 📢

🔰 #AWSIndiacustomers can now setup automated #recurring_payments on their cards

🔰 It enables to pay their monthly AWS invoices, through e-mandates on their debit and credit cards.

🔰 The automated recurring payments will streamline the bill payment process for customers. Once customers set up e-mandate, they will be automatically charged for invoices up to INR 15,000.

🔰 This will avoid the need to make manual payments for monthly invoices, and help prevent missed payments.

✴️ Key note:
✔️ As per Reserve Bank of India (RBI) guidelines, e-mandates allow customers to schedule automatic payments using saved cards.
✔️ Multi-factor authentication is mandatory to complete both the current payment and e-mandate setup.

✴️ Steps involved to setup e-mandate on cards:
✔️ Customers can log-in to the AWS console, and navigate to the Payments page through the Billing Dashboard.
✔️ Then user/customers from India can then pay for an invoice by choosing a card.

📌 Checkout this article for Recurring payment:
https://repost.aws/articles/ARdzNCmUQbQI2b_xAfa1zSLA/aws-india-customers-can-now-setup-automated-recurring-payments-on-their-cards

📌 Explore more about Managing your payments in India:
https://docs.aws.amazon.com/awsaccountbilling/latest/aboutv2/edit-aispl-payment-method.html

AWS Firewall Manager now supports retrofitting of existing AWS WAF Web ACLs

Karthik Sakthivel — Sat, 26 Oct 2024 17:09:51 +0000

What's new at AWS 📢

🔱 AWS Firewall Manager now supports retrofitting of existing AWS WAF #WebACLs

🔱 It enables customers to centrally create policies for AWS WAF that add baseline rule sets to existing WAF WebACLs associated with their resources.

🔱 With this, security administrators can now use Firewall Manager policies for WAF to insert first and last rule groups

🔱 Also centrally configure a logging destination for existing WebACLs while leaving custom rule sets intact.

🔱 To centrally define baseline protection that applies to resources protected by WAF while ensuring it is enforced by the WebACLs, By enabling the “retrofit“ setting on a Firewall Manager WAF policy.

🔱 It helps customers to rapidly deploy a standard set of WAF rules to all web applications at any time without affecting existing WAF deployments.

📌 Some of the AWS best practices of AWS Firewall manager NACL:
⚜️ Start with automatic remediation disabled
⚜️ Don't modify the value of the FMManaged tag on a network ACL
⚜️ Don't modify the rules that are managed by Firewall Manager
⚜️ Don't modify the associations for subnets that have Firewall Manager managed network ACLs
⚜️ Don't modify the pre-configured rules that are managed by Firewall Manager

📌 Complete guide to setup Centrally manage AWS WAF rules with Firewall Manager:
https://aws.amazon.com/blogs/security/centrally-manage-aws-waf-api-v2-and-aws-managed-rules-at-scale-with-firewall-manager/

📌 Explore more about AWS Firewall Manager:
https://aws.amazon.com/firewall-manager/

AWS Lambda now supports using a custom serializer with Java runtimes

Karthik Sakthivel — Fri, 25 Oct 2024 12:20:22 +0000

What's new at AWS 📢

⭕ #AWSLambda now supports using a custom serializer with Java runtimes

⭕ It replacing the default Java object serialization library with a customer-defined serializer.

⭕ It enables customers to adapt their Lambda functions to cases where the incoming event format is not compatible with the default Lambda serializer.

⭕ This feature is supported in all current Lambda managed runtimes and container base images for Java.

📌 Complete guide to build, test and deploy custom Java runtime for Lambda functions:
https://aws.amazon.com/blogs/compute/build-a-custom-java-runtime-for-aws-lambda/

AWS Billing Conductor now supports RI and Savings Plans coverage and utilization reports

Karthik Sakthivel — Wed, 23 Oct 2024 18:23:22 +0000

What's new at AWS 📢

🔰 AWS Billing Conductor now supports RI and Savings Plans coverage and utilization reports

🔰 This feature allows ABC customers to analyze Reservation and Saving Plans utilization and coverage metrics scoped at the billing group level and reflecting the pro forma pricing rules.

🔰 AWS Partners can use this feature to enable their customers to view commitment discount application on cost data that reflect the customer’s specific pricing agreement.

🔰 The billing group primary account can view pro forma coverage and utilization data for all accounts in the billing group.

🔰 Non-primary accounts in an ABC billing group can view pro forma coverage and utilization for their own account only.

🔰 Organizations should use non-pro forma cost data to drive cost optimization initiatives.

🔱 Why to use AWS Billing Conductor:
✔️ It an billing feature that enables accurate, customizable billing to meet the complex needs of business and customers.
✔️ It supports customization of billing rates, distribute credits and fees, and shared overhead costs at your discretion.

📌 Guide to start leveraging AWS Billing conductor(ABC):
https://aws.amazon.com/blogs/aws-cloud-financial-management/getting-started-with-aws-billing-conductor/

Amazon Elastic Kubernetes Service now supports using NVIDIA and AWS Neuron accelerated instance types with AL2023

Karthik Sakthivel — Tue, 22 Oct 2024 18:15:05 +0000

What's new at AWS 📢

☸️ #AmazonElasticKubernetesService now supports using NVIDIA and AWS Neuron accelerated instance types with AL2023.

☸️ It is been released with improved security features, optimized boot time.

☸️ Newer kernel versions(AMIs) of AL2023:
✅ NVIDIA GPU
✅ AWS Inferentia
✅ AWS Trainium instances

☸️ The EKS Optimized NVIDIA AMI includes:
✅ NVIDIA drivers
✅ NVIDIA Fabric Manager
✅ NVIDIA container toolkit

☸️ The EKS Optimized Neuron AMI includes, the Neuron driver.

✴️ Components included in Amazon EKS optimized AMIs:
✅ kubelet
✅ AWS IAM Authenticator
✅ Docker (Amazon EKS version 1.23 and earlier)
✅ containerd

✴️ Key note:
✅ Both AMIs also come with the software needed to use AWS Elastic Fabric Adapter network interfaces.
✅ Enables customer to choose either the NVIDIA or Neuron AMI with EKS Managed Node Groups, self-managed nodes, and Karpenter as well.

📌 Explore more about Amazon EKS Optimized AMIs:
https://aws.amazon.com/blogs/containers/amazon-eks-optimized-amazon-linux-2023-accelerated-amis-now-available/

Amazon Bedrock Model Evaluation now supports evaluating custom model import models

Karthik Sakthivel — Mon, 21 Oct 2024 17:04:04 +0000

What's new at AWS 📢

⚜️ Amazon Bedrock Model Evaluation now supports evaluating custom model import models

⚜️ This feature allows customer to evaluate, compare, and select the best foundation models for your use case.

⚜️ Amazon Bedrock also offers a choice of automatic evaluation and human evaluation.

⚜️ This automatic evaluation with predefined algorithms for metrics such as accuracy, robustness, and toxicity.

⚜️ Additionally, for those metrics or subjective & custom metrics, such as friendliness, style, and alignment to brand voice, you can set up a human evaluation workﬂow with a few clicks.

⚜️ Human evaluation workflows can leverage your own employees or an AWS-managed team as reviewers. Model evaluation provides built-in curated datasets or you can bring your own datasets.

⚜️ It enables customers to evaluate their own models they imported to Amazon Bedrock through the Custom Model Import feature.

⚜️ Importantly, it allows customers to complete the cycle of selecting a base model, customizing it, evaluating it, and customizing it again or continuing to production if they are satisfied.

⚜️ To evaluate an imported model, simply select the custom model from the list of models to evaluate in the model selector tool when creating an evaluation job.

📌 https://aws.amazon.com/bedrock/developer-experience/
Explore more about Model Evaluation on Amazon Bedrock:

📌 Evaluate best foundation models in Amazon Bedrock
https://aws.amazon.com/blogs/aws/evaluate-compare-and-select-the-best-foundation-models-for-your-use-case-in-amazon-bedrock-preview/

AWS Transfer Family now provides real-time status of file transfers initiated using SFTP connectors

Karthik Sakthivel — Fri, 18 Oct 2024 15:22:14 +0000

What's new at AWS 📢

♈ AWS Transfer Family now provides real-time status of file transfers initiated using SFTP connectors

♈ With this feature, customer can query the various status of file transfer operations on demand.
▶️ Such as file transfers are completed, in-progress and queued, or failed.

♈ This capability can also be used to orchestrate post-transfer actions based on file status.
▶️ Such as sending status notifications, triggering downstream processing of the files for have transferred successfully or initiating retries for any failures.

🔱 Key points for SFTP connectors:
▶️ It is fully managed, low-code capability to transfer files between remote SFTP servers and Amazon S3.
▶️ Also, it can only connect to remote SFTP servers and directly transfer files to or from Amazon S3 that offer an internet-accessible endpoint.
▶️ It process one file at a time and transferring files sequentially, be default.

📌 Complete guide to setup Transfer Family SFTP Connectors:
https://aws.amazon.com/blogs/storage/architecting-secure-and-compliant-managed-file-transfers-with-aws-transfer-family-sftp-connectors-and-pgp-encryption/

AWS CodePipeline V2 type pipelines supports to automatically retry a stage if there is a failure in the stage.

Karthik Sakthivel — Thu, 17 Oct 2024 15:16:03 +0000

What's new at AWS 📢

🔰 #AWSCodePipeline V2 type pipelines supports to automatically retry a stage if there is a failure in the stage.

🔰 To leverage automatic retry, set “Retry” as the result for the on failure lifecycle event of a stage

🔰 Optionally you can also configure the flag to retry the stage from the first action or from the failed actions.

🔰 When a pipeline execution fails any action in the stage, then the pipeline execution will be retried in the stage once.

⭕ Three core constructs of CodePipeline:
1️⃣ Pipelines: –
✔️ It is a workflow that constructs how application/service changes go through a release process
✔️ Each pipeline is a series of stages.
2️⃣ Stages: –
✔️ It is a logical unit, used to isolate and to limit the number of concurrent changes in that environment.
✔️ Each stage contains actions
3️⃣ Actions: –
✔️ It is a set of operations performed on application code

 ✔️ It is configured, that the actions run in the pipeline at a specified point.

⭕ Important key note:
✔️ Retrying a failed stage retries all actions in the stage from the first action in the stage
✔️ Retrying failed actions, only retries all failed actions in the stage.

⭕ Key benefits:
✔️ Useful at stage with actions that can experience transient errors.
✔️ It helps to retry at failed stage, instead of pipeline execution failed.

📌 Explore more about automatically retrying a stage:
https://docs.aws.amazon.com/codepipeline/latest/userguide/stage-retry.html#stage-retry-auto

Assign billing of your shared #AmazonEC2 On-Demand Capacity Reservations

Karthik Sakthivel — Tue, 15 Oct 2024 16:00:07 +0000

What's new at AWS 📢

⚜️ Assign billing of your shared #AmazonEC2 On-Demand Capacity Reservations

⚜️ With this feature, customer can assign the billing of unused Amazon EC2 On-Demand Capacity Reservations (ODCR) to any one of their organization accounts with which the reservation is shared.

⚜️ What is Capacity Reservations:
It is a cost optimization technique, which help customer reserve compute capacity for any duration and share it across multiple accounts.
Also it enabling customer to centrally pool and manage reserved capacity.

⚜️ How to share On-Demand Capacity Reservations:
1️⃣ Select any shared reservation

 2️⃣ Initiate a request to assign its billing to a specific AWS account.

⚜️ Once the request is accepted by the new account, unused charges will be billed to the assigned account.

⚜️ Otherwise, billing for any instances running inside the Capacity Reservation is assigned to the respective accounts that launched the instances.

📒 Important Note:
✔️ This sharing feature is available at no additional cost.
✔️ Capacity Reservation owner remains the resource owner and responsible for managing the Capacity Reservation
✔️ Billing of all available capacity of a Capacity Reservation can be assigned to one account at a time.

⚜️ Key note for billing:
⚕️ If the Capacity Reservation is shared, each account is billed for their respective usage of the reservation
⚕️ If not shared, unused capacity is by default billed to the account that owns the reservation.
⚕️ With this feature, customer can have the flexibility to configure which account gets billed for the unused capacity.

📌 Checkout the pricing and billing for CR:
https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/capacity-reservations-pricing-billing.html

📌Creation of On-Demand Capacity Reservations:
https://aws.amazon.com/blogs/compute/automate-the-creation-of-on-demand-capacity-reservations-for-running-ec2-instances/

aws #awscloud #ondemand #capacityreservation #learning #whatsnew #cloudarchitect #ec2