DEV Community

Discussion on: What piece of tech do you regret choosing for a project?

Collapse
 
theaccordance profile image
Joe Mainwaring • Edited

I usually don't regret the piece of tech, I typically regret either the implementation, poor requirements gathering, or lack of foresight in understanding how that particular stack component evolved over time.

But onto the list:

  • Postgres: I have nothing against this DB, in fact it's my go-to when using a relational db, but implementing SQL requires an understanding of not only the syntax, but also how the DB engine operates. I've had teams bitten hard by bottlenecks we've created ourselves.
  • MongoDB: Another Database that I personally favorite, we've had regrets here with devs trying to use it as a relational db.
  • New Relic: Application Performance Monitoring is a pretty important tool when you're building at scale, but I got swindled by my account rep on cost projections when switching plans, the end result doubled my annual cost for the service. The service is good, but APM spend shouldn't be a comparable cost to cloud hosting or databases on my monthly expense spreadsheet.
  • Mailchimp: I pay this company $50k USD a year and receive zero human interaction with reps from the company. I would move off of them to someone else, but it's considered tech debt and gets prioritized as such.
  • Pre-Typescript JavaScript: I love JS, it's my bread and butter when it comes to what's driven my success, but it's 2022 and one of the SaaS products I manage is a 10 year old Node.js project. A lot of the foundation is based on outdated patterns (callback hell is an understatement) and APIs that have long since been deprecated. Given that the product is sold to the enterprise, the size of the project is not trivial and the lack of strong types and other pre-compiled safeguards means that it's an absolute pain to work on for anyone who's not intimately familiar with the codebase. Thankfully, we've stopped selling this product to new customers, with any luck we'll be able to off-board the remaining 20k users by the end of this year and turn off the lights.