DEV Community

Discussion on: What's your go-to stack these days?

Collapse
theycallmemac profile image
James McDermott

I'm a big fan of what folks have dubbed the GoLD stack:

  • Go, the language
  • Lambda, your compute power
  • DynamoDB, a database that scales really well

Who doesn't love a little bit of serverless :D

Collapse
madza profile image
Madza

It almost feels like they purposely picked Lambda and DynamoDB for working with Go, just so the stack would sound that well xddd

Collapse
theycallmemac profile image
James McDermott

Lol yeah

Step one is always have a kick-ass name! :D

Thread Thread
madza profile image
Madza

Hahah, I'm currently working on a MERN stack, I'm thinking to switch out React to Vue just so I could be the mighty VENoM dev xddd

Collapse
daveparr profile image
Dave Parr

What's your experience with Dynamo DB? As far as I've heard it it's great for record based transactions, but tricky for aggregate analytics work. (Disclaimer: I've barely even read the docs :p)

Collapse
elasticrash profile image
Stefanos Kouroupis

Dynamo DB is great nowadays....a year ago was the worst thing I could have used ....I've never written so much code for such a small project (extremely high traffic) to handle just throttling strategies. Then one day we turned on "on demand" and never cared for that service ever again

Collapse
theycallmemac profile image
James McDermott

I have never used it personally for analytics, but I have heard using it in tandem with Amazon Athena is great for that kind of stuff!

I like DynamoDB a lot, though I have heard it's not great if you have a data set which is growing rapidly.

Thread Thread
daveparr profile image
Dave Parr

I have used Athena with an S3 data store. That worked well for that particular job :)

Collapse
katafrakt profile image
Paweł Świątkowski

I don't, I guess ;) We used lambda since quite early days but in almost every case we regretted it in the end. It always turned out that Kubernetes with some application processing (eg. Sidekiq) scales better, performs better and we have better insight into what's going on if something goes wrong.

Collapse
theycallmemac profile image
James McDermott

Hahah yes, definitely there are draw backs when it comes to serverless, I think it's all about the project at the end of the day.

Your final point is the most important to me, transparency is actually key for when (not if) something goes wrong :)