4

Well,
I went ahead and tested t2.micro and lambda+dynamo(free tier)

You definitely get better performance and load handing with lambda+dynamo (5rcu+5wcu)

Tested the two with wrk and a simple GET which reads an item from a database of 90k items.

I could share more details with you if youre interested, but with 2000 requests, 100 connections and 4 threads. I got about 26requests/s on ec2 and about 260r/s on lambda.

Latency for ec2 was about 28s.
Latency for lambda was about 22s.
(max load)

Comments
  • 0
    My first time performing such a test, so I'm also not 100% sure on how to interpret the results. I'm also a little confused on how these threads affect no. of connections.
  • 0
    Of course you have to keep in mind that a t2 micro is shit-tier for any kind of loadtesting. Grab at LEAST a t2.xlarge or something with more than 1 vcore.
  • 1
    @arcsector oh I just wanted to compare what id be getting out of the free tier.
  • 1
    Lambda can be great but we have banned them at work because we went that direction and things like integration testing ended up being a shit show
  • 0
    Heh, I called it (better performance with lambda than a micro) and no-one believed me... 😁

    Ok, smug posting over :-)
  • 1
    @TheAnimatrix Duh, on free tier the cross-cloud option is worse than the specific one. I'd be surprised if it weren't.
  • 1
    @Lor-inc yeah I kinda knew lambda would be better but I just wanted to find out the factor by which it was better on the free tier. A 10x factor surprised me a little is all.
Add Comment