Scaling App Platform

Posted March 29, 2021 1k views
PostgreSQLDigitalOcean Managed PostgreSQL DatabaseDigitalOcean App Platform

I decided to use AppPlatform for my start-up.
Frontend:Flutter Mobile App
Backend: PHP API (AppPlatform 1 GB RAM | 1 vCPU x 2)
Database: Managed PostgreSQL (2 GB RAM / 1vCPU / 25 GB Disk)
CDN: DigitalOcean Spaces

I couldnt decide is this configuration is enough for production?
I think DO doesnt provide autoScaling for appPlatform.
How can i decide “I need more App” or how to test DB source is not enough? What are the limitations for Managed Postgre.
How many concurrent connection can be handled for this architecture?

I know sizing is not easy

Which documentations should I check?


These answers are provided by our Community. If you find them useful, show some love by clicking the heart. If you run into issues leave a comment, or add your own answer to help others.

Submit an Answer
1 answer

👋 @tasmcan

Welcome to App Platform, excited to have you here! As you said sizing is not easy, and it’s difficult to recommend sizes because every app is designed and architected for different needs.

With that said, trying to figure out what “units” of resources can handle (eg. 1 vCPU/1GB RAM) then scaling that to cover your current load plus some works well. The way I would attack this is look at what resources your app uses under regular load. In App Platform this can be found under insights. Once you find that you’re going to want to try and add a little buffer room so that when you get bursts of traffic your app will continue to function as expected.

One of the great things about App Platform is it’s so easy to scale up and down when testing.

Hope this helps!

  • Hello @crashoverride ,
    Thank you for your reply.
    I did some load stress test with free accounts of some services.
    with 50 Virtual User, During 1 min;
    I sent request to

    I tried it 3 different load tester(,,loadmill)
    the result shouldn’t be normal
    Is there any default limitations for throuput etc?

    The average response time of the system being tested was 1 864ms, and 728 requests were made at an average request rate of 11 requests/second.

    Insight shows %59 and %54 CPU consumption for each app1 and app2
    %28 and %30 Memory Consumptsion

    If it seems normal, using app platform is not effective.

    • Hi @tasmcan

      There is a good chance that your load tests may have been impacted by an outage at CloudFlare which we use to protect App Platform apps (ref).

      I did a quick single request test against the link you posted and had a full response in 10ms which is pretty good.

      You may want to try your load test again. Another consideration is part of our DDoS protection is a slow down of large bursts of traffic from the same place.

      Feel free to reply here if you have more questions!