locked
Service Fabric Standalone - Performance & Throughput RRS feed

  • Question

  • Hello,

    I am keen to know about the performance & throughput of SF in standalone envrionments. I have a dummy SF stateless service deployed on the "localhost" and exposed over Kestrel & ASP.NET Core Web API Framework. Using wrk on the same machine reports 22K req/s whereas calling the same service outside SF reports 31K/s. (CPU is always 100%)

    My question is, where this overhead is coming from, and is there any way this overhead can be avoided?

    In addition, testing the ServiceRemoting communication listener gives bad results (only 7K req/s)? Do you think this is expected; because i thought the TCP communication stack should be the ideal solution for low latency and high throughput scenarios.

    Please advise.

    -SM

    Sunday, April 23, 2017 3:33 PM

Answers

  • I am exploring another solution of how to do a standalone version of the Service Fabric. The main usage of it could be an on-premises installation of the software or speeding the development up. 


    http://blog.geo.bg/2017/11/21/is-service-fabric-slow-for-development-theres-a-way-to-make-it-faster/

    Tuesday, November 21, 2017 3:00 AM

All replies

  • SM, have you made any progress with this?

    Thanks!


    Ed Price, Azure Development Customer Program Manager (Blog, Small Basic, Wiki Ninjas, Wiki)

    Answer an interesting question? Create a wiki article about it!

    Wednesday, October 4, 2017 1:43 AM
  • I am exploring another solution of how to do a standalone version of the Service Fabric. The main usage of it could be an on-premises installation of the software or speeding the development up. 


    http://blog.geo.bg/2017/11/21/is-service-fabric-slow-for-development-theres-a-way-to-make-it-faster/

    Tuesday, November 21, 2017 3:00 AM