Forums › Cached › Customization › Suitescript 2.0

This topic contains 5 replies, has 0 voices, and was last updated by michoel 7 years, 3 months ago.

  • Author
    Posts
  • July 8, 2017 at 3:51 am #18379

    borncorp

    Hey guys, I did some load testing in concurrent Map Reduce vs concurrent Suitelets to figure out which one has better performance. Check out the results at:

    https://ursuscode.com/netsuite-tips/…rent-suitelet/
    This is a cached copy. Click here to see the original post.

  • July 10, 2017 at 8:34 am #18380

    david.smith

    Interesting. I’m assuming your M/R used all 5 queues? I think I would have put a little more load on the main script. Custom records are typically faster to create than say a standard transaction. I would also be curious if you had 5 deployments of the suitelet (to make things fair with the M/R) how it would do.

    Thanks for sharing.

  • July 12, 2017 at 2:18 pm #18381

    TheUsualSuspect

    Can you post your map / reduce code?

    You are testing the performance of 5 netsuite queues vs 50 concurrent connections to a suitelet? This seems a little unfair. Assuming that the map/reduce scales linearly, with 50 queues it would do 510 records per second which is higher than Apache JMeter.

  • July 12, 2017 at 5:35 pm #18382

    michoel

    Originally posted by TheUsualSuspect

    View Post

    You are testing the performance of 5 netsuite queues vs 50 concurrent connections to a suitelet? This seems a little unfair. Assuming that the map/reduce scales linearly, with 50 queues it would do 510 records per second which is higher than Apache JMeter.

    I think that the key point here is that the 50 concurrent Suitelets are available without any extra licensing fees.

    In any event, it will be interesting to see how this changes with SuiteCloud Processors in 2017.2.

  • August 8, 2017 at 7:41 pm #18383

    TheUsualSuspect

    Originally posted by michoel

    View Post

    I think that the key point here is that the 50 concurrent Suitelets are available without any extra licensing fees.

    In any event, it will be interesting to see how this changes with SuiteCloud Processors in 2017.2.

    That is a good point but the post and blog highlight performance is why I was cross. Map/Reduce is beefy but restricted it seems. That Suitelet processing power doesn’t just fall from the sky.

  • August 9, 2017 at 1:18 am #18384

    michoel

    Well in other news, concurrent Suitelets are going to be much more restricted in the next release as well.

  • Author
    Posts

You must be logged in to reply to this topic.