Rate limit in WorldServer API?

Does anyone know if there is a rate limit affecting WorldServer response? We have been plagued by a slowness issue, which we have tracked down to a change in an SAP environment that connects via the REST API to WorldServer. Ping and Tracert indicate there are no network issues; the server itself shows no unusually high CPU or memory utilization. Google Inspector shows times in the ms-range for receiving requests. What we do see are long TTFB (time to first byte) times. Could it be that there is a rate limit causing responses to be buffered and sent out after a delay, the length of which unfortunately causes the connection to the SAP system to time out?

We are taking steps to reverse the change in the SAP environment that led to this issue, but it would still be good to know for the future!

Thanks in advance!

Dave

Parents
  • Hi Dave, there is no limit. The cause of the bottleneck needs to be identified.

    WorldServer api and Tomcat logs should assist in making this determination. Please raise a support case for this providing all relevant information.

  • Hi Caterina,
    We've actually had a case open for quite some time now (00589863). It took us forever to convince the various at teams at Bayer that the issue began after the front-end SAP app was moved to the cloud. When they did that, they also placed that SAP on the launchpad of every Bayer user.
    Ryan and Thorsten Kahlich have both joined numerous calls with the various teams at Bayer and the only relief comes from reducing the number calls to WorldServer from this SAP app.I'll upload the latest statistics taken from the relevant localhost access log to the support case.

  • I think it may be the Custom Endpoint that needs optimization here, it was never tested in the early stage of conception with such a large user base and with the time that has passed since its creation there is every chance it could be streamlined somewhat to meet modern day WS changes.

  • That would be great - right now our solution has provided much-needed relief, as the number of unique users each day has dropped from a peak of just over 16,000 on 10 May to under 100 since the change. Concerns were voiced at the operational meeting yesterday re. what happens if the number of active users increases. I think we have a good buffer, but optimizing the custom endpoint would be welcomed.

Reply
  • That would be great - right now our solution has provided much-needed relief, as the number of unique users each day has dropped from a peak of just over 16,000 on 10 May to under 100 since the change. Concerns were voiced at the operational meeting yesterday re. what happens if the number of active users increases. I think we have a good buffer, but optimizing the custom endpoint would be welcomed.

Children
No Data