topic Re: Performance Metrics for Services in ArcGIS Enterprise Questions

Performance Metrics for Services

Sunnywaygis — Thu, 02 Sep 2021 12:31:19 GMT

Hi There, I am looking for a way to generate some metrics around time taken by a map service query at ArcGIS Enterprise 10.6.1.

The chrome log tells me time taken but I want to know if there is a way to breakdown which part of query (GIS Server, database) is taking the longest and way to optimize it.

Thanks

Re: Performance Metrics for Services

HenryLindemann — Thu, 02 Sep 2021 12:55:41 GMT

Hi @Sunnywaygis
So usually, we would work from the base up if your service was created with a .mxd then you can have a look at mxdperfstat it generates metrics on query and drawing time and will indicate where the service is struggling.
mxdperfstat - (arcgis.com)
The Next step after this is to setup monitoring on ArcGIS Server
The url is https://your.dns.com:6443/arcgis/manager or https://your.dns.com/webadaptor/manager
In server Manager you go to Logs then statistics and create a New Report
Here you have the option to see Total Requests, Avr. Response Time, Timeouts and Maximum Running Instances
Hope it helps

Re: Performance Metrics for Services

Sunnywaygis — Thu, 02 Sep 2021 13:18:49 GMT

Thanks, will try these options.

Re: Performance Metrics for Services

berniejconnors — Thu, 02 Sep 2021 13:44:41 GMT

Hi,
This is not what you asked for but it may be helpful. I cannot help you identify the timings for ArcGIS Server response vs database server response but here are a few things that we do to help improve performance on map service requests:

1. Review the web logs on the web adaptors to identify the top 20 slowest map service requests. Our map services are open to the Public so we are susceptible to poorly designed queries. Looking at the slowest map service requests we were recently able to help two municipalities reconfigure their WAB search widget for much faster performance.
2. When configuring a WAB search widget do not use the "LIKE" operator in the query unless it is absolutely necessary. Using "LIKE" negates any indexing on your tables. Use the "=" operator and your queries will be much faster.
I hope this helps.
Bernie.