ARS 7.4.3 Slow WebUI Performance

We've been noticing slow WebUI performance, particularly in the chromium browsers.  The search results, and action panes take a long time to render (20 seconds or so).  There does not seem to be any real load on the web/admin servers (6 CPU's/24GB of ram).  I tried adding the Disable500VA attribute and setting it to 1, but that didn't help

Any ideas?

Parents
  • Have you the checked the response times between:

    1) Your AR Admin service(s) and the Managed Domains

    2) Your AR Admin service(s) and the SQL servers hosting the config DBs

    3) The web service IIS host(s) and the Admin service(s) 

  • I did, I'm getting <1ms to everything which is not surprising.  All the dc's are local to the same data center that hosts the sql db's, the admin service and the web server.

    Queries to the admin service, as compared to native ad

    420 ms to the admin service (get-qaduser using proxy and service params)

    36ms to the dc using get-aduser

  • 36ms to the dc using get-aduser

    Did you perform this test from your Admin service host(s)?  That's key.  Should be moot though since you mention that all your DC's are co-located in the same data centre as your admin service(s).

    I don't worry too much about get-qaduser -proxy being slow because the Quest cmdlets have some built-in inefficiency that goes beyond the connection to AR itself.

  • i tested from the admin services, and a host that is also on-prem that has access to the web, admin, and sql hosts.  They all produce similar results.  Firefox is smooth, chrome, and edge frequently get the spinning ball of patience rendering search results, and the right hand frame of actions

  •  

    Were you able to get some resolution on this? I seem to be having the same issue with navigation in the web portal. Just opening the domain and OU's is slow. Please post anything you may have done to help. Thank you for your time. Lu

  • We have a BigIP F5 load balancer in front of our webUI servers (prod and dev).  Both env's are exhibiting the same issue.  We notice when we bypass the F5, the speed is much better.

    No resolution yet.

  • ARS seems to render much better in Firefox.  The chromium browsers (edge, chrome) seem to be on the slow side.  I don't know if it's that the version of AngularJS that Quest uses is old, or that ARS evaluates access templates on the server side what you're allowed to see, and then the client browser has to wait on that to render the frames based on that.  

  • or that ARS evaluates access templates on the server side what you're allowed to see, and then the client browser has to wait on that to render the frames based on that

    Yes - this is definitely the case.  At least the first time you navigate to a particular location.  Thereafter, I believe your access is cached for the duration of your session.

  • For some comparison, using Edge 93.0.961.47, on a 940/860 FIOS connection, and AzureAD as our SAML provider - without the F5 in the way it takes about 2.5-3 seconds to open and browse the domain and the sub OU's underneath. 

    When we have the F5 passing traffic we can see minute long lags, or broken frames (unavailable retry)

  • When we have the F5 passing traffic we can see minute long lags, or broken frames (unavailable retry)

    So doesn't this point to the F5 being your choke point?

    Over the years, several customers of mine have struggled with tuning their F5s to work well with Active Roles.  I just stay away as I have enough to worry about just making Active Roles work they want it to. Slight smile

  • Thanks all for the information and providing things for me to test. My company also uses load balancer and it does seem to be slower through that opposed to direct to the server. Although for me it is still significantly slow going direct. We however did not have this to be this slow in 7.3.1. we upgraded to 7.4.3 and that is when things started to get super slow. I build a managed unit and a custom web page so that users can manage their own groups. In that page i have a link directing them to that managed unit which is delegated to Primary Owner. This one after the upgrade is taking a tremendous amount of time to load. I do have Disable500VA and PerformanceFlag in place. Not certain if there are any other fixes i need to consider. THANKS

Reply
  • Thanks all for the information and providing things for me to test. My company also uses load balancer and it does seem to be slower through that opposed to direct to the server. Although for me it is still significantly slow going direct. We however did not have this to be this slow in 7.3.1. we upgraded to 7.4.3 and that is when things started to get super slow. I build a managed unit and a custom web page so that users can manage their own groups. In that page i have a link directing them to that managed unit which is delegated to Primary Owner. This one after the upgrade is taking a tremendous amount of time to load. I do have Disable500VA and PerformanceFlag in place. Not certain if there are any other fixes i need to consider. THANKS

Children
No Data