locked
Poor performance on CRM 3.0 system RRS feed

  • Question

  • Hi,

    We have a Dynamics CRM 3.0 system that centrally hosted in Europe that having some performance issue. We have tried to access the server from different location and only 1 out of 3 locations is having poor/slow performance.

    I hope this doesn't make you confused, the situation is as below:-

    CRM server hosted in country A, Europe

    Access from country B (Europe) to the server have no problem at all.
    Access from country C (Asia) to the server have no problem too! It takes around 4 seconds to load a CRM opportunity form.
    Access from country D (Europe) to the CRM server is having some performance issue. According to the user it takes long time to load every single CRM page.

    We are using CRM web client (IE) to access the same page from different location. It's strange that only 1 of the country is having poor performance issue. We have run a PING test from the "problem" country but it looks absolutely fine!

    Example:
    Average round trip time from country D to CRM server: 75 ms - perfect!!
    Average round trip time from country C to CRM server: 459 ms - slow but still able to load CRM page in 4 seconds!!

    I have no idea why & I wonder if anyone have experience the similar problem before? Are there any task that I should perform for fine tuning/optimization? The users are screaming >=<

    Many thanks :-)
    Wednesday, September 16, 2009 12:40 PM

Answers

  • Since one of the locations is doing well, it seems like you are looking in the correct place, the connectivity.

    I see you've posted the milleseconds of the each ping.

    Have you assesed the router hops each location needs?
    Use Tracert and PathPing for that.

    Also trace the traffice from the slow site with Network monitor and evaluate the traffic types while a CRM session is attempting to load the page in the browser.

    Evaluate the traffic in the site that experiencing latency.

    In the past I have entered routes on the client machines Routing table when there was a need to connect to a remote server.   It's not an option that is used often but if you don't have a way to put static routes in a corporate router this is one way to confirm that there is difficulty in authentication and resolution.

    Also check the security logs to see if Kerberos authentication is failing from the slow site and defaulting to NTLM authentication.

    A DNS issue would prevent the slow site from querying the LDAP Database in an efficient manner.

    Hope this helps.
    /:>
    Wednesday, September 16, 2009 12:56 PM
    Moderator
  • There are several things to consider.

    First, tune the server.  If you can shutdown interfaces like firewire or a modem if it's in the server.   Believe it or not, the IP stack is aware of all the interfaces and even considers the firewire port as a network interface.

    Then make sure that the NICs are not set to automaticly sense the speed of the connection.  That's another decision that your asking the system to make.  It all adds up.   Then there is SQL itself.  First tune that server.   Run dianostics on the SQL server to watch queue lenths etc.

    In short, look for bottle necks.       A friend of mine has an invaluble CD for a very small price on tuning Windows 2003 and if there is lots of traffic on your network, look for ways to reduce it.

    It's a big subject, but the road and the wheels are what the application run on  and those were the things that need to be looked at before the CD goes in the tray.


    At convergence, one person called CRM the best network analyzer Microsoft ever made because if there is a problem in the network or server CRM will let you know.
    /:>

    • Marked as answer by Samantha Lau Friday, October 16, 2009 6:27 AM
    Friday, October 9, 2009 2:39 PM
    Moderator

All replies

  • Since one of the locations is doing well, it seems like you are looking in the correct place, the connectivity.

    I see you've posted the milleseconds of the each ping.

    Have you assesed the router hops each location needs?
    Use Tracert and PathPing for that.

    Also trace the traffice from the slow site with Network monitor and evaluate the traffic types while a CRM session is attempting to load the page in the browser.

    Evaluate the traffic in the site that experiencing latency.

    In the past I have entered routes on the client machines Routing table when there was a need to connect to a remote server.   It's not an option that is used often but if you don't have a way to put static routes in a corporate router this is one way to confirm that there is difficulty in authentication and resolution.

    Also check the security logs to see if Kerberos authentication is failing from the slow site and defaulting to NTLM authentication.

    A DNS issue would prevent the slow site from querying the LDAP Database in an efficient manner.

    Hope this helps.
    /:>
    Wednesday, September 16, 2009 12:56 PM
    Moderator
  • Hi Curt,

    Thanks for your help. I suppose I need some basic network knowledge in order to perform the task you have mentioned here. I will gather more info on how exactly to perform this task according to your suggestion.

    I will post here once I manage to sort this out. Good luck :-)

     

    Thursday, September 17, 2009 7:00 AM
  • Glad to help. 

    At Microsoft Convergence which is the Dynamics convention, one MVP commented that Microsoft CRM was the one of the best Network testing tools Microsoft every made.    In fact many of the more modern applications require a good "road" to run on.  

    Look at this as a chance to really find out what is going on in the "Pipes". 

    /:>

    Thursday, September 17, 2009 4:58 PM
    Moderator
  • Guess what!? I tried to login to the CRM system this morning and I do not experience the slow performance problem. That mean this problem only happened when they are more users login to the system at the same time. Is this issue caused by the network bandwidth instead of problem with the CRM system? What does this mean & what can I do next to perform further troubleshoot?

    Many thanks!
    Friday, October 9, 2009 3:21 AM
  • There are several things to consider.

    First, tune the server.  If you can shutdown interfaces like firewire or a modem if it's in the server.   Believe it or not, the IP stack is aware of all the interfaces and even considers the firewire port as a network interface.

    Then make sure that the NICs are not set to automaticly sense the speed of the connection.  That's another decision that your asking the system to make.  It all adds up.   Then there is SQL itself.  First tune that server.   Run dianostics on the SQL server to watch queue lenths etc.

    In short, look for bottle necks.       A friend of mine has an invaluble CD for a very small price on tuning Windows 2003 and if there is lots of traffic on your network, look for ways to reduce it.

    It's a big subject, but the road and the wheels are what the application run on  and those were the things that need to be looked at before the CD goes in the tray.


    At convergence, one person called CRM the best network analyzer Microsoft ever made because if there is a problem in the network or server CRM will let you know.
    /:>

    • Marked as answer by Samantha Lau Friday, October 16, 2009 6:27 AM
    Friday, October 9, 2009 2:39 PM
    Moderator
  • Thanks Curt, the problem has been resolved. Changes on firewall that cut down a lot of traffic & upgrade internet connection line help a lot.

    I have follow your suggestion and made change to the NTLM authentication, performance has improved a lot after I done the step suhhested here http://blogs.technet.com/industry_insiders/articles/407751.aspx

    I have also installed the CRM roll up & refered to the CRM optimization white paper for performance tuning.

    More info available here:-
    http://www.crmlady.com/2009/02/tackling-performance.html
    http://niiranen.eu/crm/?p=13

    Many thanks!

    Friday, October 16, 2009 7:28 AM