none
Content Distributor "Average dispatch time" should be less than 10ms RRS feed

  • Question

  • Consider the following topology: 1 SharePoint Server (Content SSA, Query SSA) and 1 FS4SP Server (all roles).

    http://technet.microsoft.com/en-us/library/gg604768.aspx#CrawlingAndIndexingPerformanceAnalysis states:

    Average dispatch time

    The time that the content distributor uses to send a batch to an item processing component. This should be less than 10ms. Higher values indicate a congested network.

    Isn't the Content Distributor on the FS4SP sending a batch to the local server in this topology ? So in that case network congestion cannot be responsible for high values of this counter ?

    Or do I have things backwards ?

    Tuesday, August 14, 2012 1:22 PM

Answers

  • Hi,

    For a single server FS4SP topology that is correct. It is sending it to another component on the same machine, so latency should be far less than 10ms :) But, if all docprocs are working, they cannot receive new batches. Perhaps the counter also reports when it has to wait to dispatch a batch.

    How is the cpu working during indexing.. if it is working at 90% +, then cpu is the bottleneck. Also check disk queue length to determine if your disk cannot handle the reading/writing, and is causing indexing to back up.

    Thanks,
    Mikael Svenson


    Search Enthusiast - SharePoint MVP/MCT/MCPD
    http://techmikael.blogspot.com/
    Author of Working with FAST Search Server 2010 for SharePoint

    Wednesday, August 15, 2012 1:00 PM

All replies

  • Hi,

    For a single server FS4SP topology that is correct. It is sending it to another component on the same machine, so latency should be far less than 10ms :) But, if all docprocs are working, they cannot receive new batches. Perhaps the counter also reports when it has to wait to dispatch a batch.

    How is the cpu working during indexing.. if it is working at 90% +, then cpu is the bottleneck. Also check disk queue length to determine if your disk cannot handle the reading/writing, and is causing indexing to back up.

    Thanks,
    Mikael Svenson


    Search Enthusiast - SharePoint MVP/MCT/MCPD
    http://techmikael.blogspot.com/
    Author of Working with FAST Search Server 2010 for SharePoint

    Wednesday, August 15, 2012 1:00 PM
  • Hi,

    The CPU is pretty high but under control. I scaled down the docprocs from 7 to 5 to compensate. Disk is no issue.

    I was just wondering about that counter, but it makes sense that it also reports the local wait. The Technet article confused me.

    We're looking at adding another server to scale up again.

    Thanks again!

    Wednesday, August 15, 2012 7:11 PM