locked
Moderator - Please delete this post/thread RRS feed

  • Question

  • Moderators - Please delete this post/thread


    • Moved by Stanfford Liu Friday, July 30, 2010 2:06 AM 2010q (From:SharePoint - Setup, Upgrade, Administration and Operation (pre-SharePoint 2010))
    • Edited by Sreekanth Suthoju Tuesday, November 27, 2012 8:27 AM a
    Thursday, July 29, 2010 3:00 AM

All replies

  • Hi Sreekanth,

    Index server actually indexes all the content inside your sharepoint farm in order to provide one of the most powerful feature i.e.Search.

    The only data that index server stores is the 'Index data' for which the location you can set while configuring the indexer and mostly its on the same server itself.

    In order to crawl the data, indexer also needs the WFE role; just the thing is that it will not serve the web requests becoz you are not going to add it to the load balancer.

     

    Hope this helps,


    Regards, Rohan
    Thursday, July 29, 2010 11:37 AM
  • Hi Suthoju,

    This is the picture of the love story in a SharePoint farm.

    Within a SharePoint farm, one will typically have a server that performs the index function of the farm in order to manage crawling of searchable content and create an index file. I was always under the impression that this Index server performs all of the crawling for the crawl-able entities, but recently discovered this is not the case. The Index server MANAGES the crawl of all content, performs the crawls for items external to the SharePoint farm , BUT makes requests to the available Query or Search servers (this is another server on its own) in order to crawl the internal content of the SharePoint farm. The Query servers in turn, return the results to the Index server to add to the index file itself which is then passed along to the Query servers again.

    Since crawl operations can potentially cause performance impacts to your SharePoint environment, Microsoft has recommended as a best practice, having a dedicated query server available to crawl data.  This query server should not be part of the potentially load-balanced front-end servers that actually serve up the web pages to end-users and should only be responsible for crawling the SharePoint content internal to the SharePoint farm.

    Now, you probably are thinking this seems like overkill to add another server to your farm merely for crawling, and in many scenarios it probably is.  Microsoft has recommended that crawling of content less than 500 GB (keep in mind this is data internal to your SharePoint farm) of data will not result in any performance-related issues without having a dedicated Query server to perform crawls.  Once your site has approached this level however, you may want to look into the potential for creating a dedicated Query server to crawl your content.

    Including the Indexing capabilities onto this dedicated Query server to perform crawls will help alleviate some potential performance issues as well by not having to pass the crawled data across your network which your NT Admins won't like. 

    So, Mr.Suthoju your Index Server will still have to pass the index file and incremental files to all of the Query / Search servers , but you'll at least eliminate some network chattiness by throwing the Index server on the same box as the crawl-dedicated Query server.

    Considering all of this, the last thing you want to do is to that machine is adding a SQl server on it...DON'T DO IT ...it isn't the recommended best practice. The only database server that you need is the already existing SQL server existing separate from all and every other box, and if I'm not mistaken, you have already CLUSTERED them, YOu DO NOT NEED ANY OTHER DATABASE SERVER on the index server.

    Let me quote you, "My client is focussed to have crawling database in Index server rather than in Web Front End Server. "

    Web front end servers SHOULD never have any database role running on them either. In summary:

    1. Web Front End is web front end period.

    2. Index/Query/Search is index/query/search period.

    3. Clustered SQL is clustered SQL period.

    The onus is on you to educate them as the professional and when you do make reference to authorities who said so i.e Microsoft (creator of the product).

    Hope I was able to be of little help.

    Cheers!


    You've only got one life; help as many people as you can, and enjoy it while it lasts.
    • Proposed as answer by NateAnne Monday, November 7, 2011 4:42 PM
    Friday, July 30, 2010 6:44 PM
  • There is a lot of confusion on this thread - Clearing it up now

    Search in Sp2010 has a number of components which can be located in a number of 'tier' or places within the farm.

    The items you are most concerned with and provide the answer to your question are

    Crawl Component - Hosted within MSSearch.exe and lives on what is typically then tagged as a Crawler Server (Index Server Term is Deprecated)

    Crawl Database - Hosted in SQL and preferably on separate spindles and possibly in its own SQL Instance due to the required IO and Caching

    Query Component - Hosted in MSSearch.exe and is responsible for Generating the results from a full text query against the Query Partition it is associated with

    Query/Index Partition - Subset of the entire index (on disk) and contains the keywords and references to the source document ID as an Inverted Tripoli Catalogue

    Query Processor - This role lives inside W3WP.EXE and is the component responsible for orchestrating the full text and sql queries plus security trimming, duplicate removal and generation of the final result set to be sent to the Web Front End where the query was sourced.

    Property Database - Is the source of metadata/properites and summary text for the submitted query

    Admin database and Admin component - In a nutshell controls the configuration and management of the search service

     

    When Indexing SharePoint Content - Crawl Components use the sitedata webservice to check for changes or to retireve akk items on a full crawl (The Query Component is not involved in this process)

    When Indexing Web Content or File System content the crawler essentially goes direct to source using the required protocol handler.

    NOTE: The Indexer / Crawler servers do not store local copies of the index. This was a 2007 limitation that resulted in us being able to only support one indexer. In 2010 as soon as we flush the level 1 indexes and have recieved acknowledgement from the index partition that a successful merge has taken place we remove these local indexes.

    When Querying - see the query processor above

    Lots more things have changed in 2010 search. Reply to the thread if you want more detail.

     


    MSFT - SharePoint PFE
    • Proposed as answer by Alpesh NAKAR Thursday, August 5, 2010 2:22 AM
    Wednesday, August 4, 2010 7:15 PM
  • Thank you Xavier, This was helpful

    Regards

    Nataraj

    Monday, November 7, 2011 4:38 PM