SharePoint 2010: The Enterprise Search product stack – Overview

Posted on October 9, 2010 by

0


I delivered another session at the SharePoint User Group on October 7th 2010 that covered the Enterprise Search product stack within the context of SharePoint 2010.  What I always find very interesting is how some folks consider search a ‘black art’.  With that in mind, I’m going to provide an overview of what there is, when you should consider each and basic functionality.

Microsoft Search Server 2010 and Search Server Express 2010

Search Server, and it’s ‘free’ Express version sister, provides an entry level search solution that is often found connected to a SharePoint Foundation farm to extend the foundation level search functionality.  Search Server 2010 Express provides the ability to index up to 10 million items, across up to 51 languages and 31 file types.  Combining SharePoint Foundation with Search Server 2010 Express provides a platform on which to build entry level search solutions with the most minimal barriers to adoption and implementation that you will find anywhere.

SharePoint Server 2010

Enterprise Search in SharePoint Server 2010 is truly enterprise in terms of scalability and features.  There is support for 100 million items per Search Service Application and the decomposition of search components means scaling is now extremely easy.  We can assign individual crawl components to individual crawl databases, we can leverage host distribution rules to specify that specific content crawlers are accessed by specific crawlers.  This is useful when we consider that in MOSS 2007 a content that resided on slow disks could very easily impact the index time.

Custom ranking models in SharePoint Server 2010 provide the ability to leverage a two-layer neural network model to override the default ranking rules to deliver content in the way you want for your specific organizations and search applications.  There is little tangible content out there at the moment and I’m striving to get the message out that this functionality exists as well as producing examples that promote understanding of the underlying architecture.

Search Refinements here, as with Search Server, provide the ability to drill down into and refine search results rather than a search term.  The major deficit with Search Refinements in both SharePoint Server 2010 and Search Server 2010/Express is the lack of exact counts.  Unfortunately, users come to expect this kind of functionality, especially those which leveraged the CodePlex Faceted Search project within MOSS 2007.  Exact refinement counts can be enabled but because SharePoint doesn’t perform analysis of the entire result set, these are difficult (if not impossible) to calculate.  They can in fact be switched on, check XML Schema for Refinements Panel web part at http://msdn.microsoft.com/en-us/library/ee819920.aspx, but the lack of accuracy here is only likely to confuse users.  It is for this reason, I guess, Microsoft decided to switch them off by default.

FAST Search Server 2010 for SharePoint

FAST Search Server 2010 for SharePoint (FAST) provides best-in-class Enterprise Search functionality that is truly leading edge.  FAST scales to support in excess of a billion item search corpus and, as with SharePoint Server 2010, the architectural components mean you can scale independantly based on what’s important to your business… e.g. Query Latency, index generation time.

In terms of features, the 5 big ticket items are:

  • Integration with Office Web Applications
    When our users search for content, specifically Word documents and PowerPoint slide decks, they often know what the content actually ‘looks like’ however this would be almost impossible to articulate textually.  FAST Search, integrated with Office Web Applications, provides in-line thumbnails of Word documents and PowerPoint slide decks.  Furthermore, users can cycle through slides within the slide deck directly within the search results.  This is very powerful and provides a hitherto unseen level of integration between common office applications and Enterprise Search.
  • Deep Refinements
    Deep Refinements build on refiners in standard SharePoint Server 2010 by providing exact counts of content.  This is very important and, quite honestly, what we maybe should expect within the base product.
  • Contextual Search
    Contextual search allows profiles to be setup such that the search results that are returned are based on who search for them.  The idea here is that, a sales director who searches for ‘SharePoint Projects’ is probably interested in very different results to perhaps a technical consultant like myself who issues the same search.
  • Dynamic Sorting
    Sort on any managed property, very important and very flexible.
  • Visual Best Bets
    A cool feature, provides the ability to integrate imagery into best bets to provide visual signposts to content

So there we have it, from Search Server to FAST, there’s certainly something for everybody.  The main drivers around what you should implement are primarily around scale, resilience and functionality.  If you have a small departmental site, or are running SharePoint Foundation 2010 for a couple of hundred users in a collaboration scenario, consider bolting on Search Server Express 2010.  You will no doubt be surprised at the level of functionality you get for effectively very little effort.  If you need to scale and add additional functionality, and really build a true Enterprise Search application, consider SharePoint 2010.  FAST Search, let’s be honest, is going to be relevant for the minority of organizations due to pricing.  However, if you have serious scale-out requirements, a very large corpus (billion+ items) or want to offer your users the highest level of features and integration then you should consider it.

Advertisements