Searching Questions

With users increasingly using Search to find files, the hunt is on for efficient search-based filing systems. One of these is Microsoft Search Server, with its free Express version. So how well does it do the job?

Users have more or less given up heeding any advice on how to organise their documents into folders and have opted for the easier “search” method. After all, it doesn’t matter where you put something as long as you can find it when you need it. You may not approve of this approach but, as I have commented in previous columns, it is the way of the future. Microsoft in particular is moving to a more search-based filing system with each revision of Windows, but at the moment its efforts aren’t particularly coherent.

Currently, we have a desktop search system that is available as standard in Vista and Windows Server 2008 and can be retrofitted to XP. Unfortunately, for reasons that are difficult to fathom, this is very much a single desktop search and not suited to working in an enterprise setting. In fact, the difficulty of using it on network shares and the need to maintain a local index might even encourage users to store documents on their local hard disks. When Windows Desktop Search (WDS) works, it works well and users come to rely on its ability to find what they didn’t bother to file adequately.

You need to find a coherent approach to search that can be centrally administered and tailored to suit your needs. This is where the new Microsoft Search Server (MSS) comes in. A new and free-to-use version, Search Server Express 2008, is an ideal way to find out if enterprise search is for you.

Both search versions are smallish applications that you can download and install on a suitable server. They can be set to examine the contents of any network share or a range of other data sources and they create a single central index on the server. This can then be used via a web interface to locate a particular file by content or using a range of metadata – filename, creation date, author, etc. The index can be kept up to date by setting a refresh schedule for any of the data sources and any user you care to give the necessary permissions to can use the index to search for data.

Point of Access
Search Server Express is an integral part of SharePoint but you can download and install it without having to explicitly deal with SharePoint. However, its user interface and all its management are based on SharePoint. You have to install it on a Windows 2003 or 2008 server and it uses IIS to create a customised web site. You use the web site to manage the Search Server and the end user browses to the search page to conduct a search.

In some ways this might be disappointing, especially to the end user. Windows Desktop Search, after all, is built into Windows and readily available within every folder. On the other hand, if you think users are familiar with web-based searches, then the extension to searching your internal network in the same way is a plus point.

The overall design of the search engine is distributed and you can have multiple search and web front-end servers. The free Express version can only be installed on a single server, but you can still have multiple web servers providing the user interface. The good news is that the administration facilities in the free version are similar to those in the full version and this lets you find out how it all works – and of course helps sell the full version! There is also no limit to the number of documents you can index in either version. However, the free version comes with SQL Express, which is limited in the size of database it can work with. Even so, you should find the system capable of indexing 400,000 to 500,000 documents without difficulty. If you exceed these limits, you can always swap to a full implementation of SQL Server, either on the same server or on a separate one.

In Administration
All admin and configuration is performed via an admin web site. You can set up a range of search sources using installable connectors and the search engine immediately begins to “crawl” the data, examining each document and also following any links that it contains. A number of connectors are supplied – for file shares, web sites, SharePoint sites, Exchange public folders and Lotus Notes. A range of other connectors are available for download.

As well as linking to data sources, connectors can make use of “federated” search sources. Basically these are other search engines that implement the Open Search standard and effectively share their indexes with the local server. Sample federated connections are available for various news servers, Flickr, YouTube, MSDN, Wikipedia and so on. This makes it possible to provide customised search pages for the wider Internet as well as for your own intranet. However, if you don’t want to use these facilities, you don’t have to.

You can customise what is indexed by location as well as by document type. The contents of documents are examined using IFilters. If you have a suitable IFilter for a type of document, then its content, as well as its metadata, can be indexed. IFilters for all Office documents, HTML, XML and e-mail are provided. Others are available as downloads. The IFilter format is rapidly becoming widespread as it is used by WDS, SharePoint and SQL Server as a way of gaining access to the data within a document.

Once you’ve indexed a data source you can let users access the search web site. The search page looks familiar and an optional advanced search page allows users to refine their search. The results are presented as a list, complete with small abstracts in familiar “web search” form (Figure 1). Users can also request alerts to be sent to them if a query result changes.



Figure 1: Using Search Server is much like using any web search engine

As everything is based on SharePoint you can customise the user web pages using web parts either interactively or using one of the standard tools – Visual Studio or SharePoint editor. Management is made easier by a comprehensive dashboard presentation of the current state of the system. You can arrange for users to see different search pages depending on their identity and security credentials. Security trimmed results present only the documents that the user has access to.

The index can be used as it is built up and you can dynamically remove any documents that shouldn’t have been included without having to re-index. “Crawl” rules are automatically created to stop the offending documents being re-included. You can also modify crawl rules to tailor what is indexed and to make things generally more efficient. After a full index you can schedule incremental updates to make sure that the new items are added as soon as possible. Once again the dynamic nature of the index means that users are never without a search facility.
Management facilities include the ability to discover what is being searched for. This information can be used to configure the index so that searches are more successful and identify weaknesses and inefficiencies.

As well as scheduling indexing, you can schedule backups of the configuration and/or indexes. As you’re working with a web interface you can use the search facility and manage and customise the site from almost any location.

Search Me
So is it all a good idea? Well, one big advantage is that Microsoft Search Express is free to use and provides facilities that its rivals charge thousands of pounds for. In this respect if you want to try an enterprise search solution, even if you are considering an alternative, you should try the Express edition. At the very least, the experience will make your exact requirements clear. You might even find that it, or its more scalable, paid-for version, does the job well enough for you to stick with it.

Next month we’ll take a closer look at how Search Server Express behaves in practice.