Microsoft drops open-source birthday gift with FAST
As birthdays go, it's not a bad present. Microsoft's decision to sacrifice FAST's Enterprise Search Platform (ESP) development on Linux and Unix for Windows potentially gives open-source search providers like Lucid Imagination a free pass.
In announcing the news Monday, FAST chief technology officer Bjørn Olstad was clearly cognizant of the implications of the move. Olstad tried to preempt the potential customer hemorrhage with a standard commitment to interoperability with Linux and Unix and vague references to some "hosted solutions."
"Many of our customers run FAST ESP on Linux and Unix today, and we recognize that our future focus on Windows means change," Olstad wrote. Microsoft is giving you until 2018 to stick with the existing versions of ESP - the date when extended support ends.
Olstad is right to be concerned. Customers interested in open-source search are moving to Apache Software Foundation's (ASF's) Lucene and Solr, and Lucid has spend the first year of its life charging people to provide support along with consulting to get the move right.
On Tuesday. Lucid's chief executive Eric Gries claimed 80 per cent of the FAST's customer base ran on Linux or Unix, and Microsoft's decision will simply mean those who'd been previously unsure, concerned about the future under Microsoft, or interested in using an open-source architecture like Lucene or Solr will now move elsewhere.
"This news is the final nail in the coffin," Gries said of FAST. "I'm puzzled by the thinking at Microsoft. But for us it definitely helps. Customers who were on the fence thinking about staying with FAST and Linux and moving to Solr - their decision is very clear."
Lucid counts those who've moved to Lucid or Solr in the last year as Fortune 500s that don't want be tied to a single company's roadmap or paying expensive licensing fees. Microsoft's plan with FAST is to integrate the search capabilities into Office and SharePoint 2010. Since January 2009, Lucid has picked up Ford, Nike, Sears, and Macy's.
Lucid's average annual subscription charge is $30,000, which gives customers pre-integration and support for these ASF-based projects. It targets those who like open-source but don't want to get their hands dirty and that want the comfort of having somebody to call or help build their system. Lucid's services include updates and sweeps of customers' installations to look for bottlenecks and check the accuracy of the data.
There are other alternatives to those who don't want to become dependent on Window, Office, or SharePoint or be locked into a single vendor such as Microsoft or even rival Oracle that offers its Secure Enterprise Search. These options include Google's Search Appliance and also Sphinx.
While Google might have brand recognition its appliance falls down because it's closed - you can't tamper with it. Also, Google charges increase as you add more documents, which hurts organizations that want to grow. Among Lucid's customers who've dumped GSA for Solr are investor's web site The Motley Fool.
As for Oracle's Secure Enterprise Search, you're again locked into a single vendor offering a closed product. And then there's the money: Oracle's list price is $34,500 per processor. Gries, meanwhile, called Sphinx more "generic" search that's designed to work on multiple databases.
If there's a niggling concern in all of this, it's the fact Microsoft felt confident enough to cut ESP's Unix and Linux users in the first place. Microsoft dropped a cool $1.2bn on FAST in 2008, making it one of just four acquisitions in Microsoft's 30-odd-year history to break the billion-dollar mark. Only Visio on $1.3bn in 2000, Navision on $1.3bn in 2002, and aQuantive for $6.3bn in 2007 beat FAST.
That's a lot of money to feel you could gamble on converting Linux and Unix users to Windows or simply throw away by dumping a large proportion of the customer base, if Gries' 80-per-cent number is correct. Even the politics of putting Windows first would make little business sense.
In 2007, the year before Microsoft's acquisition, FAST was hit with revenue recognition problems that challenged its previous claims of 60 per cent year-on-year growth. Prior to acquisition, FAST reported net losses on revenue that was falling - this before the sub-prime bubble that everyone agrees helped kill IT spending in 2008 and 2009 was far from bursting.
The figures suggest that maybe the Unix and Linux business wasn't paying so well for FAST, that Microsoft is justified in doubling down on Windows, and that Microsoft bought FAST purely for its search technology capabilities instead of its customer base.
This means potential trouble for Lucid, which is faced with the classic problem of persuading users of open-source code to pay it for those value-added consulting and support services. But Gries is confident users leaving Microsoft and eschewing Google and Oracle will want the proven capabilities of Lucene and Solr and find their way to Lucid.
Gries claimed one reason Lucid has done well and is able to beat Google among coders building search - and why it will attract former FAST users - is because Lucene and Solr are 10-year-old, proven technologies. "The largest install based in enterprise search is Lucerne," Gries said
Lucid imagination, meanwhile, brings the confidence of a support partner - especially if search is important to the customers' commodity and without it, they risk losing money.
"We don't have to persuade people - we have a number of leads coming into us," Gries said. "The basis is the fact this software is not easy. A lot of things can go wrong, like the query passer.
"You have many, many ways to cause trouble, but when you're in a contract with a commercial company there's somebody to talk to. When it's a project there's nobody to talk to," Gries said. ®