Microsoft Realizes Its $1.3 Billion Challenge

February 7, 2010

My wife watched the film Titanic last night. I enjoyed the picture but the scenes that stuck in my mind were the calm officers who were on a sinking ship. Nifty but not too many survivors. I fell asleep before the film ended. I assume everyone was saved, but that was a motion picture, not real life.

Last week there were some icebergs spotted and a big ship speeding among them.

Phone calls and emails from frantic 20 somethings who want to write a story before heading to yoga. A couple of clients pinging me about the future of enterprise search. Even a fax from a UK client who wanted a link to the Fast Linux integrator / consulting outfit Comperio, where some of the original Fast engineers now work.

image

Ships in iceberg territory.

Let’s take these inquiries—informed, desperate, meaningful, and clueless—and probe each.

First, a rumor surfaced about 10 days ago that there was a shakeup in the Microsoft Fast unit. The rumor I heard from a contact in Scandinavia was that Microsoft was backing away slowly from Fast’s Linux and Solaris versions of ESP and I thought I heard a reference staff who were told “don’t let the door hit you on the way out”. Apparently those bidding farewell wore penguin pins on their parkas.

image

I then read a pretty good write up by Kurt Mackie in RedmondMag.com called “Fast Search Will Be a Windows Only Product.” In my opinion, the most interesting comment in that write up was attributed to the fellow running Microsoft search, or at least one of the people running Microsoft search:

We will always interoperate with non-Windows systems on  both the front- and back-end, Olstad wrote. Our search solutions  will crawl and index content stored on Windows, Linux, and UNIX systems, and  our UI controls will work with UI frameworks running on any operating system. [Emphasis added]

Don’t you believe in categorical affirmatives? Well, I am careful with broad statements that imply “diamonds are forever”.

I don’t know what the staffing situation is, but I was surprised it took Microsoft’s business experts more than 18 months to realize the complexity, cost, and legal issues associated with the Fast Search & Transfer SA’s technology. The Fast company cost MSFT about $1.2 billion. Then there was the December 2009 departure of the Microsoft CFO on whose watch the deal went down (Chris Liddell). The whole acquisition struck me as an possible indication that Microsoft’s acquisition team was not as informed as it should have been with Fast ESP (enterprise search platform). I concluded that Microsoft did not have real ESP (extrasensory perception) when it decided to buy Fast Search. Then I was pitched enterprise search as a user experience with a tile of images is not going to deal with the brave new world of search enabled applications, eDiscovery, and seamless access to structured and unstructured data. Interface is one thing. Meeting users’ information needs in the enterprise is quite another.

image

Source: http://ar-d.com/images/emailblast/bing%20visual.png

Second, I documented in this Web log one public statement that a Fast ESP customer was done with the product due to lack of service, technical capabilities, etc. You can read that story to which I pointed in this Web log as “Microsoft Fast Questioned by Ayna”. Most licensees just grandfather a search system and move on. What made this story interesting is that Ayna went on the record on its Web site specifically pointing out the problems the company had with the Microsoft Fast system. Remarkable. This addled goose’s rule of thumb is that where there is one really annoyed customer there are probably lots more sitting in their cubicles wondering how to extricate themselves from a muddle.

Third, the whole Windows plumbing versus Unix plumbing does not require much technical acumen to understand. Trying to support an aging search technology with layers of acquired code, home brew scripts, open source software, and licensed “components” is a challenge in my experience. Now add engineers steeped in the arcana of Dot Net. The mixture triggers long meetings and slow, slow progress.

What happened before the Microsoft buy out of Fast Search in the period between 2002 and 2006:

  • Licensees discovered that the architecture of Fast ESP translated to some set up and configuration issues. Make a change here and something weird happens there. No big deal unless you have some overconfident 20 something make five or six config changes at once. Now that’s exciting and can only be unwrapped by with some serious work, roll backs to a last known good state, and – heaven help me – a rebuild of the indexes.
  • Some clients seemed to be reluctant to pay Fast’s invoices until the system met the client specifications. Heck, I heard of one client who did not pay until a “real” Fast engineer turned up who knew whom to call in Norway. Get enough slow pays and you hear, “Houston, we have a problem.” The hope of no return from a moon mission is more pleasant than what happens when multi billion dollar outfits get annoyed and start talking to their solicitors.
  • The guts of Fast Search remained a system and method to index Web pages and deliver search results “fast”. In an enterprise, the Web way is not too helpful.

Is this Microsoft action regarding Linux and Unix a surprise?

Not to me.

I believe it was a surprise for those who don’t know much about search.

Is this search challenge fixable?

Yes.

The solution I would choose would be the purchase of an enterprise search system that works. Without a reset, I would need to the wisdom of Solomon, the riches of Croesus, and the patience of Job to deliver a high performance, useful enterprise search system with the Microsoft Fast toolset.

I want to close with a comment made by an azure chip consultant who made a comment at a conference I attended in mid 2009: “Search is simple.”

Sure, anything a person does not understand can be explained as simple or magic. “Simple” applies not to search. Simple describes the “expert” who made this statement.

Why does Google hire wizards?

Easy question. Search is hard.

Stephen E Arnold, February 7, 2010

No one paid me to write this. I suppose if I had a sales sense I could have found someone who would have paid me not to put down my opinions. I will report this imbalance to the Taxpayer Advocacy Panel. It’s that time again, folks.

Comments

2 Responses to “Microsoft Realizes Its $1.3 Billion Challenge”

  1. Lucid Imagination » Shocked? Microsoft holding FAST to Windows and dot-net on February 7th, 2010 10:09 pm

    [...] search engine on Unix reject a standalone search engine altogether? As Steve Arnold observed, it must be tough. If even the FAST development team itself needed to drop everything to make FAST work on dot-net, [...]

  2. FAST to abandon Linux and Unix | Cominvent AS - Enterprise search consultants on February 8th, 2010 8:52 pm

    [...] = 'cominvent';In a recent blog post by CTO Bjørn Olstad, referenced by CNet, Beyond Search and Norwegian digi.no today, FAST announces that ESP 5.3 is the last version of their Enterprise [...]