MarkLogic, FAST, Categorical Affirmatives, and a Direction Change

July 5, 2011

I weakened this morning (July 4, 2011) with a marketing Fourth of July boom. I received one of those ever present LinkedIn updates putting a comment from the Enterprise Search Engine Professionals Group in front of me.

image

The MarkLogic positioning exploded on my awareness like a Fourth of July skyrocket’s burst.

Most of the comments on the LinkedIn group are ho hum. One hot topic has been Microsoft’s failure to put much effort in its blogs about Fast Search & Transfer’s technology. Snore. Microsoft put down $1.2 billion for Fast, made some marketing noises, and had a fellow named Mr. Treo-something talk to me about the “new” Fast Search system. Then search turned out to be more like a snap in but without the simplicity of a Web part. Microsoft moved on and search is there, but like Google’s shift to Android, search is not where the action is. I am not sure who “runs” the enterprise search unit at Microsoft. Lots of revolving door action is my impression of Microsoft’s management approach in the last year.

The noise died down and Fast has become another component in the sprawling Shanghai of code known as SharePoint 2010. Making Fast “fast” and tuning it to return results that don’t vary with each update has created a significant amount of business for Microsoft partners “certified” to work on Fast Search. Licensees of the Linux/Unix version of ESP are now like birds pushed from the next by an impatient mother.

New MarkLogic Market Positioning?

Set Microsoft aside for a moment and look at this post from a MarkLogic professional who once worked at Fast Search and subsequently at Microsoft. I am not sure how to hyperlink to LinkedIn posts without generating a flood of blue and white screens begging for log in, sign up, and money. I will include a link, but you are on your own.

Here’s the alleged MarkLogic professional’s comment:

Many organizations are replacing FAST with MarkLogic. MarkLogic offers a scalable enterprise search engine with all the features of FAST plus more…

Wow.

An XML engine with wrappers is now capable of “all” the Fast features. In my new monograph “The New Landscape of Enterprise Search”, I took some care to review information presented by Fast at CERN, the wizard lair in Europe, about Fast Search’s effort to rewrite Fast ESP, which was originally a Web search engine. The core was wrapped to convert Web search into enterprise search. This was neither quick nor particularly successful. Fast Search & Transfer ran into some tough financial waters, ended up the focus of a government investigation, and was quickly sold for a price that surprised me and the goslings in Harrod’s Creek.

You can get the details of the focus of the planned reinvention of the Fast system and the link to the source document at CERN which I reference in my Landscape study. A rewrite indicates that some functions were not in 2007 and 2008 performing in  a manner that was acceptable to someone in Fast Search’s management. Then the acquisition took place. The Linux/Unix support was nuked. Fast under Microsoft’s wing has become a utility in the incredible assemblage of components that comprises SharePoint 2010. I track the SharePoint ecosystem in my information service SharePointSemantics.com. If you haven’t seen the content, you might want to check it out.

Isn’t XML Is More Than Just Search?

MarkLogic’s system can indeed perform findability functions. MarkLogic can also deliver business intelligence applications, mash ups, computational feats beyond a key word search system, and perform data transformation. Publishers can stuff XML into the MarkLogic service, write scripts, and output “new” assemblages of the content in the MarkLogic server. But the core of MarkLogic remains a solid, XML repository enhanced with nifty code crafted by MarkLogic’s engineers.

image

Here’s one of my favorite graphical overviews of XML. Source: http://www.wdvl.com/Authoring/Languages/XML/XMLFamily/BigPicture/bigpix20a.html

Is MarkLogic a search system that delivers “all” of the Fast Search functionality? I suppose one can make that claim. Marketers of search and content processing make new claims every time there is an off site meeting or when someone with a stake in the company complains about revenues. MarkLogic underwent a management change, and I have a hunch that the reason may have to do with revenue.

Ergo: position MarkLogic as an enterprise search system.  Easy, tidy, and for me quite a surprise.

Enterprise search is not exactly setting the world on fire. The companies that are racking up sales include the giant Autonomy which offers a range of content centric solutions, Exalead with its well received search based applications and CloudView method, and Endeca with its database centric eCommerce and faceting system.

Are There Tough Challenges for MarkLogic in the Search Sector?

Can MarkLogic win sales from these three giants? Probably. Can MarkLogic win sales from licensees of Microsoft Fast? Probably. The challenges which I see include:

  1. Competition from tie ups like that between Cloudera and Digital Reasoning. (I cover this market sector in my Inteltrax information service. Most organizations don’t want the 1998 Fast Search type of findability. Organizations know that search does not deliver and many are looking at quite different approaches to information processing.
  2. Exalead has considerable momentum with its search based applications approach. In fact, I was asked to work on a 2012 conference which will feature presentations about the impact of the Exalead approach. Dassault, arguably one of the world’s leading engineering firms, now owns Exalead, and the firm’s remarkable CloudView approach is redefining “findability”. For an update on Exalead, you might find the interview with Laurent Couillard useful. Mr. Couillard is the new CEO of Exalead. (If my understanding of Exalead is accurate, MarkLogic’s approach is quite different.)
  3. Autonomy has tens of thousands of customers. Quite a few companies have looked at Autonomy’s customer list and concluded, “We can go after these outfits.” That is not a new idea, and if the last five years are any indication, Autonomy’s customers are often reluctant to dump Autonomy and embrace the next big thing. Most search vendors experience churn, but outright appeals to switch fall on deaf ears. Maybe MarkLogic will catch Autonomy’s attention? What then? Competing with Autonomy can be interesting indeed.
  4. Endeca continues to show resilience. Despite the age of the core code base which has been updated over the last decade, Endeca now offers customer support solutions, business intelligence solutions, faceted search, and eCommerce search. Endeca remains a tough competitor because it sells in a way different from other vendors. (Nope, I won’t reveal Endeca’s secret sauce in a free blog post. We are in an economic crisis, and I sell information, gentle reader.)
  5. The start ups, repositioned companies, and the interlopers from a far off land (Megaputer, anyone?) MarkLogic has a great core technology, but when it comes to licensees who are interested in the latest and greatest, there are many choices and more coming each day.

Is There a View from Harrod’s Creek?

My observations, before I forget them:

First, I think this “MarkLogic as a search engine” is a possible signal that the firm’s core XML business is underperforming with regard to revenue. The problem is that search is not exactly the new kid on the block and the big outfits may be wiling to license a new system for a project. Displacement of an incumbent can be an expensive and time consuming sale if the information I gathered for my Landscape study is accurate.

Second, I believe that marketing is now packaging MarkLogic’s core technology without making clear the upside and the downside of what I call a “commitment to XML.” XML can deliver some significant benefits, but there are some learnings, costs, and technical issues which must be internalized and addressed. Yep, learning curves still exist even for enthusiastic 30 year old marketing “experts”.

Third, search is just not exciting. Google’s “Inside Search” focused on Android, not search In fact, when my publisher, Pandia.com, and I talked about my new monograph’s title, we agonized over the phrase “enterprise search.” People want to do “social search”, perform “business intelligence”, and “discover” information without running queries or doing much work at all.

MarkLogic has now positioned itself as the equivalent of Fast Search & Transfer, what now seems to have been a deeply flawed product in the pre acquisition days of late 2007 and early 2008. Surely a more apt comparison is possible? Maybe the lure of licensing XML technology to SharePoint licensees is perceived as a slam dunk by MarkLogic’s management team? In a flash, MarkLogic is now an enterprise search system, not an XML server system and method. MarkLogic is now taking a step down a path that may be filled with thrills and chills. This is a path that Entopia, Delphes, and many other firms have taken.

Net net: This modest LinkedIn post by a new MarkLogic professional may mark a sea change in the company’s positioning of its flagship XML system and method. This is a possible shift that is definitely worth monitoring.

Stephen E Arnold, July 5, 2011

Mr. Arnold is the author of three studies of Google’s search technology, the first three editions of the Enterprise Search Report, a study on next generation content processing for the “old” Gilbane Group, and Successful Enterprise Search Management, co authored with Martin White. Mr. Arnold’s most recent study, The New Landscape of Enterprise Search, is available from Pandia Press in Oslo, Norway, coincidentally the “home” country of the original Fast Search & Transfer Web search system.

Comments

One Response to “MarkLogic, FAST, Categorical Affirmatives, and a Direction Change”

  1. Casey Green on July 6th, 2011 2:00 pm

    Hi Stephen –

    As you know, we have done quite a few very large Enterprise Search implementations on Autonomy and began partnering with MarkLogic in 2009. While I can agree with much of what you describe in your post, particularly in your assessment of the Ent Search space as a whole, I do think you’re off a bit on your specific conclusions with respect to MarkLogic.

    1. The gentleman who posted in the thread you linked to was at MarkLogic at the time we began partnering with them, thus it is a bit of a misnomer to view him as new or to extrapolate his post to be reflective of the general thinking of new management there.
    2. We’re not seeing any evidence of a push by MarkLogic to reposition directly as “the equivalent of FAST” or as a Search vendor. In fact, mouse over the “Solutions” nav on their home page …. do you see the nav for Enterprise Search under that? :-)
    IMO, this is more a case of ML simply being a very good search platform that, due to the “swiss army knife” characteristic to which you referred, can provide additional benefit that the search-specific platforms cannot. For an organization committed to moving their enterprise search off a platform (FAST or other), the notion that the replacement platform can allow the search capabilities to be replaced *and* serve as a platform for substantial applications to be created for dealing with unstructured & semi-structured content can be rather compelling in the right situation.

    I do think you’ve nailed the challenges quite well —- those are tough challenges for MarkLogic and any other vendor who might desire to wander into the Enterprise Search market. And if this were a case of MarkLogic truly targeting search for the sake of search alone, I’d be scratching my head too. In context, though, I think it is a savvy opportunistic move by them, a way to gain a foothold into organizations that might stand to gain tremendous benefit from the other things ML does so well. Nothing more, nothing less.

    BTW, MarkLogic’s formal view notwithstanding, Avalon believes the MarkLogic platform is a very solid choice for Enterprise Search. Sort of immaterial to the context of your post, but I felt compelled to share our view.

    -casey