Data Fusion: Not Yet, Not Cheap, Not Easy

November 9, 2015

I clipped an item to read on the fabulous flight from America to shouting distance of Antarctica. Yep, it’s getting smaller.

The write up was “So Far, Tepid Responses to Growing Cloud Integration Hariball.” I think the words “hair” and “ball” convinced me to add this gem to my in flight reading list.

The article is based on a survey (nope, I don’t have the utmost confidence in vendor surveys). Apparently the 300 IT “leaders” experience

pain around application and data integration between on premises and cloud based systems.

I had to take a couple of deep breaths to calm down. I thought the marketing voodoo from vendors embracing utility services (Lexmark/Kapow), metasearch (Vivisimo, et al), unified services (Attivio, Coveo, et al), and licensees of conversion routines from outfits ranging from Oracle to “search consulting” in the “search technology” business had this problem solved.

If the vendors can’t do it, why not just dump everything in a data lake and let an open source software system figure everything out. Failing that, why not convert the data into XML and use the magic of well formed XML objects to deal with these issues?

It seems that the solutions don’t work with the slam dunk regularity of a 23 year old Michael Jordan.


The write up explains:

The old methods may not cut it when it comes to pulling things together. Two in three respondents, 59%, indicate they are not satisfied with their ability to synch data between cloud and on-premise systems — a clear barrier for businesses that seek to move beyond integration fundamentals like enabling reporting and basic analytics. Still, and quite surprisingly, there isn’t a great deal of support for applying more resources to cloud application integration. Premise-to-cloud integration, cloud-to-cloud integration, and cloud data replication are top priorities for only 16%, 10% and 10% of enterprises, respectively. Instead, IT shops make do with custom coding, which remains the leading approach to integration, the survey finds.

My hunch is that the survey finds that hoo-hah is not the same as the grunt work required to take data from A, integrate it with data from B, and then do something productive with the data unless humans get involved.


I noted this point:

As the survey’s authors observe. “companies consistently under estimate the cost associated with custom code, as often there are hidden costs not readily visible to IT and business leaders.”

Reality just won’t go away when it comes to integrating disparate digital content. Neither will the costs.

Stephen E Arnold, November 9, 2015


Comments are closed.

  • Archives

  • Recent Posts

  • Meta