The S.O.L. Moment for BI Hobbyists

When the PC industry was young we used to build computers from build-it-yourself computer kits. The best known kit was the Altair 8800 but another was actually called SOL (maybe due to the condition in which kit purchasers/PC hobbyists would find themselves…).

Nobody buys and builds computers from a kit anymore. We buy professional products that are built by people who design, assemble and test thousands of these every day. Yet the “build-it-yourself  BI kit” is still the dominant way IT teams today buy, assemble and deliver Business Intelligence. That’s insane!

Since BI projects are built by hobbyists from tens of building blocks, it’s no surprise that their requirements so closely match “hobbyist” PC requirements:

  • More Important Requirements: Performance, cost, expandability, upgradeability.
  • Less Important Requirements: Reliability, availability, service, ergonomics and usability.

In an ideal world, BI Hobbyists would realize that assembling and operating a diverse set of technologies and products that come usually from number of suppliers (or from a single supplier through multiple acquisitions) would be overwhelmingly complex. But the fact that the pieces come from a large company typically gives the buyer the illusion of completeness and unwarranted optimism about the chance of success.

And we don’t live in an ideal world anyway. That’s why BI projects are so often delivered without the adequate “reliability, availability, service, ergonomics and usability”. And quite often they’re not delivered at all. Clearly, some BI hobbyists find themselves in similar position as PC kit purchasers used to…

These are the “S.O.L. moments” when we get calls from prospects today (increasingly from Fortune 500 types). As much as I would love to see GoodData as a part of large-scale BI projects from the very beginning, I understand that we need to prove ourselves first. I am more than happy to come to the rescue.

But isn’t it obvious to the industry that business intelligence should no longer be built by hobbyists? BI buyers should focus on business value (metrics, dashboards…) and BI projects should be built by people who design, test and deliver at least hundreds of these every day…

Making Good Data Level 1 Platform of Global SOA

I intentionally mixed two concepts in the title of this article. The first one is the concept of Internet platform as defined by Marc Andreessen here. And the second one is the Global SOA: the non-visual data and services portion of the World Wide Web. So what does it take to make Good Data a Level 1 platform and make it a good SOA citizen? Here is my list of interfaces:

Upstream APIs (REST/Atom):

Data integration processes access the following APIs to manage the data/metadata flow from the transactional systems into the hosted datawarehouse

– Physical model definition and management
– Bulk data import (CSV, XLS, XML)
– Event driven data load (feeds, ESBs)

Downstream APIs (REST/Atom/JavaScript):

Enterprise application (wikis, mashups, dashboards) use these APIs access the modeling, analytics and collaboration features of the platform

– Logical model definition and management (Attributes, Dimensions, Hierarchies)
– Report creation and execution
Metrics definition and modification
– Collaboration: tagging, comments, search

Federation APIs

Support for OAuth, OpenID, and others.

I am sure this is not a complete list and I even wonder what BI/DataWarehousing features could be provided by Level 2 platform…

%d bloggers like this: