BI

Data Transformations Complexified: What's the world coming to?

I went to a dotnet user group last week to hear a speaker on the subject of SQL Server Integration Services (SSIS), the next-gen replacement for Data Transformation Services (DTS), which is to say: "Moving data between points A and B." I'll set substantive comments about the presentation itself aside, giving benefit of the doubt to the youngish presenter trying to squeeze a lot of fundamentals PLUS an overview of the technology. That's kind of a lot of stuff, so suffice it to say the intro probably wasn't all it could be. I've been working with SSIS a bit, and I don't think it's actually possible to cover what he wanted to in that amount of time. Anyway. Kudos on the attempt.

So, my initial reaction is still my strongest gut reaction: WHYYYYYYYYY did you make DTS legacy, Bill??? I swear, I really don't need to wrap a lot around pull from Database1.Table1 to Database2.TableZ. I just don't. Very nice to have exception handling, etc, but what happened to the quick & dirty? Create a package, define some additional actions, and schedule a job that uses it... Perfect! And all from within Enterprise Manager! Gosh, how I miss SQL 2000... Of course, Enterprise Manager and the accompanying suite of tools was replaced by Management Studio - nice bundling, I have to admit - but you now also need to run a flavor of Visual Studio 2005 ("BI Development Studio" = VS2005 + different toolbars) to do basic data transformations in the preferred way.

Just when MS had the stroke of genius to consolidate the database management interface, they split off key functionality - nay, made it LEGACY - to usher in a new era of "no simple stuff! everything is Business Intelligence! Have some BI Cheerios! They're good for you!"

Silliness aside, I think this kind of efficiency (some call it "thought leadership") is too much - there was probably a way to keep both. Just throwing it out there. Get your fancy, exception-handling BI package to the serious people - and I'm only sometimes a serious person - and leave your stinking, procedural DTS in place for the masses. The result is NOT going to be, "More BI! Cha-CHING!!!," rather less reliance on SQL Server for moving data.

Enough with Saas, now there's DaaS!

DaaS (DBMS as a Service) is yet another acronym that I pray will not catch on, mostly because it doesn't add any value for anybody but the first-adopter vendors. This recent article in InfoWorld tells how database vendor Vertica is collaborating with Amazon on a move to offer its database management system (DBMS) as a hosted service on Amazon's Elastic Cloud Computing Infrastructure (EC2), a take on Software as a Service (SaaS), etc, etc. The net here is that what costs ~$150K/month to host a 500GB 1TB [edit] data warehouse inside an end-user company gets slashed to $2K/month for a 500GB store [edit]. Of course, these numbers come from the vendor and who KNOWS what creative licensing math went into the ROI calculation [see comment]. But it's compelling. [edit: a Vertica employee corrected a few details, but the story remains compelling. I encourage you to read the InfoWorld article (link above) and his comment.]

Here's the Donald Feinberg (Gartner Group) analysis:

Cloud-hosted database management systems (DMBS) hold little value to the end-user type IT and department-level organizations. [...] Another, although related to the software vendor, is when a BI app vendor puts their app on top of Vertica and sells to their clients as a package. [...] They could now sell just to the end-user (bypassing IT and red tape) and install it fast with the correct resources for the end-users on EC2, assist the end-user loading data and the users are off and running.

That's interesting - take IT out of the BI mix and you have end-users getting to their "one version of the truth" using a packaged software. So long as the tools / data can support reporting without an inherent "slice & dice your way to fun and profit!" capability, I think this is good. But the IT Analysts will remain involved in the setup of such instances until there exist Business Analysts not influenced by sales & marketing execs. Call me paranoid.