Search This Blog

Saturday, August 21, 2010

BI Appliances: Why does it have to be so hard?

There's lots of buzz these days on appliances in the Business Intelligence and Data Warehouse space. Twenty-eight years ago, Teradata established the first reference pattern of a database machine, later to become known as a data warehouse appliance when "data warehouse" caught on. Netezza made the term "appliance" popular in 2000 with it's marketing campaign. Now several database vendors (Greenplum, Asterdata, Vertica, Paraccel, Dataupia to name a few) offer their product as an appliance either with their own DBMS software or combined with a pre-defined hardware configuration from a hardware partner. According to Gartner, roughly 50% of of the vendors on the 2010 Data Warehouse Database Management Systems Magic Quadrant are considered appliances.



So today's concept of an "appliance" in the data warehouse & business intelligence space appears to be centric to the database management function. That's a great start, but how do I get the data into these database appliances and how do I get it out for consumption and to take action? What about the data integration, data visualization, data analytics and operational performance management functions? Will we see appliances pop up in the data integration, visualization, analytics and performance management functional areas as stand alone appliances or will we see the database management appliances expand their scope to include these other mission critical functions? Will we find ourselves inter-connecting functional-silo appliances or will we see an evolution towards bundled "BI in a Rack" multi-functional appliances? Will the multi-function appliances contain all of the compute, storage and network services necessary to seamlessly enable the inter-operability across the data integration, data management, visualization, analytics and performance management functions? Can I fill up cages with "BI in a Rack" functional appliances to facilitate “sprawl” (multi-node grid processing across these functions)?



Presuming for a moment that the short answer to my barrage of speculative questions is a resounding yes, yes, yes, let's ponder the next question. Can I simply subscribe to these functional "BI in a Rack" appliances as a set of services? Dare I say it...can I get Business Intelligence as a Service?

I'd like to hear your thoughts.

4 comments:

  1. Mike, there's quite a body of thought on these topics and it's fair to say that the market is answering them right now: yes. Every combination you suggest is being tried and most are finding some audiences.

    Appliances are about ease and time to value. Low maintenance, quick startup, minimal care and feeding. Vendors who make that happen are succeeding.

    ReplyDelete
  2. I'd agree with Merv. Yes, and yes. SAP has Explorer Accelerated, which is an appliance supporting ad hoc visual analysis. IBM Cognos has Go Now! (or something like that) which is an operational dashboard appliance. But for the foreseeable future, the broad-based appliance platform will dominate.

    ReplyDelete
  3. It sounds like that you're really eager to have an integrated BI solution that's simple and does the work (or services in your term) without complexity of layering.
    Is that right?

    I also agree with that.

    ReplyDelete
  4. Hi icm,


    I am eager to see solutions, indeed. With regard to the complexity of layering, with the right level of admin consoles to automate and mitigate the complexities, I rather like the flexibility layering can offer to the solution. hmmm, sounds like a good topic for my next post.

    ReplyDelete