Data Warehouse Appliances: definition & evolving place in the market

"It depends on what the meaning of the word 'is' is." - Bill Clinton, President of the USA (1993-2001)

As our first act of commenting on the industry, we'd like to address a topic that has seemingly stirred up quite a bit of emotion and controversy of late: just what IS a Data Warehouse Appliance (or "DWA" for the acronym-inclined)?

But first, the punch line: it's not definition of what a DWA is that matters, but - taking things a bit further - what deploying a DWA will mean to customers who use them in their analytics and BI scenarios.

Plenty of opinions to go around
In the world of BI and data warehousing, if there's one area that's nearly become an industry segment unto itself these days, it is the field of those industry analysts, pundits and other experts trying to define just what a "Data Warehouse Appliance" really is.

It's no wonder. Over the past three-plus years, the Data Warehouse Appliance market has blossomed, indeed. It has become a significant and growing segment of the data warehouse systems market. Since Netezza's initial entry & coining of the terminology for this space in 2002/2003, a number of new entrants (from industry behemoths to the smallest, new start-ups) have tried to stake their claims to it. Enter the industry pundits, to help us all by defining and making sense of things.

A growing market segment that's "here to stay"
Claims from long-standing incumbent data warehouse systems providers notwithstanding, I would hazard that in analyzing clippings from before 2002, you would be hard-pressed to find any references to a DWA in the media or analysts' market predictions about the future. Certainly, companies have built systems expressly for use as data warehouses in the past but my searches have not revealed any claims on the notion of an "appliance" in this space before the dawn of the 21st century.

Now we have an established and growing market category for data warehouse systems. According to IDC's Dan Vesset, "IDC expects the market for DW appliances to grow at a CAGR of 70% over the next 5 years from the estimated 2005 level of $75 million."

How does Netezza see the definition?
Today, with the definition of a data warehouse appliance is seemingly crying out for clarity, with a growing number of vendors' marketing claims making things more hazy. As the pioneer and the recognized global leader in the DWA market with over 75 paying customers under our belt, when it came to defining just what a DWA is we felt, "Who is more qualified than us?" - so we decided to weigh in with our views, as follow.

The key operable point here is that a DWA is fundamentally performance-driven. It allows businesses to have more clarity and more depth of analysis across ALL of their data much faster than they have been able to in the past. The fact that a DWA also delivers simplicity and economy is putting that performance well within reach for most enterprises.

Simply put, a true data warehouse appliance will put the high-performance of a super computer into an enterprise's data center at a cost-effective price point. And it will do so with an ease of installation, use and maintenance that will make much more powerful analyses and more rapid development of ideas possible than other systems can provide.

DWA Measures - according to the experts
A recent TDWI survey indicated a majority of members surveyed understand that a data warehouse appliance is defined as server hardware and database software built specifically for data warehousing - not just a bundle of commodity hardware and generic software - and that the benefits of this approach are greater performance and lower cost. But there have been many attempts to define and measure the "goodness" of DWAs. The table below contains but a few.

Furthermore, Robin Bloor and Philip Howard of the Bloor Group have set off down a path to make the definition and benefits of various DWA approaches more clear - aiming to do so even more completely in early November.

Why not bundles or "balanced" blade-servers?
Simply grouping multiple systems in loose affiliations won't really answer the mail here. The inefficient movement of huge blocks of data for analysis adversely limits performance; and the complexity of managing disparate systems and each of their upgrade and compliance paths alone will make this approach difficult to manage. But so will the fact that the systems will evolve independently and not necessarily in alignment with one another.

Unless it is rearchitected to specifically address data warehousing, a "shrink-wrapped" bundling of products from among a major player's broader suite of systems will be similarly performance and operationally limited. And it too will have to deal with the effects of each product's evolution pulling in a different direction.

Where is this all going & why does the DWA definition even matter?
The real issue of course, is that, to enterprise customers, the "true definition" of a DWA doesn't really matter at all; what matters is the impact that taking a DWA approach to their data warehousing needs can have on their businesses.

What we've seen from customers' use of the NPS product family is that DWAs are changing the way businesses use their warehouse data today and in the near term, including the following -

• enabling deep, unconstrained analytics on all of their business data, even in extremely busy mixed-workload scenarios;
  
• changing the way they think about the staffing to support it and opening up the development of whole new advanced analytics applications;
  
• changing the way they purchase data warehouse infrastructure; and
  
• helping mid-tier business solve critical data warehouse needs in compact, fully-contained business solution appliances.

In the longer term, DWAs will fundamentally change the way people operate their businesses.

Look for us to provide more on this and other of our views about the future of DWAs in upcoming postings.

Share:  

Posted by Phil Francisco at January 1, 2007 1:30 PM

Comments

Post a comment

Name:

Email Address:

URL:
Remember Me? YesNo

Comments: