When Is an Appliance a Business Solution?

By Foster D. Hinshaw, CEO and Founder

A data warehouse appliance is a brilliant packaging of innovative technology that’s focused on the business user’s experience. This is analogous to an iPhone that directs its advanced technology to enrich the lay user’s interactions.

“Rarely do organizations consider “what if” or transformational scenarios. What data sets are currently off-line or not being acquired from partners simply because of technical limitations?”

Two Views of the Same Goal

A business evaluation should complement the IT department's evaluation criteria for a data-access solution. The IT experts might touch on some of the same topics but their answers will be different.

– IT asks, "Does it scale?"
– Business asks, "How much data can I work with? Can I get to all the data that we've collected?"

– IT asks, "Does it support mixed workload?" translates into business asking,
– Business asks, "Can I run scheduled reports, have analysts crunch what-if scenarios, and my sales people looking at up-to-date data without stepping all over each other?"

– IT might ask about the migration process.
– The business side of the house wants to know how soon the solution can be in place and how quickly users can get up to speed.

– IT needs to leverage existing skills in order to apply the technology without extensive retraining.
– Business wants existing applications to run, only faster and with more data, and the assurance that new ones will launch in short order.

What are Data Warehouse Appliances For?

A couple years ago, "data warehouse appliance" was a term rarely heard outside of the data center. Now it might even be used in the corner office when the business side of the house is looking for better access to Business Information. What's changed? First, there is a lot more data and it's growing exponentially. Even smaller organizations have terabytes of data coming in through Web sites, manufacturing processes, tracking technology along with government mandates. Second, BI has become a standard mechanism to derive insight and value from data. Many more companies understand the advantage of truly knowing what's going on with operations and customers.

The key to an efficient BI solution is ensuring that data is captured and remains easily accessible. The data warehouse appliance (DWA) was invented to simplify the enormous complexity around stretching legacy technology to support new functions. Before the appliance, the common solution was to build data marts tailored to specific data sets or usages. As data marts proliferated, managing them in a holistic way became even more complex. DWAs simplify the process of implementing and changing data-intense applications. They also benefit IT by streamlining implementation and maintenance and lowering the total cost of ownership for the entire database and storage infrastructure.

We're seeing a fundamental change in the appliance marketplace. In its early days, the DWA was a performance analytics platform. Its new challenge is to bring all the benefits of an appliance to bear on mainstream data warehousing. (Mainstream data warehousing requirements run the gamut from better reporting, dashboards, operational BI, compliance, to data-driven personalization and trending, which are the fundamental engine of business momentum.)

How Do I Select the Right DWA?

When buying technology, even if you're buying it to solve a specific business problem, the same rules apply as with any other technology purchase. It is of course important that the appliance can "fit in" your data center and enterprise infrastructure, that it meets resiliency and energy efficiency criteria. Most important, however, is that it addresses the business requirements.

Following are some simple questions to consider. The questions are simple-the answers might not be.

How much data do you want to work with?

Not how much data do you now work with, but how much data do you anticipate wanting to use. Often current state plus a percentage of growth is the basis for deciding on the amount of data needed in the future. Rarely do organizations consider "what if" or transformational scenarios. What data sets are currently off-line or not being acquired from partners simply because of technical limitations?

Many of today's appliances have massively parallel processing (MPP) architectures, an industry standard for handling large data sets that provides the most scalability. MPP solutions allow you to remove the blinkers and consider what can be achieved if all your data assets are available. If you anticipate using more than 3 TB of data, look for a massively parallel processing (MPP) appliance.

How many ways do you want to use your data?

The ways you want to access data, i.e., the types of user profiles you want to serve, will further narrow the appliance options. DWAs had their origin in complex analytics serving a small set of users usually running ad-hoc queries against historical or static data. They have since evolved to satisfy general reporting needs as well as application-specific requirements.

If you plan to have various users interacting with the same data at the same time, your appliance should support simultaneous mixed workloads. Mixed workloads and the ability to load and query data at the same time are essential if you expect the data on the appliance to be available 24x7.

If you want to refresh data on the appliance frequently, you will care about how it performs when loading data. If you want to insert, update, or delete data on the appliance, make sure that level of transaction is supported. Some appliances assume that you will only load data or overwrite it.

Do you want to leverage existing databases?

To minimize change, especially for the business end-user, an appliance will allow you to transfer as much of your current databases' logical structure as possible. Migration to the new system should be as transparent as possible for users and database administrators alike. Transparency with existing databases lowers your administrative overhead and simplifies deployment.

It's not enough to ask an appliance vendor how complicated the migration process is. Consider the end result. How much will various user types have to adjust the way they interact with information? How much of that adjustment benefits the organization? How much is a distraction? How long will it take to regain a level of expertise?

When should I bring in a DWA?

When you want to serve a diverse user community with differing needs, have more flexibility in accommodating data growth, and anticipate changing ways to use that data. During the evaluation keep business value and your specific goals foremost. Just as with the iPhone, the business value of the DWA trumps technical whiz-bang, appealing as can be.