The scope for appliances

Since there is a generic article about appliances I had better say before I begin that what I intend to discuss are appliances that deal with data as opposed to, say, appliances that are concerned with networking.

Within the data realm then, there are number of appliances already available: for database monitoring (software solutions from Symantec, and there is a virtual appliance from Embarcadero), for data warehousing (various vendors, such as Netezza and DATAllegro), for EAI (enterprise application integration from Cast Iron Systems and others) and for Business Intelligence (CeleQuest). There may be others that I am not aware of. The question is: can we draw any lessons from these examples of appliances and, on the basis of those lessons, suggest other environments where appliances might usefully be applied?

It seems to me that appliances address at least two out of three main requirements: they provide (often very significantly) improved performance, they reduce complexity and make management and administration simpler, and they are (often much) less expensive than traditional solutions. They are also very simply to implement. In many cases they make things possible that were not previously viable at all.

Let us take the examples I have given:

• Data warehousing appliances meet all of the requirements detailed above and I have written about this often enough to not have to repeat myself. The only proviso one would make is that they become less simple as they move into the enterprise data warehousing space.
• Database monitoring replaces and extends database auditing. In practice, database auditing is rarely, if ever, done because of the performance overhead involved. Using an appliance here also ticks all of the other boxes.
• EAI appliances focus on reducing complexity rather than anything else and this in turn reduces costs. There is not, on the other hand, the same emphasis on improved performance.
• BI appliances resemble EAI appliances: again, the focus is on reducing complexity for a lower cost of ownership and wider adoption.

There are two main foci therefore: improving performance and reducing complexity. However, both of these have knock-on benefits. For example, improved performance means that you require less hardware, which means reduced costs both directly in terms of the tin you purchase and indirectly in terms of running costs, while reduced complexity leads to further advantages such as reduced administration and management time and costs, faster implementation (which in turn leads to faster time to value), and so on.

Ideally, you would want to improve performance and reduce complexity but one may be enough on its own. Where then are there other areas that might suitable for the application of appliance technology?

One obvious point is that two of the areas mentioned both pertain to databases and one notable omission is that there are no appliances currently on the market that either improve the performance and capacity of transactional databases (not to mention hybrid environments where you want to run queries and reporting off the transactional database) or reduce their complexity. Or, for that matter, reduce the overall costs involved. So, that’s one distinct possibility.

Another is in event processing. I mention this because I happen to know that there is one company in this area that is working on an appliance-based approach to this. Moreover, if you think about algorithmic trading for a moment, then this is the automation of complex buy/sell algorithms: if these can be built into an appliance then, potentially, anything that is algorithm-based can have the same approach applied to it. Why not, therefore, a data mining appliance, a credit card fraud prevention appliance or a matching (data quality) appliance? And, if we are talking about data quality, why not an appliance for associated activities such as ETL (extract, transform and load), data federation or MDM (master data management)—especially bearing in mind that there are already EAI appliances on the market.

No doubt readers will have other ideas as to where appliance technology might be usefully deployed and I would welcome any appropriate feedback. In any case, I will return to this topic next month when I will be discussing one of these possible scenarios in more depth.