steer clear of writing about topics like the intricacies of database structures, because this magazine isn’t circulated to a programmer-level audience. You, the CIO, VP of IT, or line-of-business manager, are more interested in answers to questions like “Can it be done?” and “How much will it cost?” than “How can I manage row-level locking in an OLTP [online transaction processing] environment?”
I did get drawn into a conversation about databases recently, though. What caught my attention was the claim by Cesar Rojas, a senior marketing executive at ANTs Software, that its Data Server could save a tier-one retailer tens of millions of dollars per year in database maintenance costs. I’m naturally skeptical of such claims, but if nothing else, the conversation underscored some database issues that are worth addressing.
DB Maintenance: A Universal Expense
All businesses are database dependent to at least some extent, and in turn, require database maintenance. In retail, where margins are thin and IT projects are scrutinized more closely than they are in other verticals, database maintenance takes on a whole new level of importance. You’re expected to squeeze every penny of return out of your tech investments. One of the best ways to extend the life and value of your mission-critical POS, CRM (customer relationship management), or inventory management systems is to revisit, and likely rearrange, your databases. Making modifications to tables to enable the capture and analysis of new or different t-log data, for instance, happens at the database level (provided you haven’t deployed an overlapping application for this purpose). Peeling off unused or underutilized fields and appropriating them for other purposes increases server efficiency and enables better functionality and reporting. Thinking of it in t-log terms, are you tapping into all the value that resides in data you’re inherently collecting? Are you using that data strategically? If not, would some changes to your database enable it?
Problem is, and to Rojas’ point, database maintenance is expensive. Outsourcing it to your database vendor typically means spending up to a quarter of the total cost of your initial implementation cost in the long term. Handing the duties to an internal IT staff requires varied and disparate skill sets. It’s not uncommon for retailers to run applications that draw from separate Oracle, SQL, and DB2 databases, to name a few. Not many retailers employ experts in all of the above.
Several companies have popped up in recent years that claim to solve the database disparity problem via middleware or by storing data in a more universal structure. Many of the same companies tout significant performance improvements, thanks in part to databases that function in-memory, as opposed to disk (caching data to make it more quickly accessible). ANTs’ solution exemplifies one that runs in- memory and changes the data structure. This requires less disk space, which contributes to ANTs’ claim that it can significantly reduce database hardware needs, and, therefore, expenses. Keep in mind, there are some potential risks to your data residing in-memory. If the database server goes down, for instance, you could lose data that existed in cache but was not yet written to disk.
My advice is to take a close look at your database maintenance contract and determine if there’s room for improvement and savings. If you’re shelling out a lot of cash for a legacy system that’s underperforming, you might want to take a look at some alternative database solutions.