Any organization that has deployed a business application has experienced the joy of procuring database licenses. Most database software licensing models are based on the quantity and type of compute processing cores in the underlying database server-the more cores in the processor and the more processors in the server box, the higher the cost of the database software license. Depending on the application and the business's expectations, the tolerance threshold for performance can vary. This is typically considered during the design and testing phases of an application deployment lifecycle.
Once the application goes into production and data accumulates in key areas supporting mission critical business processes, performance starts to take a hit. Performance tuning is an art, requiring the skill and experience of the highly-coveted and -paid performance database administrator (DBA)-an employee who has been, incidentally, identified by industry research as being problematic to retain. That performance guru will add database indexes, rearrange queries, or add additional database objects solely targeted at improving application performance. At some point, performance tuning will only get you so far and returns will start to diminish. DBAs will request more processing power to meet service level agreements (SLAs); more processing power turns into more database licenses not only in production but for every copy of the data. When data is copied to a data warehouse for reporting, to a test or development environment for product support activities, or to a disaster recovery site and there is a corresponding performance expectation for that environment, the production server upgrade now turns into several server upgrades-each with a corresponding increase in database license upgrades.
In many cases, data volumes are growing astronomically, requiring next-generation analytical platforms-more fondly referred to as 'Big Data' systems-to keep up with the quest for knowledge. While these database systems offer an incredible opportunity to change the way organizations find value in their information assets, they too have an incremental cost associated with the size and volume of data.
The fact that database volume and its corresponding costs are growing exponentially is not the big insight. Anyone working in IT gets this. What is revealing is that the vast majority of the data in these systems is dormant. In fact, industry analysts estimate that as much as 80 percent of the data in these systems is dormant. These are closed transactions and infrequently queried data that often are only retained for compliance purposes. If you knew you could keep all this data online, reduce its size 90+ percent, eliminate growth in your databases licenses, and still be able to restore, manage retention, or directly report on the data, why wouldn't you? Why keep this dormant data inside your most expensive applications, riding on your most expensive infrastructure, being maintained by your most expensive personnel? Stop the madness. By taking a good hard look at who is accessing what data over time, there is a good chance that after some inflection point, data is rarely accessed-if it is accessed at all. So why keep buying database licenses for data that doesn't justify the need?
While the concept is not new, information lifecycle management (ILM) has traditionally been associated with tiering infrastructure and archiving documents or email. Application ILM takes this a step further applying tiered services and archiving to databases.The idea is to inventory applications and data warehouses with the value of data and how end users access it. Mapping expectations of business process response time to the underlying infrastructure optimizes operations for performance and cost. It is a philosophical shift from blindly scaling up or out by adding more capacity or compute power as a reaction to missed SLAs to asking the business what it really needs and does that need change over time?
If the business doesn't know the answer to that question, maybe a good look at overall business process efficiency is in order. If there is an answer to that question, capitalize on it. Take stock of the assets used to support that business process and quantify the following: What percentage of data stored online in the production database could just be deleted? What percentage of data has retention requirements-either legal or operational-but doesn't have performance requirements? That's the opportunity to assess what database licenses are costing your organization and if you really NEED them.
Let's face it. Managing databases and procuring database licenses is expensive and there are no signs that data growth is diminishing. With the amount of high-performance compute power and storage capacity wasted due to dormant data, there is a big opportunity to control big data growth. By removing dormant data by getting rid of what you don't need and archiving what you need to keep for longer retention periods, latent DB capacity is released, ultimately improving application performance and operational efficiencies. Taking control over data growth not only can deliver savings on avoided cost associated with more database licenses, but also offers additional advantages, such as:
Improved database query performance
Ability to meet or exceed application SLAs by IT
Shorten application upgrade cycles
Reduced backup, recovery, refresh and batch windows
Controls data sprawl ultimately improving eDiscovery efforts
Focus budgets on managing data based on its value as determined by its age and access frequency
Whether you are talking about big data because data has accumulated over time or overnight, IT has always been in the business of supporting big data initiatives. By focusing on the benefits of Application ILM, organizations can be in a better position to make sure that not only is big data getting to the right place at the right time, but that big data is managed responsibly.