Whos Afraid of Self-Managing Databases?
Whos Afraid of Self-Managing Databases?
The role of the DBA (database administrator) is an important and mostly misunderstood role in many IT organizations.
Those of us who have held the position appear to some as geeks who speak another dialect. The language of buffer cache hit ratios and access paths.
The database represents that world between the logical and physical, between hardware and software and just like our parents let the VCR time clock flash until we stopped by to reset it, organizations believe the database to be a black box best left to the experts. I agree, but not blindly.
The major database vendors (Oracle, IBM, and Microsoft) have invested greatly over the past decade to improve the manageability of the relational database.
The goal is to increase the ratio of DBAs to databases supported, thereby lowering the total cost of ownership for a database.
The problem is that some DBAs and consultants that provide DBA services want you to believe nothing has changed.
They say that even though all of this built-in self-tuning functionality is in the database software, it is optional and not intended to replace standard DBA activities.
They will make statements like "automation features are aimed at shops that do not have the manpower or expertise to manually perform the tasks."
As an IT manager, you must see the absurdity in such statements. Why would anyone want to perform a task manually when it can be done through automation?
Personnel costs in 2006 are estimated to be almost 47 percent (56 percent if you count professional services) of the average data center budget, making it easily the most expensive single category.
We should expect to see those percentages trend downward, perhaps by as much as 10 percent, over the next eight to 10 years specifically as a result of automation.
The database vendors have made a good start, but it is up to IT management to demand that DBAs begin to exploit these features.
Over the past two years, the idea of outsourcing the DBA position has become a popular topic of interest, first with business owners looking at budgets and finally with DBA managers under pressure to reduce costs or deal with reduced staff and turnover.
Clearly we are in a period where dramatic change is occurring with respect to how organizations value IT operations and its impact on the bottom line.
Of course, regardless of the huge savings that companies are promised by exploiting cheaper, non-U.S.-based labor, many studies continue to indicate that the promised savings never do materialize.
Could it be that organizations are treating a symptom and not the cause? Could the DBA who refuses to exploit automation features be in essence hastening their own demise?
We shouldnt place all the reticence to exploit automation features solely on the shoulders of the DBA.
The database vendors also play a role in this as they are caught in the throws of this market transformation.
Lets face it: Relational databases have not been the easiest thing to manage, which is why DBA skills were highly sought after for the past two decades.
Because of its perceived complexity, database vendors often leveraged the DBA community as the inroad to getting installed.
An organizations choice of database software often came down to just a few factors, not the least of which was the recommendation of the DBA or the consultant hired to take on that misunderstood, yet crucial role.
Next Page: Success from loyalty.
Success from Loyalty
This means that Oracle, IBM and Sybase owe much of their success to the DBA and have counted on those strong loyalties to maintain their grasp within an account.
Its no wonder they are unlikely to come out and announce that their database doesnt need a DBA anymore.
On the other hand, Microsoft has taken the road less traveled (at least by the other major vendors) of selling its SQL Server database to smaller enterprises or departments within larger enterprises.
These are the groups that did not have the manpower or expertise to manage the database manually. In other words, Microsoft had to build its database for a different audience.
Now that the IT organizations are truly devaluing IT operation roles (DBAs included), its no wonder that SQL Server has become the fastest growing relational database.
The perception is that it is easier to manage. Indeed, many DBA groups within organizations have only just begun to pay attention to the vast amount of SQL Server usage within their organization.
This perception has also expanded out to the third party tools market where the traditional vendors (CA, BMC) have failed to find much demand for tools that were readily scooped up by DBA managers to support Oracle and DB2.
Now all the database vendors are in alignment concerning automation because the growth market for them is the same one Microsoft has been cultivating, the small to midsize business (250-1000 employees) that in the U.S. alone represents an estimated $54 billion in IT spending this year, including $18 billion in software.
Lower total cost of ownership is crucial to this market, but its also crucial to the large enterprise customer.
Its time for the DBA to embrace self-tuning features or risk becoming extinct as a new generation of DBA is born.
The next generation will again be highly sought after, not because they can write a script counting indexes that are never used but because they provide high-value expertise in areas such as data movement and reuse, data security and encryption, and data availability within a service-oriented architecture.
They will understand the business value of data archiving, and they will understand how real-time analytics can successfully accomplished and maintained.
In short, they wont be worried about day-to-day minutia that many of todays DBAs feel comfortable with.
Database software is simply too complex for most of us to grasp all of the possible implications of the most minor change.
In other words, the DBA is guessing most of the time. Yes, some of that guessing is highly educated; some, however, is based on knowledge that may have been relevant in a prior release but is no longer.
Some are even based on pure urban legend without taking into context the capabilities of the current hardware, optimizer changes and workload patterns.
Now the database vendors provide you with the tools to let the database manage itself.
We should let it try, after all, most companies would pay big dollars for an expert from Oracles development team to come and figure out how to squeeze even 10 percent performance improvement out of an important SQL statement.
Why not then allow the knowledge that the developers built into the database (and used by consultants anyway) do the job?
Now, we are not suggesting zero supervision, but certainly we should strive to raise the ratios of DBAs to instances managed so that we can lower operational (and low value) tasks.
Organizations that want to leverage these self-tuning features and automate lower-level DBA tasks must start by evaluating their own database management process maturity.
After all, automation has existed via third-party tools for more than a decade, but little of that functionality was ever fully used.
The reason is that the organizations own processes were poorly documented and not repeatable. This is crucial in taking advantage of any automation.
It is my belief that some organizations will look to outsource jobs rather than take the time to consider how they organize their support staff to enable improved process maturity.
Certainly an organization can always hope that the outsourcer has mature and repeatable processes.
Of course the outsourcer will be the only one that benefits from that (at least from a cost savings basis).
So if you are a DBA or DBA manager and you want to make a positive impact on your organizations IT operations and bottom line, begin the task of evaluating how to better organize your DBA staff.
Identify which processes in use today are best practices document them and standardize on them.
Most importantly, use that as the basis to take full advantage of the databases self tuning features. Believe me, there is plenty of other work that must be accomplished, and the DBA is the role most capable of tackling these new challenges.
Charles Garry is an independent industry analyst based in Simsbury, Conn. He is a former vice president with META Groups Technology Research Services.
Check out eWEEK.coms for the latest database news, reviews and analysis.