Thursday, September 28, 2006

Commoditization of Database Administration - Will DBAs and IT Managers "Get it"?

Over the last 5 years, database administration is slowly but steadily turning into a commodity exposing numerous niches that need creative input and hard work by the DBAs whose bandwidth has suddenly increased. There are three primary market forces working together to make this happen:
(i) Advances in DBMS technology from IBM, Microsoft, Oracle and open-source databases (in that order) ;
(ii) DBA outsourcing and offshoring;
(iii) Automation and virtualization technologies that create a layer of abstraction over DBA tasks ranging from simple ones like database server provisioning to relatively complex ones such as error-free nightly load process management.

The first stems from the effect of databases themselves turning into commodities. For instance, DBMS software is being given away for free not only from open source DB support hawkers such as MySQL but also the big 3 (Oracle, IBM and M$). IBM announced a free version of DB2 back in Jan '06 ( . Likewise, Oracle has been offering its free Express Edition for a little while now (

As databases turn into commodities, what is the impact on database administration? Rather than being thought of as something intrinsic to the company, it is being rightfully seen as a regular IT function - a necessary evil that is not necessarily a core competency of the business. Database administration has increasingly been separated from data management with the need for specialists in each area being felt by the business. This separation has resulted in the DBA not having to deal directly with the data within the database. DBAs used to be viewed as someone with the keys to the kingdom - in terms of their promixity and access to business-critical data. But with newer audit capabilities and security models offered by Oracle (sysdba versus sysoper, etc.) and other vendors, it is easy to separate the Systems DBA role from data-centric responsibilities. This commoditization and separation has further made it conceivable for companies to entertain going after boutique/best of breed outsourcers and hand over support for the traditional Systems DBA role in a cost effective and secure manner driven by metrics and service level agreements.

Many of these DBA outsourcing specialists have managed to add value via strong process models such as ITIL and monitoring/automation tools and integrated themselves within the companies they provide services to - further adding credibility to the fact that database administration is something that can be successfully separated and outsourced by astute IT managers.

The state of automation and virtualization frameworks and supporting sub-technologies have matured over the years - making automation of many mundane yet labor-intensive areas a reality. DBMS and tools vendors have largely shed their inhibitions about supporting solely their proprietary platform and have taken on support for competiting platforms - starting with basic monitoring support. For instance, Oracle has recently added support for SQL Server and DB2 databases within their OEM/Grid Control monitoring tool. This was plain inconceivable just a few years ago! But as the industry evolves, mainstream DBMS vendors have no choice but to lean more and more towards offering better editions of their databases for free and include support for competing databases within their own tools.

While these events bode well for the average customer due to lowered product costs and higher administrative efficiency and quality, how does this impact the individuals themselves - the DBAs who have been depending on the various DBMSs and assorted tools for their livelihood? Do they have reason to start looking at alternative careers as mail-carriers or petting zoo attendants?

My opinion is, the commoditization of database administration is a golden opportunity for DBAs to finally separate the mundane administrative areas, outsource and/or automate them and elevate themselves up the food chain to focus on areas that they previously never had the time to work on. The more user-visible stuff that if worked on, will yield measurable results in terms of reduced # of SLA violations, performance dips and outages. Areas such as quantitative performance management, service level management, data management, application/database interplay management, database/infrastructure optimization, new DBMS feature evaluation and other proactive management areas. In other words, the optimal career movement for the successful DBA needs to be "management" related (not necessarily people management, but managing performance and related areas indicated above.)

It will be interesting to see in the coming years how many DBAs actually "get" this, embrace the change and position themselves at the forefront of this once-in-two-decades kind of revolution versus how many continue to sit in their temporary trenches with arcane tools and scripts, fight progress and eventually miss the boat.

No comments: