GridApp Systems - db Connections

JULY 2010

ENABLING THE SELF-SERVICE DATABASE
BEYOND THE FUD OF THE 'SCRIPTERS'
BY ROB GARDOS, CEO

When you are successfully selling database automation solutions to the largest companies in the world you see some baffling themes. What’s been most baffling is how long and how much money enterprises spend on attempting to make a generalized automation product actually handle any complex component of the application stack, like the database. The story is simple:

Vendor promises everything (they have to, it’s competitive)
Millions of dollars and countless months/years are wasted trying to make it work
Beyond the most basic uses cases, that are most likely already outdated by the time they are scripted, it doesn’t work
People yell - partial credits issued (if you’re lucky)
In the end the enterprise is no better off than they were before the project started
And of course the aftermath; assuming any of the original sponsors are still around, they suffer the consequences (or find the ‘sacrificial lamb’)

How can you not love the enterprise software business?

As organizations look at self-service, we’ve noticed that customers are becoming more discriminating in their decisions. Perhaps it’s the impact of virtualization. This technology actually works for image deployments and the leader in the space, VMWare, tends to be a bit more honest on what its product truly does and doesn’t do. It could also be that the hype around private clouds is large enough that technology folks are being more careful. Don’t forget, these same people have gotten burned already with the ‘all-encompassing’ Opsware/BladeLogic/Tivoli decision and know firsthand what the pitfalls are. While the FUD may be losing some impact, it’s still out there and the biggest challenge is justifying to the powers that be what these products do and don’t do. When you’re talking to your boss who says, “Well we’re spending millions with [insert big IT Name here] and they say they do this so why shouldn’t I believe them,” what can you say? If logic can prevail beyond the FUD here are some reasonable responses.

These products actually do nothing out-of-the-box when it comes to managing databases. They are frameworks that require someone to create and maintain all the content. I can sell you scissors but does that mean you can cut your own hair?
Demos and POCs are deceiving. Anything can be made to work in a simplistic setting that is not indicative of reality
Who is going to pay for the ongoing support of new platforms? If you’re going to deceive yourself into thinking I can assign 3 DBAs to this and it will be handled you are doomed. This is a massive task with complex dependencies
The initial set of scripts provided by the vendor will quickly become out of date. Even if you get this so called content, what then? Whether it is in the form of a script or workflow the only thing worse than building your own code is maintaining someone else’s (especially when it keeps changing)

What makes GridApp Clarity so successful is that the approach minimizes the need to script. Our model-based approach requires users to define an end state and the software does the rest, independent of architecture, operating system, configuration and version. When Oracle launches a new clustering solution (ala 11gR2 RAC), the software handles it. If Microsoft introduces ‘high availability patching’ for SQL 2008 clusters it is taken care of. Imagine having to rework code/workflows to actually support this incessant set of changing requirements. It is simply not practical.

So don’t buy into the FUD of the products that do everything and nothing. If you’re not sure, try giving yourself a haircut and see what happens.

ENABLING THE SELF-SERVICE DATABASE

GETTING DIRTY WITH PATCHING

A DBA IS WORTH MORE THAN AUTOMATION