Theory of Constraints: the DBA

November 15th, 2013

 

9627525993_af43154455_z

photo Celestine Chua

“Do you think the great Skeeve has nothing to do with his time but guard your borders? Do you want to tie up your high-cost magician doing the job a low-cost soldier could do?” – Aahz, MYTH Conceptions by Robert Asprin

The book The Phoenix Project is the story of sorting out a company’s development and IT problems. In the book there is a character Brent. Brent is the go to IT guy who can fix things no one else can, like broken databases. Brent is also the bottleneck for required deliverables to move the company’s Phoenix Project forward. The Phoenix project is a make or break project. Either the Phoenix project succeeds and the company succeeds or it fails and the company goes down.

For the project to move forward there are very crucial deliverables, such as environment builds for QA and development. Without these deliverables the project gets blocked.

Brent is the gating factor on these deliverables and therefore the entire project.

Unfortunately Brent is used for all sorts of tasks. He’s like the go to guy for everyone.  Everyone’s requests turn into unscheduled interruptions and all these unscheduled interruptions cause delays in Brent delivering the crucial project deliverables for the Phoenix project.

Part of the book is how this crucial resource and bottleneck, Brent, is optimized. A clear first step of course is to take anything off his plate that someone else can do. Unfortunately much of what Brent can do, only Brent can do. One solution is to get Brent to train other people to do what he can do.  Couple of problems with that. One, Brent often doesn’t  know how he accomplishes some of the things he does (or so he says) and two, he doesn’t have time to train others because of the enormous amount of backlog and pressures to get the Phoenix project moving forward.

Now my question is this: if there is a huge, complex, time consuming task that only Brent can do, would you ever make Brent do it when there is an automated, self service solution that does a better job than Brent and can do it faster and be run by anyone?!

Of course not ! Right ?!

Well, wrong. It’s happening all the time. Companies are relying on expert DBAs to copy and clone massive databases over and over again.

DBAs are generally the most expensive resource in the IT department and often the hardest to justify because the business just doesn’t get what the DBAs do, but DBAs are crucial and foundational to companies. The business on the other hand only knows that it wants this data now and if the DBAs can’t give it to them then the IT department must be broken.

Speaking to companies like RBS, they said that DBA teams spend up to 50% of their time making database copies. Macys said their DBAs spend 4,000 hours a year making database copies.

But guess what. All that can be eliminated with database virtualization. Macys went from manual copying to database virtualization and reduced that 4,000 hours of expert DBA time down to 8 hours of Junior DBA time.

 

8456186741_437a7234a7

photo Steve Jurvetson

 

 


Uncategorized

  1. Trackbacks

  2. No trackbacks yet.
  1. Comments

  2. Noons
    November 18th, 2013 at 05:36 | #1

    Hmmmm…. Show me a company where dbas spend 50% of their time copying databases and I’ll show you a company where IT management hasn’t got a clue…

  3. khailey
    January 15th, 2014 at 21:05 | #2

    At one point in time around 2000, I was the DBA at a startup and there were periods of time where I spent 50% of my time copying databases.
    Given time I would have automated this more and more but still copying around databases is still slow, tedious and scripts are fraught with potential errors often due to resource constraints. Copying databases is definitely something in demand when there is a lot of dev and qa work to be done on an app stack that works principally on data in the database, especially at the start of a project.
    We’ve talked to a number of customers who have use to spend considerable amount of time copying databases before Delphix. One customer who used subsetting because they didn’t have the resources for full copies and their DBA team spent 50% of their time making these subsets. Another customer told us they went from 4000 hours of copying databases a year down to 8 hours.


+ three = 12