From database server to virtual machine to the cloud
The tasks of the old-fashioned DBA are disappearing more and more. In fact, the Cloud not only provides all kinds of new possibilities, it also makes everything much more dynamic. It all used to be relatively simple. There was a server and on it you installed database software.
That was basically it and it was pretty static. When virtual machines made their appearance, things became a bit more flexible and you could set up multiple VMs. Now with the cloud, you can basically do anything you want. You can get machines up and running in no time, which makes it tremendously scalable.
Scripting and automation
These benefits are only useful if the rest is also taken care of quickly. Installing database software every time again and performing the same post-installation tasks is not only time-consuming, but also very tedious work. ICT wouldn’t be ICT if you couldn’t automate this and with the help of scripting and automation languages like Terraform and Ansible, you don’t have to do those repetitive tasks over and over again.
Job title DBA will disappear
So the duties of the DBA have shifted over time. From someone who installs the database, creates users, backs it up and updates it when there is a new version to someone who also takes care of deployment automation, infrastructure automation and orchestration.
Thus, the job title DBA will slowly start to disappear. We have been seeing it for a while in the United States, where in Silicon Valley large platforms like Facebook and Twitter are mostly no longer looking for an old-fashioned DBA, but are posting job openings for a DBRE, the Database Reliability Engineer.
Reliability, availability and uptime
And that’s not surprising, of course. Because of the enormous scale at which these platforms operate, it is already practically impossible not to script and automate the infrastructure and database deployment. Because they deal with privacy-sensitive information, the reliability of the data must be guaranteed at all times.
And because users scream blue murder if a platform is down for a while (and this can quickly cost millions due to lost advertising revenue), server uptime and data availability is also paramount.
Preventing configuration drift
Of course, the DBRE always strives for the best deployment of infrastructure and associated configuration, because that way he can prevent or even limit “configuration drift. Configuration drift is anything you still have to do on a server after deployment.
Changing settings for a database or tweaking configuration options every time after you do a deployment? Those are repetitive tasks, and repetitive tasks are annoying, but they can also encourage errors, which is exactly what you want to avoid.
Immutable infrastructure
So the key is to script and automate everything in such a way as to create the most perfect possible image of the database, software and settings, the Golden Image. If this is done well, then we are talking about an immutable infrastructure. No changes need to be made to it after deployment. If changes do have to be made, they are made in the version-controlled configuration definition. Then a redeploy is done.
Simplicity, predictability and recoverability
This way of working has several advantages. First, it provides simplicity. Because no modifications are allowed, permutations of infrastructure status are significantly limited. The second advantage is predictability. Because the status is always known, investigation and discovery will occur much faster and any problems are easy to reproduce.
And finally, there is the advantage of recoverability. The status can be easily restored by redeploying the Golden Image. This reduces the MTTR (mean time to recover) quite a bit. The images are known, tested and ready for deployment at any time!
Every DBA will transform into DBRE
In America, the term DBRE was already introduced a few years ago, but also in the Netherlands in recent months you see more and more vacancies for the Database Reliability Engineer. I am also convinced that every DBA will transform into a DBRE, some to a greater extent than others.
Because as with the old-fashioned DBA, not every company will have enough work for a full-time DBRE either. It’s certainly not something you can “just do on the side. If you seriously want to transform as a DBA into a DBRE, it requires not only a new way of working, but also of thinking. You have to start thinking differently about your work and in everything you do, you have to think about how to avoid repeating the same thing manually again.
Is your business migrating to the cloud soon?
Is your company migrating to the cloud? Don’t forget the database! We have extensive experience with the big three cloud platforms Azure, Amazon and Google. But even when it comes to private cloud solutions, we know what the best practices are. And we know how to tweak configurations for best performance. Are there problems? Then we can solve them. At OptimaData, we have all the knowledge and we’re happy to support. Contact us if you want to talk further!
Inspiration:
The book by Liane Campbell and Charity Majors on Database Reliability Engineering and Designing and Operating Resilient Database Systems.
The review by Brent Ozar (the very best SQL Server DBA/Specialist worldwide) on the transition from the “old” DBA to the “new” DBA in the book Database Reliability Engineering.
Videos:
Laine Campbell’s keynote on Modernizing the DBA role
This blog on the DBRE and the transition from the old to the new DBA in keywords: The next iteration of the DBA