IT leaders have, so far, undersold themselves by focusing excessively on database management. Database industry expert, Martin Gaffney, shows why it’s time to shift their focus back to their true calling — being business leaders.
Gartner defines the role of CIO as overseeing people, processes, and technologies within a company’s IT organisation to ensure they deliver outcomes that support the goals of the business. If you do a business school course at Cambridge, you’ll learn how to deliver impactful results by harnessing new technologies and positioning your organisation to swiftly respond to change.
I believe that for many CIOs the reality is sadly often less inspiring and glamorous. Too many of us are acting as the recruitment agency lead for database administrators (DBAs). Given that the average DBA’s salary is around £50k, you can feel proud that you’re making a lot of these people very happy by recruiting them.
But why is that most of us can’t fulfil the role of CIO as defined by Gartner and Cambridge University?
Why IT leaders employ database administrators
Forrester says the current industry average for a large enterprise is one DBA per 40 databases. This ratio is determined by the total size of the databases being managed – “One DBA can effectively manage 25 databases of 200 GB each or five 1 terabyte databases,” it notes, and these may include production and non-production databases. Therefore, the more databases you need to manage, the more staff you will need to hire and pay to maintain them.
I’m going to go out on a limb here and say that we didn’t need as many DBAs on-call pre-cloud. While moving to the cloud allowed us to trim headcount in terms of internal ops people and developers, somehow the number of database wranglers increased. How come?
Why the database cloud fix has stopped making sense
There’s a long answer and a short one. I’ll give you the short one first, to get you back onside after calling you a glorified recruiter!
It’s a result of the way we decided to work with databases when we moved off prem. Essentially, what we now call ‘monolithic’ databases couldn’t just be lifted and shifted in one go into the cloud. 10 years ago, most brands had one great big Oracle database that supported all their services, but that didn’t work for a distributed, virtual environment.
Instead—and I do want to say, this made absolute technical sense at the time—we solved the problem by breaking that database into smaller, more tractable chunks. Some bright spark—was she a DBA? —said, “instead of one giant database, we’ll have 50 little databases that will talk to each other. Each database will serve a different part of the business, a different microservice or a different family of microservices. So, if we need to change the database behind a service—no problem, we just need to change that little data engine, do a quick update, and the rest of the business is unaffected.”
It’s time to shake things up! The era of tons of people tending endless expensive software instances is no longer sustainable.
Martin Gaffney, Vice President, EMEA, Yugabyte
Where the cloud fix for databases went wrong
Ten years on… things have gotten messy. What once seemed like a brilliant escape plan has proven to have consequences. There’s still plenty of Oracle and DB2 databases out there; in fact, Oracle’s business is still booming; I know of a major US bank which has approximately 10,000 Oracle databases spread throughout the organisation.
We need multiple databases as we see them as the must-have engine for customer services, customer accounts, loyalty schemes, the shopping basket, stock control, and so on. We’ve reached the point of one database per microservices—a huge volume of little chunks of functionality apps that all want their own database. Now, we fear they’ll get tangled up with other apps, so you can’t change this app without changing that one, and so on. It’s become a minefield of complexity.
Hence: DBAs, as far as the eye can see. And database vendors that love finding more and more reasons for you to buy more licences (and auditing you to show you how many more you use, then adjusting the bill in their favour). This is a graph line that will never stop going up and to the right. It’s like the spread of the infected in ‘The Last of Us.’
Before you say it: yes, you tried NoSQL to curb the zombie database plague. But anyone who’s gone down this path knows that if you have a serious need for transaction and ACID compliance-level data robustness (hint: any business of size), it just doesn’t work without a lot of back room patching. So, you get to employ more developers too! What a generous budget holder you are.
It’s time to shake things up! The era of tons of people tending endless expensive software instances is no longer sustainable. There must be a better way?
Is distributed SQL your way out of the job agency business?
Up until very recently, that would just seem like a dream. Technically, ‘chunking’ seemed the best approach, supplemented by NoSQL for non-core enterprise business stuff.
You had to just shrug and accept the technical debt—having your people beavering around in the background writing/fixing code, patching things up, finding clever kludges to connect things that don’t really like talking to each other… all ‘busy’ work that doesn’t add value to the business (remember that CIO pipedream of ‘delivering outcomes that support the goals of the business’?)
But there is good news at last. You will reduce your costs and hire fewer DBAs because you can now use an open source, standards-based approach to package up all the best features of monolithic in the form of distributed SQL (think Google Spanner, or my own company’s YugabyteDB).
Distributed SQL is your best bet to truly modernise your database and make it work for the cloud. It means a lot less micro-database, and a lot more microservices. It will also stop you being the best recruitment agency boss for DBAs—and allow you to finally take on the job you signed-up for, which (according to this inspiring definition from IDC) is to ‘co-create products and envision new business models with senior business leaders.’
Don’t get me wrong—I love to hire people. I just prefer to hire ones that directly help me innovate and grow the business. I’m sure you feel the same.
ABOUT OUR GUEST WRITER
Vice President, EMEA, Yugabite
Martin Gaffney is the Vice President, EMEA, at Yugabyte, the company behind open source, high-performance distributed SQL database, YugabyteDB. He has a long and successful track record of being the first person-on-the-ground of the very early EMEA team in stand-out companies such as Netezza, ThoughtSpot, Sequent, Tivoli and now, Yugabyte.