DevOps – it takes a village to do data management

Connect with us
Hillary_Clinton_in_Silicon_Valley DevOps – a portmanteau of Development and Operations – seems to be flavour of this month. Briefly defined, it is a methodology that forces these two often fractious disciplines closer together.

Many people who have worked in large organisations will recognise the friction that often occurs between other departments and IT. Very large IT departments tend to insulate themselves behind their helpdesk. In order to gain access to key IT people it is necessary to observe helpdesk protocols. Anybody who has ever tried to get something done urgently by IT and has run foul of lacking a helpdesk reference number will attest the frustration caused in the name of change control. Sometimes, the person who’s help you need may only be sitting a couple of desks away but first you have to leap through hoops of bureaucratic fire. The IT function needs to track the requests it receives, but change control forms seem to have got longer and longer and require more sign-offs. Complexity tends to creep in to helpdesk processes and eventually this suffocates the business. Agility goes out of the window.

I recently met with the IT team of the asset management arm of a very large international bank. Demonstrating the supreme agility of our Data Hub’s rules engine, there were gasps around the room. Unfortunately, not gasps of delight! The IT team were horrified at the idea of ceding control to users to change business rules about how to validate their data. “We can’t do it like that. All changes have to go through us because we need to do UAT on it”, came the objection. Now, to be fair, surely, it cannot make sense to let users completely loose on data infrastructure, letting them make changes in an ad-hoc way. But I did wonder if we have forgotten what the “U” in User Acceptance Test stands for?

Now, Misato Data Hub is rules based so IT can have all the permissions if that’s really what you want.  Do IT really have to make the change that tweaks a business rule that defines if a data value is acceptable or unacceptable? Since Operations own the data and they are the experts in it, why not let them set their own validations? The system must have role-based user accounts that let you get quick feedback on whether the change was as expected, revert it if necessary and audit who changed it and when. Also, it must prevent radical changes to shared services that could impact other consumers of the same data. However, this kind of co-ownership is in-keeping with the DevOps revolution. If we work together more closely, the demarcation will blur and the business will benefit from faster time to value. Of course, IT can still keep ultimate control of upgrades and access to the platform. Lets do what we are best at and not hide behind departmental barriers. To egregiously paraphrase Hillary Clinton, “it takes a village to do data management”.

This kind of thinking is the heart of the DevOps revolution. IT, Operations and QA/Process Control teams need to come out from behind their bureaucracy and work together so the business can be more agile.

Comments are closed.