Part 7 in a series of posts about Data Protection as a Service. This is the first of two posts on automation… (Also posted on Cobalt Iron’s blog)
Mike Matchett, Small World Big Data
From breathing to paying bills, from good ideas to great habits — what’s better than automatic? Intelligent automation means never having to miss an opportunity, obligation, or requirement. When thinking about the best way to approach enterprise data backup, smart automation ranks high on our list of goals.
Automation is especially critical when dealing with the challenges of deploying limited expertise over a widening scale of mission-impacting data and the growing complexity of hybrid infrastructure. It really is a case today of sink or swim!
Inside any IT operation — but especially those concerned with availability, performance, or security — an ultimate goal really should be 100% automation. Some folks call this “autonomous operations.” Fully autonomous data protection always creeps just out of practical reach given the increasing volume of important data and emergence of new architectures and applications. Even if just to keep pace, the goal should always be to increase the level of operational automation.
Where to Start?
It is impossible to automate everything at once, so which parts should be addressed first? Automate any manual and continually repeated data protection task or responsibility, especially where human consistency (or lack thereof) affects reliability.
If someone has to stop and think about what they are doing each time they approach a repetitive task — remember key command details and gotchas, recall small steps and perform them in order, and never make a typo — there will be mistakes. When these kinds of tasks concern data protection, one or more mistakes eventually will prove quite costly to the business.
In cases in which full automation isn’t yet feasible, there is usually a smart way to automate away risk and provide intelligent, accelerating assistance. As in many IT disciplines, one can capture and encapsulate best practices, aggregate big data sets if need be, then leverage intelligent analytics to apply both ongoing learning and deep expert knowledge consistently.
Of course, it doesn’t make sense to spend a lot of effort automating mediocre (or outdated) processes; that will just deliver worse results faster. Implementation of automation best practices requires experience and expertise. One must automate not only the right things but also maintain and evolve all automation over time. It’s not a static world, and automation that is hard-coded, embedded, and forgotten can become a thorny legacy problem later on when environments change and key assumptions no longer apply.
Existing practices might very well be those relied on to just get by, and they are not necessarily considered the best. There are probably big gaps and more exceptions than we want to admit. Again, keep in mind there is little point in automating poor practices. It is important to consider that the very best source of expert automation and ongoing maintenance may not be found in-house, particularly in smaller IT shops.
…(Continued in a second post that explores automation oversight.)