Blog Last week I had an interesting chat with Andy Warfield (CTO and founder of Coho Data). We started a debate about his latest article and the pros and cons of custom hardware design in modern storage products. However, the conversation quickly got side-tracked to another topic: the role of the storage admin, if any, in 2016.
It all started from the role of hardware in the design of new storage solutions, but ended up in a discussion about what end users really want from storage systems today … and what vendors are doing to make it happen.
What do you want from your storage?
The answer is very simple. The wish list is not the same for everybody, but I’m sure that in the top five positions “ease of use” or, more generally, “ease of operation” is always included. “Ease of use” or of operation, which incidentally are not the same, are on the top for several reasons. Sometimes it’s about scalability (scale-out is easier to operate than scale-up).It could be about protocol choice (multi-protocol gives you more choice and freedom), user interface (not just a good UI but also strong APIs), integration and so on. Even all-flash could be considered a way to simplify your storage because, to a certain extent, it avoids complications with performance tuning. Everything is now designed to bring simplicity and avoid any sort of complexity … especially at scale. The goal, for many vendors, is to make the storage layer transparent, as is happening for the rest of the infrastructure.
And lately we are starting to see evidence of this trend, withmany newer storage products now operated by non-storage administrators.
Many newer storage products are operated by non-storage administrators!
The storage admin is dead
Sorry to say, but this is true. I know it’s hard to deal with your own death but you, storage admins, are a dying breed. You are the last person standing, the guy defending the last hill. Thinking about storage as storage estate, the arrays and storage area networks to manage, it’s no longer a feasible job. The quantity and quality of data is so large now that the same tools can no longer be implemented nor can the infrastructure be looked at in the same way as in the past. It has already happened to your colleagues a long time ago. The server admin is dead and their old job now comprises two different roles: the rack guy and the virtual infrastructure admin (or simply, SysAdmin). The server admin who hasn’t evolved and insisted on dealing with servers is doing it in the data centre now (with all due respect to those poor guys working in that inhospitable environment), racking servers, wiring them, switching them on and doing maintenance.
The evolved sysadmin is actually doing another job … in fact, he or she is now in the automation and orchestration business, with hundreds of servers under management (if not more), a bit of coding and a lot of fancy foot work … They probably don’t even know where these servers are physically located in the data centre.
The sysadmin just has a virtual view of them, but that doesn’t mean that they don’t have complete control over them. The truth is that now all these virtual servers are much more under control than in the past and the whole infrastructure is more flexible and manageable. The split between the server guy and the modern sysadmin started 10 years ago … when VMware and hypervisors in general were finally accepted as a valid option for enterprise workloads…
Now, thanks to flash memory, virtualisation, software-defined storage and above all the cloud, it no longer makes sense to micro-manage storage systems … otherwise the risk is that storage admin folk become another just rack guy.
Thanks to flash memory, virtualisation, software-defined storage and above all the cloud, it no longer makes sense to micro-manage storage systems…
Long live the data admin
As with the server/sysadmin, it’s time to evolve. Are you ready? In a similar evolutionary process, as with what happened to the server admin, the storage admin needs to become a data admin. Concepts regarding automation, orchestration and even coding are applicable to storage: from migration, configuration and provisioning, the data admin will have to think more about applications, data policies, security, data protection and re-usability/recycling of resources.
At the end of the day the goal is the same, it’s all about data management (protection, performance and SLAs …), it’s just that the data growth rate (and diversity) is quite unmanageable with traditional tools, and you can’t do that by working on the single storage array. You have to apply your knowledge and do it at scale.
Some tools are ready, others are still in the early stages of their development but the whole data centre is heading in that direction and storage should too … The difference is to start moving away from the concept of the storage systems and start thinking about storage infrastructures and data virtualisation – APIs, integration with hypervisors, strong automation and much more: now that storage is much smarter than in the past, it’s time to change the way we use it.
Closing the circle
With more and more private and hybrid cloud deployments, storage has to follow the trend and become smarter. I know the conservatism of storage administrators but they need to think more about data and less about bits and pieces. Containers and other technologies up in the stack are taking data availability, resiliency and durability for granted. Meanwhile users, no matter where they are accessing data from, expect uncompromising performance and capacity. Last but not least, storage is more competent than in the past and software-defined systems allow you to build much smarter and flexible storage infrastructures now.
In some cases, they are also capable of off-loading some compute tasks directly to the storage infrastructure. Examples are all around us but, again, this means that you need to think more about data and less about the box containing it.