In-Depth

Flash Storage: a Virtualization Game-Changer

All new technology is promoted as transformational. This one truly is.

I am both a systems administrator and a writer. In moments of hubris I like to imagine that this combination enables me to pass along qualitative data from my personal experiences, instead of cold hard numerical precision demanded by traditionally quantitative technology journalism. Today's topic is modern storage; more specifically, how I feel it has changed my views on virtualization.

The Psychology of Systems Administration
Were this a typical qualitative article, you could expect a ream of numbers: speeds and feeds, number of VMs fit into a smaller space and so on. I'm sure there are innumerable flash vendors that would love that. What I want to talk about -- what I think actually matters -- is how introducing flash has affected the psychology of my systems administration.

As a sysadmin with reasonably severe ADHD, I don't tend to take on clients where there are a whole lot of formalized processes. I don't do ITIL; there aren't formal checklists to go through when I create a new VM. For the most part, things are done from memory. At best, I go by some notes I've taken down and shoved into a text file in the documentation directory. In small shops, you can get away with this sort of thing.

The one thing that has always saved me is that the systems I've worked on have been relatively slow. Installing, patching and configuring an operating system, applications and relevant services took long enough that I had time to go through my mental checklists, peruse my informal documentation and even verbally double-check with one of my co-workers to make sure I didn't miss anything.

Flash changes all that.

Close Encounters
My first real encounter with datacenter-class flash in production was my Tintri T-850. I loaned it out to a customer whose SAN had blown up, and it proved to be difficult to get used to. Everything was so much faster that, if I'm being perfectly honest, I ended up making mistakes during VM builds and migrations. I was so used to trucking along at the speed of the progress bar that when the progress bars gained a warp drive, I didn't really have any good habits to fall back on.

The Tintri came into my life in mid-2015. Enterprise flash wasn't exactly new at this point, but capacity flash was egregiously expensive. The Tintri's impact on me wasn't just about flash. Tintri put a lot of effort into making storage "fire and forget" simple. It's kind of eerie how often I forget it's even part of my lab. This would be repeated in other storage systems for the next 18 months or so.

Earlier this year, Micron sent me over a pile of PCI-E NVMe drives. I sprinkled them about my lab and suddenly I had to cope with changing how I did things not just for one customer, but in my day-to-day operations as well.

During 2016 flash prices plummeted. Arrays loaded with Dell 3.84TB drives are the new normal. I've even seen my first order for a NetApp all flash unit with Samsung 15TB hard drives! The mind boggles.

Amidst this cacophony I began to slowly realize that the change I was adapting to was not simply about the speed at which I thought or reacted. The solution to my problems wasn't the adoption of more formalized processes, to-do lists or so forth.

Underneath it all, the combination of the blistering speeds offered by flash and the quantum leap in manageability offered by the startups has changed how I react to infrastructure as a whole. I used to think of myself as an infrastructure administrator. I grew up liking blinkenlights, idolizing Geordi La Forge (of "Star Trek: The Next Generation" fame), and wanting to be someone who made the machines go.

No Longer That Guy
Somewhere, I stopped being that guy. Scale Computing came out with a hybrid hyper-converged solution that most of my customers ended up buying. I stopped using VMware every day. I stopped logging into my storage solutions and assigning LUNs. I stopped hand-building VMs and relied on desired state configuration tools. In short, I started treating the IT infrastructure I maintained less as my baby, or my personal empire, and more as this ephemeral thing that only sort of mattered because it helped me do my real job: deploying applications and services.

Proper enterprise storage changed everything about being a virtualization administrator for me. And they did it by removing almost all the times I need to actually administer my virtualization infrastructure.

Now, about networking…

About the Author

Trevor Pott is a full-time nerd from Edmonton, Alberta, Canada. He splits his time between systems administration, technology writing, and consulting. As a consultant he helps Silicon Valley startups better understand systems administrators and how to sell to them.

Featured

Subscribe on YouTube