This is part of my series of posts on Tech Field Day 18. You can read more of them here.
My favourite part of Datera’s presentation at Tech Field Day 18 was the section on infrastructure in use.
Too often, technology vendors focus on greenfields deployments, or the initial installation of new infrastructure. IT systems are purchased with a depreciation period of three to five years, so even a full week of install time represents about half a percent of even a relatively short-lived set of gear. For the far more common scenario of systems that aren’t replaced until they reach technical end-of-life, focusing on the initial setup tasks ignores where most of the value of the equipment comes from: being used.
A better way to think of installation is as one special case of a change. Minimising the cost of changes throughout the life of equipment is a vastly more useful way of framing the challenge, and that’s what Datera presented here.
The Cost of Change
When the cost of a change is high, the change tends to be given much more scrutiny, and the value likely to be gained must be that much higher and more assured. A high risk change with a high cost of failure but little payoff simply isn’t worth it. However, if the cost of change can be reduced, then less value need be delivered with each change. This is the driving force behind the “ship to production thousands of times a day” rhetoric we hear from cloud-native DevOps devotees.
Low cost changes also help to overcome the human bias of loss aversion.
Datera’s approach to storage provides these advantages: it’s easy to make changes, so the changes get made without hassle or fuss. If the changes turn out to be not as great as hoped, it’s easy to change them back and try again.
This encourages experimentation, which accelerates the discovery of improvements. Each incremental improvement gets rapidly accepted as the new status quo on which yet further improvements can be built. This is the simple mechanism behind the enormous benefits seen by those organisations that shake themselves out of the rigid and controlling processes that have consumed them for far too long.
What was once adopted as a retreat to safety has locked these organisations into a kind of stasis.
Ironically, this mechanism makes the storage layer much less important on a daily basis. It disappears into the background because it’s no longer getting in the way, petulantly insisting that its special fragility needs to be coddled before any progress can be made. The brittle IT systems of yesteryear required armies of priests to mediate their interaction with the lay public in order to protect each from the other. The more flexible and robust approach of systems like Datera opens up great possibility of what can be done with them.
Because as important as storage may be to the steely-eye storage admins who cut their teeth on hand-configuring RAID layouts, no one else really cares. The job of storage is to make data available when and where it is needed (and nowhere else, because security). Today’s applications change their requirements rapidly, in response to the changing demands of the wider world. Storage that can’t adapt quickly enough slows everything else down.
We need systems that are as adaptable as Datera appears to be. Changes should happen online, all the time. It should simply be routine, and that means concentrating on the value of making change easy and low-risk.
More of this, please!