Good god, where will the new storage experts come from?

The value of experience

While the younger, hipper crew busy themselves with the next Social-enabled Bewjelled Bird as a Service public cloud-backed mobile-wearable-of-things super application, previous generations are deploying and managing SANs, wrangling virtual infrastructure, patching some ancient cobol application for the umpteenth time, and manually entering the 1s and 0s directly onto hard drive platters with a force electron microscope and the sheer power of rugged determination.

Every time I have to teach some greenhorn how to format a floppy disk so that they can do a BIOS flash a part of me dies inside. Soon, we'll have these doe-eyed children of the public cloud looking at us in terror and asking "sir, please, what's a LUN?"

The skills base for running internal IT is going to collapse. "Systems administrators" will, in a very real sense, soon be obsolete. Oh, there will always be a few on hand; on-site gear isn't go away in the enterprise any more than mainframes evaporated overnight, and someone has to run those clouds…but there are a lot fewer "someones" required for that than handle the job today.

Today's sysadmins will "reskill" for this cloudy future. Many will simply retire. Others are headed for the exits; they're seeking their fortunes in a different occupation all together. Slowly but surely the skills base for internal IT – and perhaps most critically, in-house storage – will collapse.

Once that happens, we are collectively screwed. Only the richest corporations will be able to afford the nerds required to maintain their own IT and thus only the richest corporations* will own IT. They'll rent it out as they see fit, insulating themselves from economic risk and downturn by turning the knobs on the rest of us.

Surely there's a path forward? Ownership and rental both have their advantages and their risks. Wouldn't it be great if we could find a balance between them?

Hybrid vigour

The solution to our ills is to keep a hand in maintaining our own infrastructure. The hybrid cloud is necessary not only so that we have a fallback position when things go pear-shaped, but so that we maintain a minimum skills base within our organisations. We need to run our own "private" clouds, but design and implement them in such a fashion as to be able to put workloads on the public cloud as well.

The ability to "burst" workloads up to the public cloud allows us flexibility while having a minimum of equipment locally allows us to mitigate risk. For those interested in this there are three basic options.

The first option is Microsoft from top to bottom. While it’s got a reasonably mature hybrid offering that you can buy off the shelf today, I have to admit to being less than enthused about the idea of getting into bed with it for another decade or two. If and when Microsoft's goals are aligned with your own, it’s a fine business partner. Of late, however, its goals seem to run counter to the interests of customers, ecosystem partners and developers. I'm not sure who that leaves aligned with Microsoft, but I'm pretty sure it's nobody reading this article.

Next up is VMware; like Microsoft, it offers an off-the-shelf hybrid option. The hybrid portion of the equation is a little less polished than what Microsoft has to offer, but that gap is closing rapidly. Like Microsoft, VMware seeks to be the vertically integrated stack that owns your future. I am leery of lock-in.

The last viable alternative is Openstack. Openstack has gone from utterly irrelevant also-ran to "proper infrastructure" in the past 18 months. Redhat, IBM, HP and Rackspace have been dumping quite a lot of resources into the various projects that make up Openstack and there are now a number of public and private clouds running it.

Like Microsoft and VMware's offerings, Openstack allows you to easily build hybrid cloud services. Unlike Microsoft and VMware, Openstack is incredibly modular. Individual elements can be swapped out to suit your needs.

Storage is a great example; whether your desire is object storage, file storage or block storage, there are many companies that offer off-the-shelf storage solutions for Openstack. These range from server SANs such as HP's Lefthand to traditional offerings like NetApp. Similarly, numerous projects exist to provide new-generation storage specifically built for Openstack, the most obvious of which is Swift.

In an Openstack world, if you don't like one provider you can simply swap them out for another. All that matters is API compatibility and seemingly every vendor is getting on board. Moving your workloads from your local environment to any other Openstack environment is just as easy as with Microsoft or VMware, but the whole thing was designed from the start with a paranoid dislike of anything that smacked of lock-in.

None of the hybrid cloud offerings on the table are perfect. All three have a lot of growing up to do, but they represent what is likely to be our best hope for actually building and maintaining a relevant IT skills base in a future that looks set to be dominated by the public cloud. When most of the world's businesses are renting their critical infrastructure from someone else, choosing to own looks to become a competitive advantage.

*Yes, yes, "the richest corporations and [insert edge case die hard nerd here]." If you made it this far into the article without wanting to stab me in the eye with a fixie you're probably a special case. Don't eat me for using a generalisation. ®