The OS for X

Blogs

(Source: Stockfresh)

There has been a recent trend among some of the larger IT vendors which is an ambition to become the OS provider for X.

Now this is not some hark back to those algebra classes where you had to work out X, rather X is interchangeable, depending on the specific direction of the vendor in question.

For years, for example, VMware has been talking about, and actually delivering, the operating system (OS) for the data centre. But now Intel is talking about the OS for Big Data, and Microsoft is putting the finishing touches to Cloud OS, the OS for…you’ve guessed it!

But what does this all mean? Does the analogy actually carry? Or is it just a load of old marketing hyperbole?

What seems to be happening is something of a backlash against some of the early trends in both virtualisation and cloud computing. First of all, when virtualisation began to get popular in around 2005, there were precious few vendors in the space. VMware was head and shoulders above everyone else, IBM was calling old hat (though not in x86 space) and Citrix was the smaller player, with Microsoft a distant third. Red Hat was in there too, but that was very much in its own niche. Irrespective of this small cast of players, what happened was that the unfettered access to infrastructure led to exactly what had happened previously: sprawl.

When x86 servers became cheaper to buy than their yearly running costs, many companies became careless about their deployment and stacked them high with little care or thought to management, as demand had to be met. Then demand fell off a cliff and wasn’t seen again for quite some time (2008). Suddenly, power and cooling were an issue and server consolidation accelerated rapidly, facilitated primarily through virtualisation. But then the same thing happened again, as the people commissioning a VM had often never seen the physical server they were given before, now the VM responded in exactly the same way so they didn’t care — except for the fact that many could now provision themselves, self-service, and so they did, in droves. This gave rise to VM sprawl. Suddenly storage capacity was eaten up, back-up times went through the roof and an altogether different crisis was born. But, it was an easier crisis to solve as features such as auto-tiering, enforced expiries and other environment management features made possible in virtualised environments brought things under control.

All the while, the creeping appeal of cloud computing was making people see that there was another way to go, apart from having whacking great servers in massive data centres that cost a lot of money: have someone else do it and simply control it all down a big fat pipe. But the early days of such things saw much ambivalence, as on the one hand there were stories of Amazon Web Services allowing jobs to be done with its elastic infrastructure in days that would have taken months and costing peanuts, to horrible stories of vendor lock-in where once something was sent to the cloud it experienced the equivalent of indefinite incarceration, as now it couldn’t get out again.

Vendors worked hard to allay those fears and interoperability, open standards and the development of hybrid capabilities all conspired to make users realise that actually, it is more of a case of risk appetite and comfort levels with vendors and skillsets that determined just how locked-in you were, or were prepared to go.

“In reality, none of this adds up to an operating system for X. But it does successfully convey what the vendors are trying to do for customers: provide a fully integrated, but interchangeable, stack of tools that can combine to provide a robust infrastructure to achieve the desired balance of on-premise and cloud, self-owned and managed and outsourced, and purpose-built or multi-purpose”

So now vendors looked to fill out their respective stacks and offer as much as they could at every layer of infrastructure and service to try to catch those wavering, or undecided about what bit to put where. The end result is that the fear of vendor lock-in was being countered with the idea that a comfort level and experience with one vendor could be capitalised upon to allow that single vendor to be used in the full stack, whether it was Oracle, IBM, SAP, VMware or Microsoft, and now even Intel, depending on the primary purpose of said stack.

In reality, none of this adds up to an operating system for X. Even Microsoft’s Cloud OS, which in fairness makes a good case as a tight integration of the critical elements for running either an on-premise IT infrastructure or one in the cloud, the analogy of an OS for that purpose is loose at best. But it does successfully convey what the vendors are trying to do for customers: provide a fully integrated, but interchangeable, stack of tools that can combine to provide a robust infrastructure to achieve the desired balance of on-premise and cloud, self-owned and managed and outsourced, and purpose-built or multi-purpose.

This is a manifestation of the realisation among vendors that the old, often self-imposed, labels of ‘IBM house’ or ‘Microsoft house’ are largely consigned to history and the reality is that brand loyalty is not what it used to be when margins are tight, competitive advantage is everything and agility in all things is one of the only mechanisms to deliver the former. In this environment, if something isn’t quite up to scratch, say an API for an on-premise application or a cloud service, then it will be binned at the earliest opportunity in favour of something else that will do the job. These days an application, a service or a piece of kit not only needs to work well in itself, but it also needs to be able to integrate with every other component in its environment much more than ever before. Therefore, vendors seem to be pinning their hopes on the well-worn idea of the OS to express just how well they have achieved this for your benefit.

Are people buying this idea then of the OS for the data centre, the cloud, Big Data or indeed, the agile enterprise? Well, yes and no. As is often the case, many organisations are being told that they are already using the X OS because they have this app on that platform talking to these services and so they are, in effect, doing it. Organisations are rarely in a position to declare ‘We will deploy X OS’, but many may deploy several of its elements as it goes about its own strategy. Doubtless, we will see marketing campaigns that will put forward some company or other as running the OS for X, and lauding its abilities. But the underlying trend seems persistent. Tight stacks, carefully integrated without losing interoperability. Just like an OS, only better.