Enterprise IT Will Be Out of Infrastructure Biz in 5 Years

If your blood is boiling and you want to skip to the comments section and let me have it, that's fine. Just keep in mind that I purposefully wrote the title to be controversial and to open up a conversation that I believe is very relevant and that will shape enterprise strategies and the entire ecosystem of supporting product and services.

Let's examine the facts, shall we? Today, most (and, no, not all) datacenters are colocated and typically "lights-off" datacenters. That is, while we still purchase, lease and manage the infrastructure, it is very much hands off. Most of our operations are focused on the virtual infrastructure and what is left of legacy systems, like AS400. If you compare and contrast this situation with what enterprise IT was like in the late 1990s, you will find it very different. Back then, the focus was heavily on the hardware. I still remember that all the jobs I was applying for had requirements to have experience on specific hardware. Now, that requirement has almost disappeared.

Fast forward to today, and the cloud is finally staking a claim. The explosive growth and success of Amazon Web Services is a good indicator of a market shift. Couple that with: Google's plans to enter the public cloud IaaS space; Microsoft's Azure already battling AWS; IBM is already has a strong cloud presence; and VMware is about to enter the market. That's just to name a few.

Now, I know that when I bring up this subject I am immediately going to be confronted with security, regulation, cost, and a whole slew of other very valid concerns or reservations. But remember, I give it five years before enterprise IT exits infrastructure. So, if we assume that in five years most of the regulatory compliance, security, connectivity, and other issues would be resolved, what other barriers do you have for running your business completely on an IaaS?

You might be asking, what about legacy systems, large data sets or storage requirements? As far as legacy systems are concerned, in five years they'll probably be where they are now, considering that they can't be virtualized.

Storing large data, now that's a different story. Let's take the Amazon strategy on storage. One can easily make the case that storing large data on Amazon today is far more cost-effective given their sophisticated methods of moving the data into their infrastructure. But let's not assume that we can move large amounts of data today, so would it be reasonable to think that in fie years and given the tech landscape and amount of competition in this particular area, the cost of hosting your data in a public cloud will be far more appealing than the hassle and operational hassle of managing and refreshing storage on premises? We have examples today of enterprises moving large amounts of data into Amazon. So that's what is happening today, and in five years with all the momentum, I would say it'll be a given.

Let me give you more evidence that this shift is actually happening. Most vendors now, from Cisco, NetApp, EMC and others, have approved cloud service providers lists which they use to compensate their sales teams. That is, if a NetApp rep puts a customer into an IaaS cloud provider, he makes a commission. Now, you are probably wondering, why would they do that? For starters, NetApp recognizes that this trend is unstoppable and that they need to find a way to continue to monetize their hardware. In turn, they then find a new type of customer -- cloud service providers. Vendors will make a deal with cloud service providers to compensate their teams on bringing customers into the cloud, provided that these cloud providers dedicate some or all of their spend with these manufacturers. Now EMC can sell massive amounts of storage to cloud providers, and same goes with Cisco and the rest of them.

By now you are at least thinking it's not absurd and I have a point. Now, what happens to current jobs? We have millions of architects, engineers and experienced folks that design these large, complicated networks, storage networks, and so on. Well, their skill sets will evolve to manage everything in a "software-defined" infrastructure, or they will go work for these large cloud providers who will need their skills as they scale out massive infrastructures in multiple regions.

But will we really abolish all infrastructure from the enterprise IT? I don't think so. With the advent of the Internet of Things (IoT) and the potential for billions of new smart devices to be connected and be under the control and governance of IT, the network infrastructure will still exist in the form of wireless and LAN. Management could be cloud-based, but equipment will reside on premiseS to provide connectivity. Here's what it means in a nutshell: New roles and jobs will be created in IT to manage, govern these smart devices, but you can see there still is a role for traditional networking engineers. It'll just be a bit different than what it is today.

I am eager to hear your thoughts. What would prevent you as an organization that is 80 percent or more virtualized from completely moving to an IaaS provider? If there are constraints today, will there be fewer constraints in five years that infrastructure will be abolished? What have I missed that could be a show stopper? Please note that I am huge believer that enterprise IT will never be able to build infrastructures as reliable or secure as public cloud providers simply because of the sheer size and expertise that goes into building them.

Posted by Elias Khnaser on 05/01/2013 at 1:29 PM