The enterprise IT landscape: Five key ways it will change
Dave Cartwright peers into tech crystal ball
Blog The more IT managers (of which I'm one in my day job) I talk to, the more I head people wondering where IT will be going over the next five years. The times are, as a Mr. Zimmerman sang in 1964, a-changin': but what will be different? I reckon things will evolve in five key ways.
Storage in the cloud is already gaining traction, and it's only going to get bigger – at an accelerating rate. We'll still have a requirement for some on-site storage, but the motivations for adopting storage in the cloud come from two angles.
First, there are disincentives to having on-premise storage. It takes space and power, both of which are often expensive and/or scarce. It's a pain to upgrade: unless you have spare capacity you end up having to shift data to temporary storage while you replace disks.
You have to concern yourself with the resilience of your storage – although RAID protects you against individual disk failures you need to consider distributed and replicated storage for protection against site failure. And if you use tapes for archiving then you have to store them carefully.
Second, there are incentives for moving to cloud-based storage. The cost of renting cloud storage is now very reasonable, and is highly competitive compared to on-premise disk. And there is a growing number of appliances you can buy that let you connect transparently to cloud-based storage - and in some cases do automated or semi-automated tiering and shuffling of files between local and remove filestores.
The cloud is a big deal, and more of us are going to be using it to satisfy our processing needs. Similarly any company with a modicum of sense and more than a handful of servers will already be using a virtualised infrastructure in-house, or at least starting to move to one.
Despite the hype companies that move their entire application infrastructures to the cloud will be few and far between. Instead many organisations have - for whatever reason - a kernel of applications or data that they don't feel inclined to move to the cloud. This will lead to an increasing number of hybrid installations where the on-premise systems connect to cloud-based systems.
Over the next two or three years we will see an increase in appliances and management software that lets you control all elements of a hybrid environment as if it were a single entity.
The network as a commodity
Because the on-premise infrastructure will become simpler with systems being moved out to the cloud, the need for complex networking will reduce. This doesn't mean that networking will become trivial and everyone will start buying cheap-as-chips LAN switches from car boot sales, but what will happen is a shift in the network configuration from the switching and routing hardware and into the virtual server infrastructure.
Call it Software Defined Networking if you wish, but in fact people having been doing this kind of stuff for years until someone gave it a funky new name. So you'll probably still have routing being done in the layer 3 network infrastructure, but the config of the average router or layer-3 switch will be relatively simple: they'll route between VLANs and will present those VLANs up to the virtual server hosts to be consumed by the server admin guys.
IT becomes Business Systems
Although the server infrastructure won't need to grow particularly in either size or complexity, the same isn't true with the applications. Do you really think you'll have an on-premises email server in three years' time? I've not had one for five years now, and my email has been way more stable than it used to be.
Some applications just don't exist in any form other than in the cloud (Salesforce, anyone?). And others are both better and cheaper in the cloud than on-premise (Websense is my favourite example). Where apps remain in-house the support requirement is likely to be greater at the application level than with the underlying servers.
The traditional emphasis of the IT department - a crowd of geeks applying TLC to a fleet of servers, storage and operating systems - is changing shape. The infrastructure support team will still be there but it'll be much smaller and will comprise multi-skilled individuals that look after servers, storage and networks.
The headcount will move toward the application support people, both those with skills specific to particular apps and those who focus on application integration and directory services, as app integration is getting more and more extensive with time.
All of the above is something that my peers and I discuss from time to time (generally over a pint – well you would, wouldn't you?). But quantum computing is a concept that's starting to see more and more written about it – and is in fact being done in real life in some instances.
Will we have quantum computers in the next few years? Not in the average business, no. First of all the concept's in its infancy so while the big compute companies will be doing it in their labs we won't be able to afford to do it in our own businesses.
More relevant is the statement I just read in a description of Shor's algorithm (one of the most popular quantum algorithms in experimental setups): “This section may be too technical for most readers to understand”.
So IT will evolve over the next few years, with on-prem kit being easier to support and even easier to interface to the inevitable cloud-based storage and servers. But you're not going to hang your corporate IT hat on quantum computers in the corporate network for a few years yet ...®