With 2013 fast approaching, IDG Connect is serializing commentary from industry experts on Gartner's 2013 predictions over this week. Read our full series of pieces dedicated to the technology trends of 2013.
According to Gartner, the market in 2013 is expected to shift toward integrated ecosystems, driven by user demand for lower cost and simplicity. While it's true that the IT industry is seeing greater and greater convergence of the components making up the IT stack, those same vendors are hedging their bets and aligning themselves as part of several other reference architectures, meaning there are even more options available for IT purchasers than ever before. With this, there is now the distinct advantage of acquiring a switch that is pre-integrated with a server, which is pre-integrated with a storage component and so forth. Often these integrated systems are from a single vendor, leading to lock-in with that particular vendor.
On the flip side, not all vendors contribute to the lock-in. There are behemoths like Intel that play well with Dell and other hardware manufacturers, as well as with the OS and hypervisor providers who are all at war with each other. Yet those same hypervisor providers such as VMware and Windows and Xen, all play nicely with each other at the application layer. It's even gone so far in the cloud space that we see iterations of database applications from IBM, for example, within the AWS cloud; and even the newest version of Windows now available on AWS; which in turn competes with Microsoft's own Azure platform. Add to that the ecosystem of reference architectures, and cataloguing of virtual instances, and the variety of applications available, and the choices can become quite confusing to enterprise decision-makers.
So does that mean greater simplicity to the operators? Aside from the obvious complexity in decision-making, there is enhanced deployment simplicity due to the pre-integration of systems and associated applications in the ecosystem. However, to assume that this has driven out cost is to incorporate only the high costs associated with IT staff and integrating these different IT components, rather than the cost of the technologies themselves.
What is often missed is the fact that the last point of convergence is at the management layer of these infrastructure systems. This is where new costs are introduced, including the cost of inefficiency on having to maintain multiple management systems. Add to that, the fact that often, theoretically integrated stacks are a result of recent acquisitions by a tech vendor, and the markitecture may be well ahead of the actual delivery mechanism. There is still the risk of incompatibility between components. Similarly there is the lock-in factor and inability, often, to utilize one's pre-existing IT infrastructure - meaning that it is every bit as important to ensure that you are buying into highly extensible technologies that allow for the maximization of investments made in other pre-existing technologies, unless you're willing to wait for your vendor to introduce those capabilities themselves.
Needless to say, the growing trend is to move higher up the stack and away from the capital investment of hardware, and toward the self-control of an application or instance, made available via IaaS, PaaS or simply from an application bazaar or marketplace. The latter is most familiar to end users of mobile app stores.
Something similar has been happening over the years in the ISV space, and amongst developers (including enterprise DevOps teams) that are increasingly looking for agile ways in which to solve business productivity needs for an enterprise organization. But why recreate the wheel when someone else can offer you either the mainstay of the code you're looking for, or the entire application itself?
While in the mobile space those apps tend to be fairly rigid in their usability by the end consumer, in the business world the expectation is to have some level of customizability. Salesforce.com's developer force.com has done a commendable job of opening the door to innovative development to answer that need for customizability, yet locks developers into its own platform protocols. To be fair, one could say the same thing about Azure or Google AppEngine, although in both those cases, the door has been opening to allow for numerous app runtimes to be facilitated.
The next step in this game is for more thorough control over the non-rugged (read insecure) code being deployed so rapidly into these ecosystems, and enhanced mapping of dependencies and management capabilities from the code-level all the way down to the datacenter facilities, to enable smarter workload operations, and with it, cost efficiency.
By Antonio Piraino, CTO, ScienceLogic
Phil Muncaster reports on China and beyond