Five-year Plan: 8 Problems IT Must Solve
There's a reason that so many businesses create five-year plans: If they're reasonable, they're achievable. Setting goals within that timeframe allows room for prioritization and opportunities to deal with the unexpected.
Of course, it's a little harder to develop a five-year plan for the entire IT industry. But the best place to begin is to consider the needs of IT people who map out the plans and get the job done every day.
[ Also on InfoWorld: Get off on the right footing -- put together your IT A-Team to crack the toughest tech assignments | Read Bill Snyder's Tech's Bottom Line blog to find out what the key business trends mean to you. ]
That's what we've attempted to do in creating this agenda of eight needs IT must attend to over the next five years. You won't find a lot of lofty talk about cloud computing or exotic technologies in the lab -- just a wish list for solutions to the big problems that get in the way of IT doing its job.
We don't expect everyone -- or even a majority of people who read this article -- to agree with our picks. Each person has his or her own axes to grind. We hope to hear about yours in the comments to this article.
Organizations large and small have been dealing with the increasingly creaky model of fat-client desktops since the dawn of the PC age. Admins still rove from desk to desk to upgrade and troubleshoot, and despite endpoint security advances, each desktop or laptop remains a big, fat target for hackers.
A variety of potential replacement technologies have made waves, but none with the all-encompassing feature set required of a no-brainer choice.
Take thin client computing. It fits some usage models very well, such as call centers and data entry applications, but most knowledge workers won't tolerate what amounts to a share instance of Windows. Other forms of centralized client computing, such as ClearCube's blade workstations, are in the same boat, meeting 100 percent of the requirements for a few markets and 0 percent for others.
The solution for this problem may come in the form of VDI (virtual desktop infrastructure), a slicker and higher-performance thin client model, an amalgam of the two, or something completely different. Bottom line: We need a new paradigm that extends the PC tradition of personal empowerment, yet sustains centralized IT control, security, and management.
Ultimately, portability should be part of the package, too. That way, users can take their desktop environments with them and work without a connection; when that connection is restored, so is IT control. Will some kind of secure, client-side virtual machine be the solution? Will users bring their own laptops or tablets and run that "business VM"? Maybe. But nothing like that is close to getting widespread traction yet.
If you were to go back in time just 10 years and mention that a 64-bit 48-core server with 512GB of RAM would be available for relatively cheap in 2010, they'd look at you funny, then wonder aloud about the possible uses of such a beast. Almost overwhelmingly, the answer to that question today is virtualization.
There's no doubt that virtualization is the path of IT for the foreseeable future. An essential part of that vision is huge multicore servers, each housing dozens of virtual servers, but the default configuration of those servers is nowhere near purpose-built for virtualization. It's time to change the defaults.
Most servers may now be aimed straight at the virtualization market, but they're still constructed for a single-server role. The additional hardware, heat, power, and size of these boxes don't do any good in a virtualized environment, and we could easily do away with them. Virtualization hosts need only three items: CPU, RAM, and I/O. Hypervisors can and should boot from internal flash devices or at the very least a 1.8-inch SSD, but the need for physical disk -- along with all its cooling and power requirements -- can be jettisoned.
A few entries in the server market fit this model to a degree, but they're all blades meant to reside in the appropriate chassis, all with local disk. In five years, I expect that ordering a blade chassis or a server with local disk to be the rarity, while diskless virtualization host servers will be the norm, with virtualization and SANs as common as keyboards.
Bring those servers to the smallest reasonable size possible, then go forth and virtualize. In 10 years, we'll tell our kids about way back when you could buy a server with an internal hard drive.
Far too many remote offices in today's world remain connected by ancient TDM technology. When dialup ruled the scene, those 1.54Mb T1 circuits looked huge, but now they're abysmally slow, yet cost at least as much as they did 15 years ago. There's no excuse for it.
If Verizon and others can roll out fiber to the home, they can certainly roll out fiber to the business. Whether your remote offices are in the middle of a city or the middle of the woods, there's bound to be fiber nearby. Barring that, the strides made in delivering high-speed data circuits over copper in the past decade make that lowly T1 look even older and slower.
The major problem there is that there's no impetus for carriers to move away from the T1 and T3 cash cow. They've been milking those circuits for eons and have established them as high-price, highly reliable circuits -- and they are. However, the wheel of technology has moved well beyond their capabilities.
In five years, it should cost no more to connect an office in the Michigan suburbs to an office in Virginia with a 100Mbps or 1Gbps pipe riding over a common carrier than it does to set up today's T1. And these links should be just as reliable as the T1 ever was.
I can probably count on one hand the number of IT professionals and end-users who've ever read an entire EULA. No doubt software licenses will always be written for the lawyers first and the users second, but the array of licensing schemes used by the huge range of companies is far too complicated. They can even interfere with IT's ability to keep the lights on.