PC Alternatives for the Future

We covered software alternatives in last week’s column, so now it is time to look ahead at changes coming in hardware over the next few years, and the players driving those changes.

It is certainly time for a change. In 1984, we had two PCs in the business space. Apple enjoyed 40 percent market share and had two innovative designs, the Apple II and the Mac, neither of which survived in business.

Meanwhile, IBM had a large desktop computer that took up much of users’ desk space. That form is clearly in the minority now.

Since that time, Apple’s volume has tripled and the PC market has grown nearly 3,000 percent … which kind of explains Apple’s market share, doesn’t it? And yet, for the last 15 years or so, we have had largely the same tower and laptop designs with little real variation. That is changing — and for all of the right reasons. This week, we will look at a number of those changes, from the moderate to the extreme, and the vendors who are driving them.

Wyse/HP: Thin Clients

Thin client computing was largely a brain child of the CEOs of Sun and Oracle, Scott McNealy and Larry Ellison. The problem with this initially was that not only were these two guys unable to agree on their approach, neither of them understood the PC they were trying to replace. To make a better something you first have to understand that thing.

Oracle has long since abandoned the PC market while Sun remains with its SunRay 1, which has become an industry joke that laid-off employees use to partially explain Sun’s inability to compete. Unfortunately for them, Sun employees have to use this thing, and, as deployed, it is painfully slow.

Now, however, the thin client is enjoying a resurgence. Wyse is the leading independent vendor of thin clients. Of the large vendors, it is interesting to note that Hewlett-Packard is dominant here. After IBM’s exit, HP remains the only vendor in its class actively in this market.

As deployed, the thin client works as an updated terminal. It is nearly instantly turned on, enjoys a low hardware cost, and is generally more secure than its PC counterpart. It is also among the least frequently stolen devices — in some cases, thieves have actually returned them.

However, the thin client is also substantially less powerful and very dependent on the network it is on and the server that contains the data it needs to access. Its biggest problem is shared resources, as servers generally are not designed to handle the kind of processing that productivity workers require, at least not on a shared basis. That makes these things ideal for light work, data entry and kiosks.

Dell: Diskless Workstations

To gain the performance of a PC with the data security of a thin client without incurring excessive cost, Dell has rolled out diskless workstations. Much as they sound, these are workstations that access a server for their OS, applications and data.

These workstations enjoy similar security to a thin client, address the performance problem and can be redeployed as PCs if needed. They also tend to be one of the less expensive and least risky alternatives available. Like PCs, though, they put a lot of complexity at the end points where the expense is highest, they have more theft risk, and they are noisier than their thin-client counterparts.

ClearCube/HP: PC Blades

A PC Blade is very similar to a blade server and represents a card that contains a functioning PC. Some designs have hard drives, and some, like the diskless workstation described above, use a server for this.

PC Blades come in two flavors: Direct, which is currently only offered by ClearCube (partnered with IBM and Lenovo). ClearCube’s device simply uses a Category 5 cable to remote the ports and monitor to the desktop. It enjoys full PC performance, but requires a direct cable connection back to the blade (you can’t use switches, routers or hubs).

The other, built by both ClearCube and HP, is similar to the thin client and often uses a thin-client product at the desktop end. It has performance advantages over the generic thin-client product because the user has a dedicated processor, but it typically can’t handle rich graphics due to network bandwidth constraints.

This class has two primary advantages — uptime and resource allocation. Blades can be dynamically allocated and assigned much like trunks in a phone system. Thus, you don’t need one for every employee. You could, depending on implementation, share a single blade much like you would a more traditional thin client device. If there is a hardware failure, the user should be able to recover simply by rebooting, which causes a shift from the failed blade to one of the backups. Ensuring that the user doesn’t lose any work by retaining “full state” is possible, but it depends on the implementation.

PC Blades carry the highest initial cost of the solutions mentioned so far — with the possible exception of the SunRay 1 — but they also offer the best blend of performance, security and uptime, and are second only to thin-clients in maintenance costs. These are best-suited for environments that require PC-like performance but demand low noise, enjoy high labor costs and have a substantial security exposure factor, such as in healthcare and finance markets.

Cubix: Optical Hybrid

The Cubix, which just recently entered the marketplace, is a small rack-mounted PC connected via optical cable to a desktop device that splits the PCI bus. Kind of a hybrid between the Dell and direct connect ClearCube products, the Cubix optical hybrid provides the performance of the Dell with many of the benefits of the ClearCube solution, like low noise, low theft and higher security.

Because of the massive bandwidth of the dedicated optical line, this solution would be best for trading floor implementations, electronic signage, operating rooms and call center room displays. One other market is taking to this — the U.S. government — due to the vastly more secure nature of an optical cable.

Limiting Adoption

There are three things limiting the broad business adoption of these technologies today:

• Standards: The PC market is defined and was largely driven by a high degree of standardization. These products, even in the same class, are often too different from each other to swap out and most companies will avoid single source contracts like the plague these days. This is probably the critical path; once this is addressed, the others will likely follow.
• Fear of Change: Every one of these devices is different. In fact, the one clear advantage Dell has is that theirs is the least different of any of them. Companies don’t like change. The greater the change, the more difficult it is to get them to move. Often, a bridge needs to be created to help them on their way.
• Mobile: With over 50 percent of the new PC sales expected to be laptop computers, the fact that none of these solutions effectively address the mobile worker is problematic. HP used to have a line of Clamshell PocketPC devices that likely could have filled this need, but they were discontinued some time ago.

Given that wireless technology is becoming more ubiquitous, I believe the mobile problem is solvable in all but the Cubix solution — we won’t have the kind of bandwidth that the optical hybrid requires in a wireless solution anywhere but in our dreams.

Standards have been addressed in the past, but this issue has been incredibly difficult to address in the present. Still, there is progress on blade servers, suggesting there can be progress when it comes to PC blades as well. Change is difficult, but given increasing security exposures and how much governments like most of these solutions, I believe people will be forced to move whether they want to or not.

While every one of these solutions provides better uptime, only the thin client and BladePC solutions point towards a future where the PC is like the phone used to be and just works. Isn’t that something to look forward to?