The Way of the Dinosaur

With the cry, “The mainframe is dead!” the PCs entered the mainstream corporate computing world. Mainframes were dinosaurs, and their inevitable extinction was the next page in their history. But let’s look at this analogy more deeply.

The dinosaurs became extinct following a global environmental catastrophe. There was no war between the mammals and the dinosaurs. The mammals did not out-compete dinosaurs in any ecological niche. An external event radically changed the environment, eliminating the dinosaurs, who had successfully dominated the planet for 180 million years. Dinosaurs suffered a huge environmental catastrophe 200 million years ago. Only half of all dinosaur species survived, while a species which showed both mammalian and saurian characteristics failed, leaving true mammals to evolve in the shadow of the dinosaurs.

And so how does this inform our understanding of the competition between the PCs and the mainframes? The mainframes did dominate the corporate landscape for generations; they were big and capital-intensive. PCs provided an alternative processing mechanism initially for spreadsheets and ultimately many traditionally host-based processes. PCs learned how to connect for client/server computing and eventually to use the Internet for browser-based information access and analysis. So PCs competed successfully for a presence in a series of ecological niches that mainframes had once dominated.

But there is another environmental catastrophe forming. This shock will transform the computing landscape. The transformation is the green revolution. Fossil fuels are becoming a costly and undependable energy source. While the transformation to alternative energy sources is underway, conservation is now an imperative. Business will consider any measure to reduce energy consumption. Building design, telecommuting, and outsourcing all shift the energy burden away from the core business. IT consumes significant power. Measures that IT can take to reduce energy consumption get high marks, are scalable, and have measurable impact on a business’s overall energy use.

Personal computers consume a significant amount of energy. A desktop computer consumes 200 watts, and generates additional energy costs in removing its heat from the building environment. A thin client workstation – and the energy in the data center to support its computing – consumes 25 watts. Converting from desktop computing to thin client computing cuts energy costs by a factor of eight. While one physical server can host five to ten virtual servers (based on typical CPU utilization), that same physical server can host a hundred or more virtual desktops. Virtualization can reduce energy consumption by an order of magnitude or more in a medium to large enterprise.

What other benefits and risks does the organization face when converting from desktop personal computers to thin clients? On the benefit side, the data remains in the data center, so data loss because of a desktop error does not happen. The firm can deploy comprehensive backup and disaster recovery mechanisms. No user’s data would be lost because they forgot to connect and run a backup job. Also, support and maintenance costs drop. The firm will not need to keep a spare inventory of parts for multiple generations of PCs, and the tech support staff will have all the diagnostic information in the data center. Users won’t have to bring their PCs to the help desk to get software installed, and they won’t have to run virus scans or software updates to stay secure or remain current. These processes can be built into the virtual desktop environment inside the data center. Information cannot be stolen from a thin client, since it does not leave the data center. No user can insert a USB drive and download files, or lose a laptop with a hard disk full of customer records.

What risks does a firm face when migrating towards virtual desktops? There are some applications that don’t play well when virtualized – heavy graphics and 3D modeling, for example. These need an unencumbered host with a huge amount of available capacity and may not render well across the link between the virtual desktop and the screen. If the user works at locations where connectivity is problematic, he may need the entire project on his laptop, a fat client device.

More importantly, when firms virtualize their networks they may not have as much visibility into the network activity between virtual desktops and servers. And, in this virtual network, they may not be able to track which users are where. Users’ virtual desktops may move to balance load or recover from an interruption in service. Most importantly, the traditional reliance on a perimeter, whether for security, systems management, capacity planning, or compliance, vanishes in the virtual world. This requires clarity in defining business service level objectives. Policy cannot be imbedded in network topology as it was in the 1980s and 1990s.

So the battle between the mainframes and the PCs may turn out a bit differently than that between the dinosaurs and the mammals. The impending environmental catastrophe threatens the power-hungry PCs, and the large hosts, which efficiently parcel out computing, storage, and bandwidth, across a broad population of users, may prove to be the more adaptable and responsive creatures in this cyber landscape.