Where are all the thin clients then? Back in 1988, “thin client” was a buzz phrase, and these “fairly dumb” terminal devices were supposed to take over the world. The days of the complex and costly desktop PC were supposed to be numbered.
Somehow, though, it never happened. The much touted benefits – low cost, easy manageability, longevity, security and so on – were never enough to convince most organizations that server-based computing would be sufficiently flexible and reliable. There was also the teensiest bit of unspoken suspicion that the whole idea was an anti-Microsoft caper, designed to hurt Redmond’s revenues rather than aid potential customers.
Although thin client adoption is still very low (about 6 percent of the commercial desktop market), sales are increasing strongly from this low base, up 22 percent in the first half of 2006 compared to the previous year, according to a recent IDC EMEA study. The signs are that that thin client model may finally be catching on, in a small way at least.
Once reason for this is that there are now more ways than ever to use them. In the past, thin clients have typically been used in conjunction with applications running on a Microsoft Terminal Server or Citrix MetaFrame, the classic “server-based computing” model. (In fact MetaFrame has been replaced by Presentation Server 4, which runs on top of Terminal Server.) Now it’s also become common to access Web-native applications through a standard browser, which can run on a standard desktop machine or a thin client device.
But that’s just the beginning. The blade client model enables users sitting behind thin clients which are little more than smart monitors to run applications on rack-mounted PCs which are physically housed in a data center. Many would argue, however, that this isn’t really thin client computing – it’s more a case of placing the keyboard, video monitor and mouse in a different location to the PC.
Another innovative approach is to connect a thin client to a VMWare server running a personal computer in a virtual machine, which is similar in approach to the blade client solution. One could argue that a single server is hosting applications access by users with a thin client (a one to many server-based computing approach) but in reality the server is hosting multiple virtual machines, and each user is accessing a single (virtual) PC running programs which only he or she has access to (many to many).
Both these models offer the same benefits: users only require thin clients, so there are fewer costly and insecure PCs to manage around the workplace.
Before an organization can move its users to thin clients, it needs to be certain that they will be able to access all the applications they need. When applications are running on blade clients or in virtual PCs this shouldn’t be a problem, but what about server-based systems such as Citrix’s? While in theory it’s possible to run all but the most graphically intensive applications over a network using a traditional thin client architecture, in practice it’s often less than satisfactory for common productivity applications like Microsoft Office, says Chris Ingle, a research director at IDC. “It is possible, but they do run very slowly as there are too many disk read/writes,” he says.
Ideally, all applications should be Windows terminal services compliant, and in the past in-house applications which were not multi-user aware could present problems. But Citrix and others now have application isolation technologies so that multiple instances of the same application can be run without too many problems. Client-server applications should in theory work much faster, as the client and server are running on the same machine and can communicate much faster. Only the screen data – and changes to it – are sent over the network to the thin client.
It turns out that thin client architectures have been most successfully adopted in select industries including banking and call center environments, according to Ingle. “Thin client is at its best where users work with a restricted set of applications, like in banks or call centers, or in manufacturing where users only need specific supply chain apps,” he says.
To an extent thin client computing has been superseded by three tiered Web native applications running on browsers, page generation engines and databases. This is convenient because they can be accessed by any browser-equipped device on any platform (from Windows, Linux and MacOS based desktop or laptop devices to PDAs and even smartphones). More traditional server-based thin client systems need some access software such as Citrix’s ICA client, which is not nearly as ubiquitous as the common browser – even though versions do exist for these platforms.
But Web-native applications are making a surprisingly small impact on the server based computing model. Where developers are writing applications from scratch, building Web-native applications is feasible, but in reality most programs currently developed are standalone or client/server applications, Ingle says. So when organizations want a thin client approach, they have little option but to go down a Microsoft Terminal Server or Citrix Presentation Manager route.
In terms of hardware, the market leader is CA-based Wyse Technology, with HP and PA-based Neoware also strong in the market. Sales of Sun Microsystems’s Sun Ray thin client device, which acts a frame buffer and renders output sent from an Sun Ray application server, have failed to take off, however, according to Ingle.
Will thin client computing break out of the call center and a select few industries like finance? For it to do so the economic argument will have to be clear. Vendors stress the benefits of lower support and maintenance costs, cheaper thin client hardware and better security. But there are organizational and cultural barriers to be overcome – users like PCs on their desktops – and any thin client implementation will require appropriate skills to transfer applications to servers. There’ll also be new server requirements and additional licensing costs, and quite possibly extra bandwidth fees if WAN links need to be upgraded to accommodate traffic generated by branch office users.
Thin client computing may be on the rise, but it is not going to take over the world for some time to come.