We’ve already written about how converting standalone PCs to dumb terminals can simplify and standardize your computing environment. A great way to enhance the terminal strategy is to use a Linux LTSP server to “power” the dumb terminals.
Dumb terminals require a basic operating system (OS) that provides basic keyboard, video, mouse, network, and a remote desktop client application (for connecting to the Windows session). This basic OS can be “embedded” on the terminal hardware (typically as a miniature installation of Windows or Linux), but an embedded OS still requires individual installation and configuration on the terminal hardware. In our view, it is far superior to download your terminal’s OS over the network using an LTSP server.
LTSP stands for “Linux Terminal Server Project”. LTSP’s terminal implementation is different than that of Microsoft Remote Desktop Service (RDS), because LTSP includes the basic OS functionality described above. In other words, you can “boot to” LTSP over the network… something you cannot do with Windows.
But, this is not about an alternative to Microsoft RDS. It is about making it simpler and easier to maintain terminals that connect to Microsoft RDS. LTSP replaces only the basic OS that otherwise is likely embedded on the terminals.
LTSP works around the shortcomings of embedded terminal OS. Embedded OS terminals can be slow to flash a fresh OS. They can be slow and tedious to make configuration changes. Software to centrally-manage embedded OS terminals can require additional licensing expense. Eventually, the embedded OS may become obsolete (no longer secure) and you may then have to purchase entirely new terminals. While embedded OS terminals are still better than maintaining standalone PCs, they are not completely ideal.
With LTSP terminals, the OS is centrally managed. Just like the Microsoft RDS server, you can apply updates and configuration changes just once, and all of the terminals benefit. Because LTSP is free and compatible with a wide range of devices, you can convert otherwise obsolete desktop PCs to viable, up-to-date terminals.
We do have a few technical tips for anyone planning on deploying LTSP clients. First, consider configuring LTSP to default to “Fat Clients” (Google for how to do this). The difference between a “thin client” and a “fat client” is that the thin client depends more on the server’s computing resources. With a fat client, the OS is downloaded and executed on the terminal, which tends to increase performance. Unless you are using exceedingly ancient terminal hardware, there is no reason to forgo the benefit of fat clients.
Another technical tip that we have is to use terminals that have Intel processors and video chipsets. Particularly at the low-end, performance suffered when using non-Intel video adapters. The slower screen refresh rate was particularly troublesome when scrolling in Excel or in a web browser. The video drivers available in Linux for Intel adapters seemed to be superior to the drivers available for the competing adapters we tried.
Our final technical tip is to consider using LIKEWISE-OPEN to authenticate LTSP clients to Microsoft Active Directory (AD). Integration with AD will further reduce administrative overhead.
While not a required part of a terminal computing strategy, deploying an LTSP server and clients can further reduce the cost and complexity of maintaining your IT infrastructure. It also provides flexibility to convert your existing PCs, which helps manage cost and risk when first converting from standalone PCs to terminals.