2. Introduction

If you or your organisation has been working with computers for some time now, then it is pretty much assumed that over the years, you have retired many a computer, and now have a warehouse full of those supposedly useless machines. Did you know that those discarded boxes could be put to better use? Or that each of those boxes could be made to run a modern operating systems at nearly no cost, whatsoever?

Surprise, surprise!! Linux can breathe life into that pathetic 486 of yours! And what's more: it could run at a decent clip compared to a Pentium III! The idea is as simple as it gets. Take a simple bare-bones PC with some memory, a cheap display card and a decent network card, and turn it into a workstation for office work or surfing the net, or for scientific computing. In educational organisations you could use it as a terminal for students to perform their practicals. Interested?? Read on...

2.1 A historical perspective

Way back in the late 1970s and 1980s, memory and hard disk space were insanely expensive, and there arose a need to find ways and means to make workstations more affordable. However, as displays became more and more sophisticated and preliminary GUI applications replacing their legacy character-based counterparts, this task seemed more and more daunting.

In those days of UNIX, universities and large corporations found a method to do this. They used the X Window System and gave birth to a new breed to displays - X Terminals. These systems had no local storage and very little memory. All they were meant to do was to run an X display for their own purposes in order to support applications running on the server itself.

As workstations became more powerful, the need arose to run applications using the processing power of the workstation alone. Storage, however, continued to be expensive and hence servers continued to handle the storage for the networks.

Soon these terminals, were the "in thing", finding favor with corporates and small organizations, alike. Management of such networks was simple with a centralized server being all that needed to be configured and maintained. A single point failure in the network was easy to rectify, and terminals that failed could easily be swapped with working systems.

But as processing power increased to deal with computing demand, networks could not cope with the immense traffic generated by simulataneous read-write operations from various hosts. With secondary storage becoming cheaper, diskless systems were soon things of the past and the benefits of centralized computing were lost.

However with the recent increase in network bandwidth, diskless systems have seen a sort of a revival and have found usage in many a computing applications, such as public reservation systems, information kiosks, set-top boxes, etc. Diskless nodes are once again being considered as a tremendous cost-saving exercise and centralized computing is seeing a renaissance.

2.2 Look Ma...No disk!!

A truly diskless system can eliminate the following storage devices:

• Hard Disk
• CD-ROM Drive
• Floppy Drive

2.3 The pros and cons of going diskless

The Advantages

Diskless systems have many advantages to offer:

1. Diskless sytems offer a very low-cost solution.
   • The RAM on the central server is shared by many clients. A single copy of a program is kept in memory and used by multiple clients. For example, in an internet cafe, despite the fact that various users on the network use a web browser, only one copy of the web browser is kept in memory, thereby eliminating wasted memory. Thus, by pooling memory on the server, memory costs can be reduced tremendously.
   • There is no need for a UPS or airconditioned environment for the diskless clients. Providing the server alone, with a UPS, serves the purpose.
   • The cost of the workstations, with typically low requirements, is minimal.

2. With no local storage, diskless systems can be made very secure

   These are ideal machines for students to experiment with. There is no local filesystem to corrupt if the machine crashes. Files on the server can be easily controlled or replaced. Diskless workstations are also safe from viruses, which function by copying themselves to the hard disk. The risk is limited to the central server.

3. Administration of a group of diskless systems is simpler.

   As discussed earlier, because administration is centralised, workstations can be added and removed with great ease. Backups need to be maintained only at the server end, and all configuration can be performed centrally. This is particularly useful when the network is spread over a campus or over several offices in a company.

4. Diskless computers offer high-speed performance.

   With the elimination of program loading time, diskless computers tend to work very fast. Once someone has loaded a memory hog like Star Office/Open Office into memory, the program loads in a flash for subsequent users.

The Disadvantages

With the advent of modern 100 MBit and 1 Gigabit ethernet cards, network bandwidth is no longer a consideration for the performance of diskless systems. However, with distributed computing having played a significant role in the proliferation of desktop PCs, the totally server-based solution represents a single point of failure. However, when compared to the central dependency that many modern systems have on various specialised, central servers, the diskless scenario doesn't look much worse. The bottom line is: Take care of your server!!