Changing the destiny of man and machine
A few years ago, while I was an undergraduate majoring in CS, our department received a shipment of desktops as donation from Intel Corporation. They were good systems for that era: up to quad processor Pentium servers, Pentium Pro clients, and the works. Of course, all were pre-configured with one of Windows NT Server, Windows Workstation or Windows 95. Till then, the department had a few HP and Sun workstations, and contention for computing resources used to be high, in general.
I was supposed to be the department's most "enthusiastic" student, so I was asked to suggest as to what exactly to do with the machines while a big room was being renovated (which would be the Intel Lab). As for me, I ran Linux everywhere at that time (well mostly, except in places where I tried *BSD, HURD, Minix and the like). I even ran Linux on some tycoon's laptop who was seeking my help regarding some "issues" with Microsoft Office - unbeknownst to him, courtesy UMSDOS. Naturally, I suggested Linux, but there was skepticism. Some like minded and well-informed professors fortunately agreed, and everybody thought it wouldn't do any harm to put Linux on one or two machines (there were about twenty five in all). The Intel Lab was inaugurated, with Windows on most machines.
It wasn't long before the academic community started feeling itchy due to the user fiendliness of Windows, and I was quick in converting a few more PCs to Linux. The Linux bandwagon guys in the department would go about explaining to whosoever listened the utility of Linux. The number of Linux users started increasing ridiculously as people found the system a joy to use - documentation was available aplenty. Performance was great. We promptly introduced NIS. Home directories over NFS followed. Linux captured two SMP machines (formerly running Windows NT Server) soon as we made them Linux servers. SMP was big deal at that time. Very big.
Managing the large cluster was becoming harder because only a few students and a couple of professors had the inclination (not necessarily the time!) to do so. To reduce upgrading woes, /usr was put over NFS (coming from a Linux server). This worked fine, so we put even / on the client machines over NFS - that is, the root directories of all clients resided on the servers. Adding a new machine to the cluster became a breeze - add its entry to a few configuration files on a server (DHCP was being used), use a boot floppy on the client and it would come up with a graphical Linux login - all configured. One client's hard disk went bad, so I programmed the boot PROM on its Ethernet card to make it boot off the net without a disk or a floppy.
By now, the Intel Lab had grown tremendously - in sophistication and in user base, and of course, in reputation. We attached a few modems to the servers and thus had dial-up services in the department for the first time. PAM was there for access control. Auto-mounting was used in most cases. Requests for access to the Lab used to pour in from people in other departments - it was ironic (but encouraging!) to observe guys with hefty Silicon Graphics, Sun, DEC and HP servers in their labs craving to use Linux!
All this activity was academically motivating - people were doing numerous projects - from inside-the-kernel ones to Java applets (not to imply that in-kernel and Java are two ends of a spectrum!). Many people discovered a new (at least hitherto unknown) meaning of computing: here was a system that you could understand, modify, be constructive and creative with, and not worry about somebody suing you for violating their intellectual property rights. For example, somebody wrote a centralized boot-up/shutdown utility. Several original and innovative projects were conceived and successfully implemented in the Intel Linux Lab - come to think of it, several hundred CS guys: undergrads, grads and doctoral candidates - all doing their minor and major projects, and research work on Linux! Areas such as voice recognition, image processing, operating systems, high speed networking (including ATM), multimedia conferencing ... you name it, and there was somebody doing something on it (a group of guys worked on process migration too!). No wonder the Intel guys seemed pleased whenever they visited the Lab to see how their equipment was being used.
Some of us began writing Linux related articles for magazines and trade journals. These were extremely well-received by the readers. There were some who started offering consultancy - a few even set software companies! Often a Linux enthusiast from amongst us would give a talk on some aspect of Linux, and each time the conference room would be full. Perhaps the biggest contribution of the Lab is the number of geeks it has given birth to.
Those in charge of the Sun/HP machines asked us to do something about them too - so we put their machines on the Linux cluster. Soon after, we installed SPARC/Linux on most of the Sun workstations, remote booting on some of them. I wanted to convert the religion of the Ultra's too but they wouldn't let me! Today, the Intel Lab is certainly state-of-the-art. Intel has recently donated new machines - including Xeon servers. The University's Web server, the Email gateway, the DNS server, the NIS server, the NFS server, the DHCP server, the PPP server, in fact, almost absolutely everything - is housed in this Lab. Linux reigns supreme here, though Microsoft CDs and manuals can still be seen inside glass cupboards.
May, 1998