Paul Durrant's Weblog

Paul Durrant's Weblog
Stream of (un)consciousness...

All | General | Music | Solaris

Main | The best album I've... »

Friday September 10, 2004

peer group pressure...

...leads me to participate in this apparent blogging revolution. It's not that I'm reluctant to participate, it's just that it's never been high on my to-do list.

So, as is customary, a short resume or, as we say in the UK, curriculum vitae...

I graduated from the University of Cambridge in 1994 and went to work for a small Token-Ring network adapter vendor called Madge Networks. Back then Madge was a fast growing business and Token-Ring was actually beating Ethernet in the performance stakes at 16Mbps to Ethernet's 10Mbps. Madge's business plan was pretty much 'Do it faster than IBM, for less money, in less space', the space aspect referring to the amount of memory the device driver's resident code would occupy under the good old 640k limit under DOS. This involved doing some pretty cool stuff, such as running a TCP/IP stack on the card and having the host based device driver code modify itself post-initialization to release all the memory occupied by the initialization code that was no longer required. Whilst at Madge I became involved in a small effort to write devices drivers for our Token-Ring hardware for the popular flavours of UNIX of the day: Solaris, Unixware and SCO Unix. This experience stood me in good stead when Madge's Token-Ring sales started to rapidly dry up (mainly in the face of the new, cheap 100Mbps switched Ethernet) and it became apparent that it was time to look for a new job.

I came to Sun in 1996, and joined the IMP business unit which was formed from the company Integrated Micro Products, which had recently been acquired by Sun. I worked in the Slough office on the ftSPARC and Netra-ft 1800 fault tolerant systems. These were pretty cool technology too. The basic principle was that they ran Solaris in lock-step on two independent sets of CPUs. An external bridge was used to compare the behaviour of the two sets and, if they differed, a fault was deemed to have occurred on one set. The boxes also were also at least dual-redundant in everything else, including network and disk. I gravitated back towards networking and ended up maintaining the Ethernet multiplexor driver. Since then I've been through quite a few other projects in Sun and now work for the Solaris Networking/Security Technology group where I maintain the Generic LAN Driver (GLD) module and am technical lead on Project Nemo which will take the straightforward interface that GLD provides and provide enhanced functionality and performance. We're seeing some very encouraging results from our prototype code but, since it's a work in progress, better not say too much more.

(2004-09-10 05:39:00.0) Permalink Comments [2]

Comments:

I remember IMP well, infact I nearly joined them in 1994. I was interviewing with Sun and SCO at the time. Their offices were a 3 level house with the servers on the ground floor!

Posted by Calvin Austin on September 10, 2004 at 11:26 AM PDT #

I remember IMP well, infact I nearly joined them in 1994. I was interviewing with Sun and SCO at the time. Their offices were a 3 level house with the servers on the ground floor!

Posted by Calvin Austin on September 10, 2004 at 01:28 PM PDT #