Archive for Starlink

Personal Internet History

Posted in Biographical with tags , , , on April 9, 2023 by telescoper

The above graphic, which comes from here, has been doing the rounds on social media and has been eliciting quite a few responses from people of my age or older, about when they started using the internet, so I couldn’t resist a comment or two.

My first experiences of anything like the modern internet was using the computers at the (now defunct) British Gas On Line Inspection Centre (OLIC) in Cramlington, Northumberland. I wrote about this here. That would have been in 1981, in between leaving school and starting University (which I did in 1982). I also went back to OLIC work in the summer holidays while still an undergraduate.

At OLIC did quite a lot of coding (on projects related to pattern recognition), most of which was on VAX computers (and also the odd PDP 11/45). Incidentally, when I later started as a research student in 1985 I was delighted to discover that the STARLINK system in use at Sussex and throughout the UK was also VAX-based because I was already fluent in the command language (DCL) as well as the database software DATATRIEVE. Another reason I am grateful for the experience I gained at OLIC is that, working in that environment, I had to learn to make my code (which, incidentally, was all in Fortran-77) conform to various very strict standards which is no doubt why I am a bit of a stickler when it comes to scripts written in my Computational Physics lab!

Anyway, the Vax computers in use in OLIC and in STARLINK were connected by a thing called DECnet. This allowed users to send emails to other machines. The format of email addresses was much simpler than in use today, being of the form “host::username“. One could send files that way too; the alternative, via, FTP was terribly unreliable. DECnet provided a fast method of communication, but did require the receiving system to be accessible when you tried to send and would fail if this were not the case. Later email protocols would keep trying to send messages if at first they did not succeed.

While working at OLIC, where the powers that be were paranoid about industrial espionage, sending things this way essentially required the sender to log into the receiver’s machine, which terrified the systems people and it was soon blocked. STARLINK allowed this, however, as did the U.S. Space Physics Analysis Network (US-SPAN), the European Space Physics Analysis Network (E-SPAN), and by 1989 there were 17,000 nodes worldwide.

Sending emails outside DECnet was a rigmarole, involving including SMTP% followed by the external address. Messages inevitably had lengthy headers indicating the circuitous taken to reach the destination and often took some time to reach their destination. I was regularly using email in 1985, and if that counts as internet usage, then I started way to the left of the start of the graphic.

I didn’t use the world wide web until later. I’m not sure when, but it must have been around 1991 or so. The arXiv in roughly the form it exists now started in 1993, but it had precursors in the form of an email distribution list and, later, FTP access. Initially, it was based at Los Alamos National Laboratory (LANL) with a mirror site in SISSA (Trieste) that was used by those of us in Europe. In the beginning, arXiv was quite a small-scale thing and it wasn’t that easy to upload full papers including figures. In fact the SISSA system was run from a single IBM 386 PC (called “Babbage”).

The astrophysics section of arXiv (astro-ph) started in April 1992. Although astrophysicists generally were quick to latch on to this new method of distributing preprints, it took me a little time to get onto arXiv: my first papers did not appear there until February 1993; my first publication was in 1986 so there are quite a few of my early papers that aren’t on arXiv at all. In 1993 I was working at Queen Mary & Westfield College (as it was then called). I was working a lot with collaborators based in Italy at the time and they decided to start posting our joint papers on arXiv. Without that impetus it would have taken me much longer to get to grips with it.

4 Comments »

Starlink Memories

Posted in Biographical with tags , , , , , , , on December 3, 2014 by telescoper

This lunchtime I put on a black tie and went to the funeral of Stuart Keir, who died suddenly a few weeks ago at the aged of 55. Stuart had only just retired, on grounds of ill health, having worked in IT services for over twenty years here at the University of Sussex. When I first started in 1985 the observational astronomers at Sussex were primarily based at the Royal Greenwich Observatory (RGO), which was at that time located in Herstmonceux and Stuart was one of the staff there supporting the computer systems. The RGO moved to Cambridge in 1990 (for reasons which still escape me) at which point some of the astronomers and support staff relocated permanently to the University of Sussex while others went to Cambridge; Stuart moved to Sussex when the Starlink node was consolidated here following the loss of the RGO.

stuart

Stuart Keir

I remember Stuart very well from my life here as a PhD student and then as a postdoc in the Astronomy Unit, both before and after he moved full-time to Falmer campus. He wasn’t always the easiest of characters to work with – he insisted on doing everything his own way, which sometimes conflicted with what the astronomers wanted – but it has to be said that he did usually know better than us about the important things. He also worked extremely hard at his job and was definitely a great help to me during my time at Sussex. I’m very sad that he didn’t have long to enjoy his well-earned retirement.

The news of Stuart’s death made me all nostalgic for my days as a PhD student, when life was definitely much simpler. It also made me look back affectionately at the Starlink system as it was in the 80s. I had been lucky enough to get a series of temporary jobs, between school and university and during summer vacations at university, that involved the use of VAX computers of exactly the same type as that used in Starlink so I was well prepared when I started my PhD. In fact, looking back, I still think the native Virtual Memory System (VMS) and Digital Command Language (DCL) were terrific to work with. Unfortunately it seems that DEC, the company that manufactured the VAX systems, made some unwise business decisions and they lost ground to SUN. During the 1990s there was a rapid phase transition within Starlink from VAX machines to a range of SUN workstations.

Much of the value of Starlink in the old days was the connectivity it provided in what was basically a pre-internet (enabled by DECNET) which allowed sharing of software across all UK astronomy groups. In this sense it was an important step towards Open Science. The growth of the internet however let to a reduced role for Starlink from the mid-1990s onwards. Stuart transferred from Starlink to general IT services at the University of Sussex when Starlink was winding down.

In its heyday, Starlink was a great resource that provided a superb environment for astronomical researchers. It would have been nothing, however, without the dedicated team of Starlink Managers like Stuart who managed the local nodes and worked so help the astronomers that worked there. Data-intensive subjects like astronomy depend as much now as they did then on the hard work of computer support staff, an effort that is in my view not given anything like sufficient recognition by universities and other research institutions.

Rest in Peace, Stuart Keir.

5 Comments »