Then, after successful debugging your job isn't finished. The outline of "Three Questions About Each Bug You Find" <http://www.multicians.org/thvv/threeq.html> is:

1. Is this mistake somewhere else also?

2. What next bug is hidden behind this one?

3. What should I do to prevent bugs like this?

Not relevant to the article, but for the history of UK academic networking mentioned in various comments, Wikipedia's account looks about right, though I'm not sure it's up-to-date concerning the regions: https://en.wikipedia.org/wiki/JANET

I was using SRCnet in 1981, when Liverpool Physics had a dedicated link to Daresbury (national lab) whose speed I don't remember at that stage. Unfortunately the infamous PDP11 "terminal concentrators" for interactive use then were horribly unreliable. RJE to the cloud, where analyses ran, worked well.

Yes, SRCNET/SERCNET/JANET were great as a physics researcher, despite what people have said about X.25. A potentially interesting point is that TCP/IP on JANET originally ran over X.25 until X.25 was finally phased out.

The "fat pipe" didn't look so fat at that stage! (I don't remember when you could first easily interact with the Internet.)

Red book was particularly interesting (modulo lack of security) long before "the Grid"; it worked between various computer centres.

The trouble was that it was quite unclear to a researcher, even in one of the research council networking hubs, how to get access to the gateway, and it may have cost. I gave up trying before going to work in Oak Ridge for the summer (where I was taken aback by the primitive computing, at least "outside the fence"). For some time (mid-80s to early 90s? I don't remember) we were generally dependent on the infamous BITNET email gateway to communicate with the rest of the world from the well-developed UK network. It was "interesting" to deal with code in a Swedish 6-bit character set sent through the EBCDIC gateway to ISO 646-GB. (The Fortran Hollerith formats were added interest...)

Jim Austin has one: https://www.computermuseum.org.uk/machines/sun_e10000.html

> a guy screaming into a rack of spinning rust and watching read latencies spike

https://www.youtube.com/watch?v=tDacjrSCeq4

That’s the one! :) Uploaded by Bryan Cantrill no less.

> Sparc is not for performance, particularly for benchmarks

It certainly was from 2011 to 2019: https://en.wikipedia.org/wiki/K_computer

> obviously you need to write your own software

Kerberos ran fine for me on cast-off SPARCstation 10(?) which was well obsolete at the time.

The last Sun systems I operated -- x2200s not up to Sun standards -- ran GNU/Linux on Opteron in an HPC cluster.