When I was a young sysadmin at an ISP in the mid-nineties, we had a shell server that would be exploited fairly regularly. This was in the days before automate

When I was a young sysadmin at an ISP in the mid-nineties, we had a shell server that would be exploited fairly regularly. This was in the days before automated updates, when the standard Sun delivery method for patches was a mailed CD-ROM. (Online patches were available, but you had to seek them out, and even then they occasionally lagged behind the exploit code being posted to bugtraq by a few days.) A big part of "securing" this particular machine was to separate it from the rest of the network so that even if compromised, it had no more access than the outside world. I set up an RSA private key for SSH authentication so that I wouldn't have to provide my password to a possibly compromised sshd. I backed up the tripwire database to removable media and checked the integrity of the machine's binaries regularly. Shell access as a standard feature of dial-up internet access died right around that time, and I don't think many ISPs were sad to see it go. That and Usenet were huge time/resource sinks that I'm not sure provided a lot of value in the end to our average customer.