So we were deep in the middle of upgrading the ONElist system to support all these new users. I was spreading out the functions of the system over several machines, and Scott was working on the new Oracle database. But in the meantime, I had to slow down the growth of ONElist otherwise the whole system would crater, to use the technical term. So on July 4, 1998 I turned off the ability for people to create new mail lists.
The last thing you want to do with a new company is stop its growth, but we had no choice. Each new user added would slow the database down a little bit. This was adding up quickly. Trying to use the ONElist web site to do anything was becoming very difficult, because each operation would take a long time. What I discovered during this time is that people would become impatient waiting for a web page to load. They would cancel the operation, by hitting the ‘Stop’ button in their browser and then try to redo it. This only exacerbated the problem. On the ONElist servers, the original operation still was being worked on, and now we had this additional operation (the repeat of the first one) to do. This resulted in a quick downward spiral to the point where the system became unusable.
July 17 – ONElist Upgrade Update
Earlier in the week, we had added 2 new PCs, making a total of 3 machines running the ONElist site. The 2 new machines were dedicated to handling email delivery. The other machine (the original ONElist machine) still ran the web site, database and archives.
Date: Fri Jul 17, 1998 8:23am
Subject: ONElist Upgrade Update
We are continuing the system-wide upgrade of ONElist. Earlier this week we completed upgrades that doubled our email delivery capacity. Email should no longer be delayed before delivery. In about two weeks we will be doubling email capacity again to ensure that we have plenty of room to grow.
Next up on the upgrade schedule is the ONElist web site. The first upgrade to that will occur late next week. Another upgrade will occur about a week after that. That will signal the completion of the upgrades. These two upgrades will require service interruptions; we will send out notices detailing this as the time approaches.
We thank you for your patience during our upgrades. We know there have been a few glitches, an unfortunate side effect of large system changes like this.
For the curious, our upgrades mean that ONElist is now a distributed, scalable system. Which is tech speak meaning that we will be able to increase our capacity easily in the future. We are using a Sun Microsystems Enterprise Server, a redundant fault-tolerant disk array, an Oracle database system, and several other high power workstations.
Thanks for using ONElist.
The ONElist Team