Upgrade task that generates externalIds (623) takes a long time to execute

We have 80,000 user in our internal directory, and it took over 5 hours to upgrade versions of crowd. After speaking with a support engineer, this was considered an acceptable downtime to Atalssian. In fact, I was asked by them to plan for another 2 hours in production, just in case.

A 7 hour outage for an enterprise application is not only unacceptable but also shameful for Atlassian to consider it a reasonable time for an upgrade.

Please consider figuring out a better was to upgrade your database than line by line update statement. Create a load file, perhaps?