Hi,
The amount of time required to migrate from 2.x up to 3.0 is insane. For our database with 200G and 2 million nodes we are looking at a couple days to successfully migrate from 2.x to 3.0.
This includes
- Full database restructure of every table numerous times (eight hours or so)
- Index Rebuild (Can be skipped partially). For our repository a complete rebuild takes about 12 hours or so (even using the MT rebuild in 3.0)
- Permission migration patch (This was estimated to take two days until we hacked around in the database to assign nodes to a default permission). If lots of nodes have alf_node acl_id as null you will be waiting for days.
- Version Store Migration Patch (A complete index rebuild (non-multithreaded of course) of the version2store. This is estimated to come in at another 8 hours.
- And we are not done yet!!!!!
How are we supposed to schedule a downtime window to accommodate this down time when the switch over to production is to take place?
Please note that we are using HP ML350 8 core machine with 6 gigs and RAID 5 SmartArray so our hardware should be pretty typical of enterprise customers.
