Subgit import performance

We are using directory layout option for a very large SVN repo (100 GB/72k commits). The configure takes minutes but the import takes days. I have increased memory on the windows server to 16 GB and have increased the java memory to 2 GB in the script.

Is there anything else I can do increase import throughport?


Hi Greg,

the configuration task is pretty fast, indeed, especially with layout directory option as in this mode SubGit only create needed file structure without even connection SVN repository. The initial import, on the other hand, does download actual data, so it’s a slow task essentially. 100 GB and 72k commits – it’s quite big repository, so it may take even days to translate.
As for he initial import performance – it strongly depends on many aspects, such as overall hardware performance – SVN server drive and network performance, SubGit machine drive subsystem and network performance, etc. From SubGit’s perspective there are some settings they can slow down the import, find the details in the following article:

Note, however, that EOL setting must be set prior to the initial import start, it cannot be changed on the fly.