The past weekend I was helping a customer perform several P2V of physical servers to VMware virtual machines. Those were critical applications with small downtime window.
Generally P2V are pretty straightforward and easy to do. Given enough time, any x86 physical workloads can be migrated. In the ones I just did, all I had was 4 hours each. On the surface, 4 hours for about 120GB worth of data was achievable over 1Gbps networks. For these, we needed a lot more time. We did a migration on one of the machines which had no issues with a tight window, it took 16hrs!
The cause, nearly 1 million small files!! Small files are the bane of storage, which are a pain to copy.
We had to use different techniques for the P2V. Block based copying was not an option, and so it would have to be just regular file based copying.
The solution? P2V the OS disk only, then copy the other drives using FastCopy.
FastCopy was much more efficient that standard file copying utilities. It is known to be very successful in copying small files, and it’s true! Even for copying over the network! Though it’s still far from being as fast as block copy, but it was certainly faster than leaving it to regular copy/migration means.