I'm reading the TR-3929 Reallocate best practices document. I've ran "reallocate measure -o" on all volumes across all our V-series filers. Most of the volumes come back with an allocation check of over 3, meaning a reallocation is recommended. Not really surprising since I've grown the aggregates over time with array LUNs from the backend storage.
The TR recommends physical reallocate, but also states that this is not possible for RAID0 aggregates. But since I get RAID protection from my backend storage, I of course run all my aggregates as RAID0. Does this mean I can't really do reallocate? Or that the traditional reallocation would be my only choice?
Unfortunately, the process relies on RAID, and since we don't do any RAID within ONTAP, the physical reallocate (reallocate -p) will not work as currently designed on aggregates using third-party array LUNs.
I just chatted with the author of that TR (sits down the hall), and he was curious about what the exact measurements were. What number exactly? He was also wondering what type of workload was supported by these aggrs. Can you share that info?
Re: Reallocate not possible in RAID0, so unusable in V-series?