You are not logged in.
Pages: 1
Has anyone made significant use of the mdadm-based RAID functionality in LVM? Apparently it's been around for a while (introduced to RHEL in 6.3), but AFAICT it doesn't get much use.
But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner
Offline
Some people use it. Even so, I prefer mdadm.
Offline
In the end isn't it the same? I may be wrong but I have the idea that the tools will end up using the same infrastructure, they are just different frontends for different needs/use cases.
R00KIE
Tm90aGluZyB0byBzZWUgaGVyZSwgbW92ZSBhbG9uZy4K
Offline
@ROOKIE - In the end it's very different I think, because the RAIDing is done at the LV level. So you potentially have numerous independent RAIDs (perhaps with different RAID levels) and un-RAIDed LVs on the same set of disks. If you only use RAID 0, 1 or 10, you would probably end up with more or less the same thing as LVM on mdadm. But with parity RAIDs, the need to calculate the parity bit(s) for multiple RAIDs could be a bottleneck. Maybe only with higher-end SSDs, but still.
OTOH as you say there are potentially different use cases. That's part of what I was hoping to get insight into if anyone was using it. In my case, I could see using it to give me access to two RAID levels without having to be locked in to two fixed size mdadm RAIDs.
Primarily though, I made the post because I'm interested in mundane things like stability and recovery behaviour/performance. Also, I'm hopefully going to have some time to do a bit of experimenting soon, so if there is interest I might be able to fulfill some test requests.
But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner
Offline
It might share code but I don't think it's the same from a user point of view. mdadm works with distinct member devices, in LVM environment those are invisible. LVM only gives you the final result (the LV itself), everything else is hidden under the hood. If it fails and needs manual recovery, I do not know how well that would work with LVM w/o manually pulling stuff out of LVM's metadata.
Offline
Pages: 1