I’m a bit looking forward to it, given that you are rebuilding the array, there is a big read/write burden on the controller, but I found that I really don’t have anything to compare it.
See Severe (frequently freezes for 5-10 seconds) expected behavior during RAID reconstruction and heavy read/write usage (performance is significantly affected during backup and when users download large [multi-GB] files via FTP).
Any thoughts on this would be greatly appreciated. This is a hardware RAID6 (LSI 9266-i8) installed locally on a 40TB array on NFS (i.e. the server is physically very close to the workstation).
< /div>
RAID rebuilds and how they work pre and post failure.
Now, in terms of my view on rebuilds, we do know that rebuilds will bring some very slow performance, and this is a matter of course. As you will As you can see from the link above, RAID reconstruction not only extracts the information from the failed disk to the well-known disk (rebuild after a failure), they also write the information to the system drive and other data/auxiliary drives while the server is running The other thing to keep in mind is that the commonly used functions that you usually see do not take time, and relatively few resources now occupy more resources than normal, and taxes are imposed on the servers that have already been taxed. In the case of pre-rebuild failures Situation (performance is slightly better, but not a lot) you can be lucky and have a drive (logical or physical) fail and rebuild the RAID before the end user (hope you as an SA should have some kind of alert system, so you don’t Should be surprised) or even know of any problems.
The 5-10 second freeze you see is absolutely normal, especially if the server you are rebuilding is any type of database server, which is written by default And reading are higher than usual (that is, the SQL server of the database containing the end user’s access rights) for a whole day; a property management company I once consulted has a program that can access their tenant records all day for viewing and Write new information, and it is always used a lot.) It will be more noticeable.
The other thing I suggest is to get any RAID utility (GUI version) that comes with the controller And install it in the operating system , So you can monitor the rebuild without loading into the controller BIOS.
Nowadays, a very small and almost non-existent problem is NFS vs. iSCSI. I know you are using NFS, and in the past The situation is that iSCSI will have better overall performance in the case of virtualization, but with recent improvements to hypervisors and hard drives and controllers, the performance of NFS is almost the same as iSCSI, so it sounds like you have a very good SAN.
I am happy to answer anything you need to know, so please feel free to comment.
Such a quick question – our RAID6 array It is currently being rebuilt, and there is a very obvious loss of file system performance (the home directory is NFS installed on the array).
I am a bit looking forward to it, given that you are rebuilding the array, so there is A lot of read/write burden, but I found that I really don’t have anything to compare it.
See serious (frequently freezes for 5-10 seconds) expected behavior during RAID reconstruction and a lot of reads /Write usage (performance is significantly affected during backup and when users download large [multi-GB] files via FTP).
Any thoughts on this would be greatly appreciated. This is installed locally on NFS Hardware RAID6 (LSI 9266-i8) on the 40TB array (that is, the server is physically very close to the workstation).
First of all, this is a good Resource, which outlines the rebuild time.
RAID rebuilds and how they work pre and post failure.
Now, in terms of my view on reconstruction, we do know Rebuilding will bring some very slow performance, which is natural. As you will see from the link above, RAID reconstruction not only extracts the information from the failed disk to the well-known disk (rebuild after a failure), they It also writes information to the system drive and other data/auxiliary drives while the server is running. Also, keep in mind that the common functions you usually see don’t take time, and relatively few resources now take up more resources than normal There are many situations, and taxes are imposed on servers that have already been taxed. In the case of pre-rebuild failure (performance is slightly better, but not a lot) you can be lucky and have a drive (logical or physical) that fails and is rebuilt before the end user RAID (hope you as an SA should have some kind of alarm system, so you shouldn’t be surprised) or even know any problems.
The 5-10 second freeze you see is absolutely normal, especially If the server you are rebuilding is any type of database server, the write and read are higher than usual by default (that is, the SQL server for the database) all day; a property management company I once consulted has a program that can access their tenant records all day to view and write new information, and it is always used a lot.) It will be more noticeable.
The other thing I suggest is to get any RAID utility (GUI version) that comes with the controller and install it on the operating system so you can monitor the rebuild There is no need to load it into the controller BIOS.
Nowadays, a very small and almost non-existent problem is NFS and iSCSI. I know you are using NFS, and in the past it was the case of iSCSI in virtualization There will be better overall performance, but with recent improvements to hypervisors and hard drives and controllers, the performance of NFS is almost the same as iSCSI, so it sounds like you have a very good SAN.
I am happy to answer anything you need to know, so please feel free to comment.