Hi there. First time using freenas, overall really impressive product. My boss and I decided to run with this in lieu of of a pre-built solution (such as buffalo). Our goal was to have a massive storage solution for both file storage (<1 TB ) and to house our backups from other servers with 20 days of rotation (currently using tape archives (<1.5TB)). Once the decision was made we ordered 2 machines with 32 GB of ECC memory an LSI card (flashed to IT mode), and 8x 4 TB hard drives each. We were expecting to have about +/- 26 TB of storage. However after reading up on using RaidZ (comparable to Raid5) I realized it is now frowned upon because of the probability of an unsuccessful Raid rebuild. My question is since we have two devices (the second one is a backup with ZFS replication being pushed to it) would it be advisable to use RaidZ anyway since we have an offsite backup in case the Raid rebuild failed?
-
Important Announcement for the TrueNAS Community.
The TrueNAS Community has now been moved. This forum has become READ-ONLY for historical purposes. Please feel free to join us on the new TrueNAS Community Forums
RAIDZ best configuration.
- Thread starter NMSU_Matt
- Start date
- Status
Well, as long as you are relying on the fact that you will have to do a full restore and you won't have one disk in each server fail at the same time, then sure.
IMO I can't find much use for a RAIDZ1 because just the time lost with rebuilding makes the cost of the extra disk a major bargain.
IMO I can't find much use for a RAIDZ1 because just the time lost with rebuilding makes the cost of the extra disk a major bargain.
- Joined
- May 28, 2011
- Messages
- 10,996
If you have a critical need to retain/protect your data I strongly urge you to use RAIDZ2 or better yet, RAIDZ3. These are nothing like RAID5 and have a huge amount of redundancy built in. I don't understand what yo mean buy an unsuccessful RAID rebuild, are you talking about resilvering when you replace a failed drive? We do have some RAID gods working here like @Cyberjock and @Solarisguy.
Thanks for the input, :) about how much longer is the time difference between between RAIDZ1 and RAIDZ2 when rebuilding? Another question if we decide to go with RaidZ2 is the performance going to decline noticeably if we set the layout pictured in the attached file? At night we plan to run out exchange backup to freenas, the users documents we plan to house on here are mostly just word and excel documents so we are not to concerned about performance in that area.
Attachments
After reading up on Raid and RaidZ I came across some articles describing how a Raid5/Raidz rebuild can become corrupt due to bad sectors on the remaining drives (same one posted above). And since 2009 its been unadvised to use Raid5 (show how long since I had to setup Raid, pretty rusty.) I'm not sure what resilvering is? Thanks for any input, I appreciate it.
Neil Whitworth
Dabbler
- Joined
- Nov 14, 2013
- Messages
- 30
How much is your data worth?
How long will it take to recover from backups? (and how much will that downtime cost?)
When a HDD fails, with RaidZ you have just lost all the redundancy in your pool. A second failue or unrecoverable read error and your pool could be gone for good.
This thread covers the chances of HDD/pool failure. Only you can decide if the cost of the extra HDD/lost storage is worthwhile for your set up.
How long will it take to recover from backups? (and how much will that downtime cost?)
When a HDD fails, with RaidZ you have just lost all the redundancy in your pool. A second failue or unrecoverable read error and your pool could be gone for good.
This thread covers the chances of HDD/pool failure. Only you can decide if the cost of the extra HDD/lost storage is worthwhile for your set up.
solarisguy
Guru
- Joined
- Apr 4, 2014
- Messages
- 1,125
As it was pointed out, you had asked a business question in a technical forum ;)
So business angle first. It is all about insurance and risk management, thus somebody from outside is not going to be able to help you a lot. There are IT departments that keep reliability data for their own environments. But theirs is not your environment... You likely have your different hardware with different temperature inside your cabinets, have a different air flow and different power quality from your UPS. Results from somebody else can tell you whether RAID-Z2 is better than RAID-Z1, but you already know that it is...
Unless you are working with a NASA or NSA budget, your bosses are not going to have a risk management study
, they are likely just looking for a technical confirmation of their business decision. One of the simpler technical outlooks, I can think of, is to consider the worst scenario out of the likely ones. Very often it would be a human error that wipes out your primary server. Assuming write speed of your ZFS pool to be 150MiB/s and 23.65Tib (=26TB) of data to be transferred, you get 46 hours for the task. RAID-Z level seemingly was not important, but the disk and network speed were. During those two days, you have only one copy of the data, so RAID-Z level does play a role... Better disks offer better insurance too... (see the thread mentioned in the post just before mine). Of course you should reverse the situation and realize that the same recommendations apply to your primary server.
If you have your systems, perform testing with your data, hardware and patterns of usage. And see what your sustained speeds are for the worst case scenarios (you can think of). For example, try users writing and reading while resilvering. Do these tests both for RAID-Z2 and RAID-Z3 with a varying number of disks. Try to have a non-trivial disk space usage (that is at least 1TiB) in your pool.
Have fun!
So business angle first. It is all about insurance and risk management, thus somebody from outside is not going to be able to help you a lot. There are IT departments that keep reliability data for their own environments. But theirs is not your environment... You likely have your different hardware with different temperature inside your cabinets, have a different air flow and different power quality from your UPS. Results from somebody else can tell you whether RAID-Z2 is better than RAID-Z1, but you already know that it is...
Unless you are working with a NASA or NSA budget, your bosses are not going to have a risk management study
, they are likely just looking for a technical confirmation of their business decision. One of the simpler technical outlooks, I can think of, is to consider the worst scenario out of the likely ones. Very often it would be a human error that wipes out your primary server. Assuming write speed of your ZFS pool to be 150MiB/s and 23.65Tib (=26TB) of data to be transferred, you get 46 hours for the task. RAID-Z level seemingly was not important, but the disk and network speed were. During those two days, you have only one copy of the data, so RAID-Z level does play a role... Better disks offer better insurance too... (see the thread mentioned in the post just before mine). Of course you should reverse the situation and realize that the same recommendations apply to your primary server.If you have your systems, perform testing with your data, hardware and patterns of usage. And see what your sustained speeds are for the worst case scenarios (you can think of). For example, try users writing and reading while resilvering. Do these tests both for RAID-Z2 and RAID-Z3 with a varying number of disks. Try to have a non-trivial disk space usage (that is at least 1TiB) in your pool.
Have fun!
- Joined
- May 28, 2011
- Messages
- 10,996
Resilvering is the name of the method that ZFS uses to rebuild a replaced drive. When you have a failed drive with say a RAIDZ3 system, you replace a drive and over a period of hours (depends on drive size and how much data is on your drives) the new drive will be rebuilt and still all your data is available for use. In a RAIDZ3 system even if you have another drive failure the new disk will continue to rebuild. You should read up on ZFS to understand what it is. It's not your fathers RAID5.
- Status
