Getting ready to install FMS on a Windows Server 2012 machine that has 6-146GB Drives. Will Eventually have about 40GB Data with about 200 GB more in External Containers? What should I consider in deciding which Array configuration to use with these 6 drives.
You can't go wrong with either one. There are pros and cons for each one. We have a general FileMaker Server hardware knowledge base article you can refer to for more information.
If you have 4 x 1 TB drives, you have 4 TB of disk space. With RAID 5, you'll get 2.7 TB's of useable storage. With RAID 1+0, you'll get 2 TB's. So RAID 5 will return more useable storage.
Studies on big SQL engine databases have shown that they perform better on RAID 1+0 than RAID 5. RAID 5 reads very fast, but writes very slow having to do with how the parity stripe is written. Databases do a lot of read and writing, so the theory is that RAID 1+0 works better for database environments. I've never seen such a study in FileMaker, but believe the theory would be true for FileMaker too. And since disk space is not that expensive anymore, the advantages of RAID 5 having more usable storage is not so great as it used to be.
And some people will argue that there is a very rare possibility of silent RAID 5 corruption that is not possible on RAID 1+0. I think it is very highly unlikely as RAID 5 has really stood the test of time as a good RAID system, but it is just one more reason to have a database use RAID 1+0.
HDD maker means 146G as 146,000,000,000 bytes.
It is about 136GB in OS, so 1+0 makes usable total space less than 400GB.
It depends on your speed of growing db size.
1 of 1 people found this helpful
If you can live with the amount of storage a RAID 1+0 gives you, it really is the best choice for performance
The problem with RAID 5 is that a FileMaker server is most likely to get bogged down while writing data, and that is also where RAID 5 is at its slowest. Reads are usually much less of a problem, since the cache, if sized appropriately, does a very good job of minimizing how much data must be read in. Since you need to write out any changes as soon as possible to disk, regardless how good your caching is, write access becomes you main bottleneck.
Of course one exception might be if your system is infrequently modifying any data, in which case RAID 5 would be your best choice.
FYI, an interesting choice that can also come up is how big a block size to use. I've usually opted for the smaller size (32K if I recall), but I don't have any hard data to back up whether that's the best choice.
1 of 1 people found this helpful
I would highly advise against RAID 5. The silent corruption issue is a real one, and the worst part is you'll never know if you've been hit by it until you experience observable data loss, and then it's too late. I don't know anyone that uses RAID 5 any longer because of this reason.
Raid 1+0 is fast, and definitely a better choice over RAID 5, but only allows for failure of one drive in the array and requires double the disks.
I personally would recommend RAID 6 at this point in time, which is what we use. Find a RAID controller that does it in hardware and is optimized for RAID 6. The advantage of 6 is that it allows for 2 drives to fail, and it doesn't require as many drives as it works off a parity system. While it does impose a write penalty, using fast drives and a controller that is meant to do RAID 6 can greatly mitigate this. We use RAID 6 on all our VMs here at this point, with excellent results.