Creating raid arrays with an Areca card is very simple.
- Go to the Areca BIOS and define the array and format it.
- Stripe defined array sets if you want a multi-level array in the Areca BIOS.
- Go to disk management and format the array with GPT as a single volume for each defined array set.
- Do not create smaller volumes. That is similar to partitioning a single disk, something that is not done. You created the array to get speed, why destroy that by making smaller volumes and introducing further overhead?
Well harm I think they should get you to write their manuals
So ive been googling volumes on an array and there are mixed opinions. The issue for me is a whole load of pain in broken links. I've been using the same drive structure for years. I presume creating 2 big partitions would give less of an overhead than 4 small ones (I can cope with dealing with broken links from one of my big three drives).?
You can define multiple array sets in the BIOS, for instance a 4 disk raid5 and a second 4 disk raid5 instead of one 8 disk raid5 or 50. That would achieve your purpose nicely in that you can used different drive letters for each array. There is one downside however and that is speed. If we take the transfer speed of a single disk as T and N as the number of disks in the array, then a parity array (3 or 5) will have a speed of around:
(N - 1) * T * 0.85 which translates to 2.55 T in the case of a 4 disk array and 5.95 T in the case of a 8 disk array.
The alternative is suffering the pain once by pointing PR to the new location of your files and saving this project.
Not just premiere: I'm a jack of all trades. Libraries, shortcuts, backups, catalogues, photoshop actions, files with links they'd all be broken. I've only got a 5 disk array.
EDIT: I just tried removing the partition and copying a 1.2gb file onto it a terracopy reported 154mb/s from an ssd. Copying to a single wd black (3/4 full) gave me 79mb/s
Is this an indicator that theres something ammis? I set the array up with the default stripe size (64k) and with >2tb support using 64bit lba
And when I test the write with no volumes on the array in HDtune I get an AVG of 434mb/s!, an access time of 0.025ms (2.1% CPU)
That corresponds nicely with my indicator:
(N - 1 ) * T * 0.85 = (5 - 1) * 129 * 0.85 = 438 MB/s.
But then why is the write so slow in reality? 434 vs 154? I read about partition alignment somewhere, could this be an issue, or so benchamrks have nothing to do with real life sequentional writes?
Thanks harm, I appreciate you expertise, I'd be flailing aroudn in the dark otherwise. Maybe I should try raid 3?
That could easily be caused by indexing, antivirus running, disk defragger watching, services running while not needed, background processes like mDNSResponder, jusched, and the like. That is why is always good to 'tune' your system, to get rid of the standard overhead.
For the fastest writes, under disk device properties you need to enable write-caching (should be done by default) and turn off windows write-cache buffer flushing on the device (this is not the default). Of course you should probably either have either a UPS for your system or Areca BBU (battery back-up unit) or both in order to push the envelope of speed with this setting.
UPS tick!, BBU no tick, Yes Im looking into it.
Well I just transferred the contents of 2 of my wd blacks onto it at the same time and at times I was getting write speeds of 120mb/s and 140mb/s at the same time, so maybe the ssd was the bottleneck. About to start using it for real so I'll get a feeling for any speed boost in the next few days.