PDA

View Full Version : Need advice on new RAID software



Steve R
02-01-2001, 12:06 PM
I've been using an external RAID 1 array on my 9600/350 as primary backup of client data for a number of years, using Retrospect Remote and FWB's Hard Disk Toolkit for the RAID formatting. The canisters are removed at night and placed in a fireproof safe.

I've just upgraded the drives to Seagate Barracuda 18XLs (ST318416N) and found that FWB do not currently support this mach. They say they are dependent on the drive manufacturers to actually give them a drive so that they can characterize it (?).

I've looked at the Charismac RAID product, but it does not have a 'mirror existing drive' option and destroys all data when initialising an array. (It also has a user interface that makes me ill to look at.) This means, presumably, that if a drive array breaks, the entire contents of the 'good' mechanism must be backed up and restored, a major hassle when working with big drives.

Possible contenders are SoftRAID and Optima's DiskArray software. Does anyone have any positive or negative thoughts about either of these, or want to suggest any others? Optima's software claims to create self-repairing RAID 1 arrays, which is nice if it works.

Steve

jorge
02-01-2001, 01:00 PM
We recommend, use and sell Softraid. My company depends on it.

j

kaye
02-01-2001, 01:27 PM
I use SoftRAID for single drives and RAID 0. I love it. I know nothing about RAID 1 but doesn't SoftRAID do the same as Optima in self-repairing RAID 1 arrays as discussed here http://www.softraid.com/core.lasso and probably elsewhere on the site? I have no experience with Optima, but I totally agree with you about CharisMac RAID. It is also slow in a striped RAID compared to SoftRAID. k

Dogstarman
02-01-2001, 06:23 PM
SoftRAID functions similarly to Optima as far as self-repairing. I use it to mirror 2 18GB Barracudas. Works great, in the background to repair and upgrade the mirror. Hey, some things HAVE to be in the background. And I am fairly certain you can mirror an existing volume. At least it looks that way to me on the site. I have had a good deal of success with it, especially compared to FWB or CharisMac.

magician
02-01-2001, 07:37 PM
the really nice thing about SoftRAID is that you can continue to use a RAID 1 volume while it rebuilds in the background. SoftRAID is the standard for professionals, which is why Apple itself bundles it with servers.

until recently, I had never heard of Optima. All our machines here run SoftRAID, and have been, for years. We love it, and we have a close relationship with SoftRAID LLC, which is why we are the lone internet vendor authorized to sell it.

http://macgurus.com/infopop/emoticons/icon_biggrin.gif

MacMikester
02-03-2001, 02:54 PM
Seems like a good spot to insert a recent personal finding.....
I boot 2x 10k Cheetahs in RAID 0 on an ATTO UL3D in a PM9600/350 using SoftRaid, no prob--great performance.

But I was also using SoftRaid for a couple of individual disks (the stock 4G IBM in the PM9600 and a stock 2G Seagate originally in a PM8500). On a whim, I switched the two individual disks to the ATTO driver, installed and formatted with ExpressTools Pro. There is a HUGE speed increase:

SoftRaid 2.2.2/4G IBM UltarStar/ ExpressTools benchmark-8MB max transfer, disk cache on:
Peak Read 5.59 Sustained Read 5.54 Peak Write 5.72 Sustained Write 5.59

ATTO Express/4G IBM UltraStar/ ExpressTools benchmark-8MB max transfer, disk cache on:
Peak Read 8.11 Sustained Read 8.06 Peak Write 8.27 Sustained Write 8.07

This drive terminates the internal SCSI-2 (10 MB/s) stock bus. I note a similar speed increase with the old Seagate on this same bus (the Hawk is a SCSI-1 device with an extended SCSI-2 command set so it theoretically maxes at 5 MB/s--results went from around 3MB/s to around 4.2 MB/s). I have experimented heavily with SoftRaid driver options and no manner of settings ever changed benchmark performance very far ( a couple of tenths of a MB/s) from the above results.

That's my story and I'm stickin' to it.

[This message has been edited by MacMikester (edited 03 February 2001).]

magician
02-04-2001, 04:00 AM
what happens when you run the Cheetahs with ExpressTools or ExpressRAID?

didn't Kaye post comprehensive results from some tests like this that he ran?

MacMikester
02-04-2001, 07:34 AM
Hey Mag,

I don't have ExpressRAID so I couldn't try it out on my array and I don't want to take my array down to try the individual Express driver on the Cheets. I was very very surprised by the performance results on the individual disks. Makes me wonder if ExpressTools benchmark is 'optimized' for the ATTO driver. In retrospect, I should compare with an alternate benchmark to be sure, but I have spent way way too much time over the last three weeks servicing the needs of my mac instead of it servicing me. Now that I'm on the upside of that, I think I'll stay here for a while and enjoy.

Kaye has indeed tested so much, I can't remember if he has done a similar comparison between just these two drivers or not.

kaye
02-04-2001, 12:27 PM
I have tried both for single drives. The difference between SoftRAID and ExpressPro Tools when initializing a drive/volume is that EPT, as part of the setup, allows you to set Mode Page Parameters optimization whereas in SoftRAID you must select the Mode Page Parameters after the drive/volume is initialized and you can only change a few parameters, EPT changes much more. Additionally, those two drives, being older, their parameters were probably an old generation and not nearly optimized and maybe not even the default. So SoftRAID as the driver was working with old parameters, EPT optimized the parameters. You could have gotten better performance from SoftRAID by leaving its driver on the drive and then, using EPT, just select the drive and use the option to optimize Mode Page Parameters.

If you recall, that was discussed before. EPT does not care what the driver is (for every driver I have tried), it will optimize Mode Page Parameters even if SoftRAID or HDT or Anubis is the driver. Personally, I found SoftRAID faster, after manually changing Mode Page Parameters myself. EPT does a good job of optimization, but I just prefer to do it myself for a particular drive. k

Louie
02-04-2001, 02:06 PM
I think Kaye and I agreed that ExpressRAID was an absolute and total pain in the ass to figure out how to install it properly.

MacMikester
02-05-2001, 10:42 PM
Hey Kaye,

The results I found below were with Mode Page settings set to the factory defaults. Like you, I don't prefer to use template style changes but would rather have access to the mode pages through a format like HDT Config, which shows you all changeable values. I have no idea if EPT changes parameters that are not readily accessible by looking at the limited listings in EPT or SoftRaid, but the listed parameters were the same for this comparison. After extensive experimenting with SoftRaid parameter changes, I never saw more than a couple of tenths of a MB/s difference and never came close to the performance numbers I am seeing with the Express Pro driver.

kaye
02-06-2001, 12:50 PM
MM,

I must have misread your post. I thought you said that you first tried SoftRAID, with the assumption being that MPP were who knows what, not necessarily default, and then initialized with EPT with, now I know at default settings, a MPP change. So, unless you changed the MPP to default before running the SoftRAID test, the playing field is not level. Does this make sense?

If you did set MPP to default before making the driver comparison, then I have no idea. Every comparison between the ATTO EPT driver and SoftRAID for single drives, and the ATTO RAID and SoftRAID for dual Miles2 and four X15s (I only tested ATTO RAID in this config that I recall), SoftRAID was faster. There must be a YMMV issue somewhere in this. k

MacMikester
02-07-2001, 12:53 AM
Hey Kaye,

What you are saying/asking makes perfect sense. I did set mode page parameters to default in the beginning when I set up my RAID (Cheetahs and ATTO UL3D courtesy of the Gurus) and switched all my drivers over to SoftRaid. Then I filled up several yellow legal pad sheets recording ATTO benchmarks for SoftRaid tweaks. Since I was running OS 9 at the time, I think I ran HDT Config (v. 3.0.2) off the HDT startup CD to set parameters to default. I did this because of the strange numbers for cache segments and buffer settings that SoftRaid was initially showing, as we have discussed a long time ago. Anyway, I did this for my RAID and my IBM Ultrastar. I only recently added the 2G Hawk to my 9600 and I used EPT to set defaults on that for both drivers, and to set defaults on the IBM for the EPT driver.

Unless I did something drastic with SoftRaid, like disable write caching, I just never the numbers change very much. I thought my drives were pretty much optimized until I switched to the ATTO driver and saw this increase. Maybe the ATTO driver works better with older firmware??

kaye
02-07-2001, 02:11 PM
MM,

I'm at a loss. One of those YMMV or perhaps older firmware as you say. Except for my PTP which is absent at the moment, all of my older drives are in NuBus machines where SoftRAID will not work and I never tried the late ATTO drivers. Not even sure they would work in NuBus. But the ATTO Mode Page Parameters choices do work in NuBus, well, at least a couple of revisions ago. k

Louie
02-07-2001, 03:33 PM
MM:

The Hawk is S-L-O-W! You may have found a supercharger. Try it again, please. How the heck did you figure out how to install ExpressRAID?

kaye
02-07-2001, 06:18 PM
Louie,

If I understand MM correctly, he does not have ExpressRAID. He used ExpressPro Tools to initialize those two non-RAID drives. So those two drives have the ATTO driver on them. k

Louie
02-07-2001, 08:33 PM
Do each of those programs have different drivers?

MacMikester
02-08-2001, 01:07 AM
Hey Louie,

Yes, you have to purchase ExpressRAID. EPT can be downloaded but my version came with the ATTO UL3D card.

Steve R
02-08-2001, 01:24 PM
This discussion has meandered off in interesting directions, but I thought it might be helpful to post a follow-up to my original request for RAID software recommendations.

To be fair to FWB, it now seems that some of the original problems I was having with Hard Disk Toolkit may have been due to termination problems. However, the time required for HDT to mirror an 18Gb drive, during which time the Mac is paralysed, was unacceptable compared with that for a 4Gb drive.

I rejected Charismac RAID because it requires two *empty* and initialised drives to form a mirror. This means that if a mirror fails, all the data must be moved off the good drive before the mirror can be rebuilt, big hassle. As I'm using a RAID 1 mirror as my primary backup, I have no way of doing this even if I was prepared to.

That leaves SoftRAID. Builds mirrors in the background, excellent technical support, excellent product, reasonable price. Recommended. I'm in RAID Heaven.

Steve R

magician
02-08-2001, 02:10 PM
well, there you go!

nice summation, Steve. Thanks for the closure.

http://macgurus.com/infopop/emoticons/icon_biggrin.gif

Mactacular
02-10-2001, 11:31 PM
SoftRAID 2.2.2 gets the job done well.

Filter413@hotmail.com
"PowerMac SkunkWorks" http://members.tripod.com/mac_man_6/index.htm