[JB72]

I'm going to be in the market for an IDE RAID card in the near future (a month or two) for my MDD, using Mac OS X only. I'm hoping to get as many opinions as I can. Some things that I am looking for include:

- The ability to sleep w/o hassles.
- Reliability.
- Speed.
- Bootable.
- Quality drivers and support.
- other stuff you guys think to mention.

The options:

1) Sonnet Tempo RAID133



2) Siig UltraATA 133/100 RAID



3) Acard AEC-6880MA6




Final thought: Why isn't there a combo RAID/FW/USB card like this available for Macs? Will there be one in the future? That would be worth waiting for.



TIA

[Sven G]

... Isn't the Sonnet Tempo Trio similar to what you are looking for...? (However, maybe it's not RAID...)

[JB72]

Originally posted by Sven G:
... Isn't the Sonnet Tempo Trio similar to what you are looking for...? (However, maybe it's not RAID...)

Yeh I saw that. But I really need the RAID . Thanks anyway though.

[tr]

the siig and the acard use the same chipset. i have the siig RAID in my G4 450. i actually updated the firmware on the card using the latest acard firmware, after reading that some people have had success doing it. there is also an upgrade to fix wake from sleep errors. sleep works fine on my G4, but i usually don't put my computer to sleep.

i works pretty well, and it's fast. i mainly use it for video storage/scratch. setup was pretty simple, but a bit tricky. for the siig, i think you have to hook up the drives first (with the card in non-RAID), format both drives, then shutdown, flip switch to RAID, reboot, and format again (now, as one giant disk, if you are striping).

i have been noticed one weird thing with my card, though. every once in a while, usually after i've had the computer on for 24+ hours, after i shutdown, and then reboot after a few hours, OS X can't see the RAID, and i get the message "you have inserted a disk that X can't read". but, if i shutdown again, pull the power cord from the computer, then plug it back in and reboot, it sees it fine. it's strange, i have no idea what's going on there.

so other than that, the siig RAID has worked great. plus, it's cheaper than the other RAID cards out there.

tr

[JB72]

tr,

Thanks for all the detailed, honest feedback. That's exactly what I was looking for.

The siig and the acard use the same chipset.

I think the Sonnet card uses the "Promise" chip, I think. I wonder is that's the same one too.

after reading that some people have had success doing it. there is also an upgrade to fix wake from sleep errors. sleep works fine on my G4, but i usually don't put my computer to sleep.

Sleep is a big one with me. I like to keep my workstation on 24/7, but I can't stand the noise when I try to sleep!

every once in a while, usually after i've had the computer on for 24+ hours, after i shutdown, and then reboot after a few hours, OS X can't see the RAID, and i get the message "you have inserted a disk that X can't read". but, if i shutdown again, pull the power cord from the computer, then plug it back in and reboot, it sees it fine. it's strange, i have no idea what's going on there.

Yikes! That scares me. Have you checked maybe XLR8yourMac for a possible solution?

Thanks again for your help tr.

I need alll the input I can get; the good, the bad, and the ugly. Anyone else using an IDE RAID card with OS X?

? - ?

[JB72]

FYI, from XLR8YourMac:

the Mac SIIG and Sonnet Tempo ATA/66 RAID and ATA/133 RAID cards are Acard based, but not the Tempo ATA/100 and Tempo ATA/133 non-RAID cards (they are Promise chip based.) If in doubt, run Apple System Profiler to check the card controller chip/ID.

There is also some info here, for those in the same market as me.

[JB72]

Alright baby, now we're getting hardcore. That's more my style...



4-CH High Scalability ATA133 RAID Adapter

"AEC-6885MS fully supports 64-bit PCI specifications, including Bus Master to increase transfer capability, and also supports Scatter/Gather to enhance multitasking effects. The highly effective AEC-6885MS is fit for Macintosh workstations and servers. It can meet your need in storage and data protection."

4 channels, 64 bit/33Mhz. Now we're talkin'.

[JB72]

Well I geuss I didn't get as many replies as I had hoped for on this one. Looks like I will probably hold out a little longer to find out when we'll see this new 4 channel card for Mac. I would imagine that 64 bit might have certain advantages for working with large video files.

[tooki]

Before recommending a product to you, I'd sooner like to hear what you plan to do with it. For DV video, RAID is basically not necessary. (Remember that RAID 1 -- striping -- significantly reduces your data security!!!) So unless you absolutely, positively need all the speed of the RAID, don't do a RAID 1. Now, RAID 0 (mirroring) slows your writes, speeds your reads, and does protect you against drive failure. But remember that ATA RAID cards don't seem to have the ability to rebuild the array after a drive fails -- you will still have your data accessible on the remaining drive, but you have to reformat it to recreate an array.

RAID 5 is a great kind of RAID (distributed parity) that offers good speed and great redundancy, but regardless of whether ATA or SCSI disks are used, these arrays connect using SCSI or Fibre Channel (and thus tend to be rather costly!!).

Another thing for you to keep in mind is power requirements, if you plan to use the internal IDE and the card at the same time. Remember that even if there are lots of drive bays, there may not be enough power to power that many drives!

tooki

[JB72]

Originally posted by tooki:
Before recommending a product to you, I'd sooner like to hear what you plan to do with it. For DV video, RAID is basically not necessary. (Remember that RAID 1 -- striping -- significantly reduces your data security!!!) So unless you absolutely, positively need all the speed of the RAID, don't do a RAID 1. Now, RAID 0 (mirroring) slows your writes, speeds your reads, and does protect you against drive failure. But remember that ATA RAID cards don't seem to have the ability to rebuild the array after a drive fails -- you will still have your data accessible on the remaining drive, but you have to reformat it to recreate an array.

RAID 5 is a great kind of RAID (distributed parity) that offers good speed and great redundancy, but regardless of whether ATA or SCSI disks are used, these arrays connect using SCSI or Fibre Channel (and thus tend to be rather costly!!).

Another thing for you to keep in mind is power requirements, if you plan to use the internal IDE and the card at the same time. Remember that even if there are lots of drive bays, there may not be enough power to power that many drives!

Ty Tooks.

I'll eventually want to be able to do some SD if possible with IDE. Also wouldn't mind blazing with large renders in DVD (and DV50 when/if that happens.) So I feel the need for speed. I thought about RAID 5 but from what I hear it doesn't favor digitizing the way the arrays are setup. Otherwise it is dang coolio.

I didn't think about the power issue too much. That's something to make me sad .

[gatorparrots]

The 2-channel RAID cards are fine if you only hang one drive per channel (say two 200GB drives), but remember the inherent inefficiencies of IDE in a master/slave relationship: you generally take a performance hit of 10-20% in that scenario. For maximum speed, you need independent channels. This is why you would be better off with either two separate cards or the 4-channel card: you could support four drives on four separate channels without the IDE master/slave performance tax. This way would be much cheaper and would compare favorably in speed to an IDE RAID array. Also, the smaller your partitions, the more disk space you regain from increased block size issues during formatting (the bigger the single partition, the more space is lost in formatting).

[JB72]

Originally posted by gatorparrots:
The 2-channel RAID cards are fine if you only hang one drive per channel (say two 200GB drives), but remember the inherent inefficiencies of IDE in a master/slave relationship: you generally take a performance hit of 10-20% in that scenario. For maximum speed, you need independent channels. This is why you would be better off with either two separate cards or the 4-channel card: you could support four drives on four separate channels without the IDE master/slave performance tax. This way would be much cheaper and would compare favorably in speed to an IDE RAID array. Also, the smaller your partitions, the more disk space you regain from increased block size issues during formatting (the bigger the single partition, the more space is lost in formatting).

Big thanks gatorparrots. I'll definately be doing the 4 indepent channel thing, with four drives. I'll just have to wait until this new card is out. I suppose there are other improvements I can focus on until then.

[tooki]

Originally posted by gatorparrots:
Also, the smaller your partitions, the more disk space you regain from increased block size issues during formatting (the bigger the single partition, the more space is lost in formatting).

Wrong. Not with HFS+ (which you need anyway for drives bigger than 2GB). HFS+ limits the block size to 4KB no matter how big the drive is. (Some third-party formatting utilites will let you use other block sizes, but there's no compelling reason to do so.)

Drive speed won't be your limiting factor when encoding MPEG2 for DVD, nor for rendering effects. The CPU[s] will be the bottleneck for most such tasks, and as such, faster drive throughput won't do much.

I agree that independent drive channels are the best for IDE, although I don't know where gatorparrots comes up with those numbers -- there will be NO speed loss if only one drive on a given bus is being used at a time, and a significant speed loss if they are both being used heavily at once.

tooki

[Cincinnatus]

Just to clear up a few points:

RAID 0 - striping
RAID 1 - mirroring

( http://snow.prohosting.com/scripts1/SectionDocs/raid/ )

Personally, I'm using the Acard 6880M, with 2xWD 40 GB drives - striped. I haven't had any issues *yet*; no hardware failures or other strange corruption of the file system (I run journaled, just to be a bit safer - though I may not be buying myself anything here). I have this RAID as my boot volume - and occassionaly I notice a stuttering in iTunes - which is odd, as all of my mp3s are on another server. Other than that, everything has been running smoothly for 6+ months

/Cincinnatus

[JB72]

Drive speed won't be your limiting factor when encoding MPEG2 for DVD, nor for rendering effects. The CPU[s] will be the bottleneck for most such tasks, and as such, faster drive throughput won't do much.

True. But I'll be squeaking buy with IDE RAID for SD as it is (I hope the new Apple 10 bit codec will allow this, as does the Igniter's codec currently.) Plus I've never heard an editor say, "My drives are too fast." FCP is a very flexible program, and I always find myself wanting to do more (w/o going back to $C$I.) Plus I want to arm myself as best I possibly can against the dreaded dropped frame, who visits those with their gaurds down first.

I agree that independent drive channels are the best for IDE, although I don't know where gatorparrots comes up with those numbers -- there will be NO speed loss if only one drive on a given bus is being used at a time, and a significant speed loss if they are both being used heavily at once.

How do channels figure into this?

With FCP, one should ideally be using multiple drives and/or partitions simultaneously for best performance. I certainly hit more than one drive at once.


Thanks for the advice all. I do need to be giving this all some thought.

[JB72]

A new potential ally:



Saddled next to the 4 channel RAID card in my MDD*, it's like a 64 bit storage orgy.

The only potential downside is that I think it's only single channel. I love channels. Might have to wait for the next gen FW card. Or maybe not.


*up here {points to head} I'm already using that sweet four channel monster.

[gatorparrots]

Originally posted by tooki:
I agree that independent drive channels are the best for IDE, although I don't know where gatorparrots comes up with those numbers -- there will be NO speed loss if only one drive on a given bus is being used at a time, and a significant speed loss if they are both being used heavily at once.

I think we agree, but are pointing out two separate issues with a master/slave relationship under IDE. I was specifically pointing out the inherent speed loss of 10-20% associated with a master/slave relationship on a single IDE channel. A quick scan of the net produced the two quotes that follow:

Let's address this two fold: First of all, running RAID on IDE drives is
almost never a good idea. You *might* consider RAID 0 (striping only),
but anything else on a single controller is most certainly a terrible
idea. The idea behind RAID is to access disks in parallel. If the system
must swap between disks on the same cable, you lose all the advantages,
plus maybe 10-20% overhead (depending n your system). If you REALLY want
to do this, you most certainly need multiple PCI controllers, with only
one disk per cable. However, I've heard of problems trying to run
multiple drives that are different speeds (because of screwy latencies
which make the point of RAID useless) and different speed/throughput
controllers (though the latter is less significant)

Additionally,

You can expect significant performance penalties when you mix different kinds of IDE devices on the same IDE channel. IDE slows down when you combine fast and slow IDE/ATA devices on the same channel. For example, combining a fast ATA-2 PIO Mode 4 hard disk on the same channel with a slower PIO Mode 2 IDE CD-ROM drive effectively cripples the performance of the hard disk by forcing it to continually wait for I/O operations when the CD-ROM is active. Under these circumstances, the fleet effect takes place: The fleet goes only as fast as the slowest ship. In these circumstances, a better solution is to place your slower IDE peripherals (such as CD-ROMs and tape drives) on IDE's slower second channel, keeping them off the channel that contains your fast hard disks. However, if you use more than two IDE hard disks on your system, you won't have a choice; you will be forced into the fleet effect situation. With SCSI, these issues are non-existent.

As if these issues aren't bad enough, you can find a significant number of IDE/ATA controllers that use only a single command queue shared between the primary and secondary IDE channels. This situation can result in major performance delays when devices on both channels try to talk to the system simultaneously. The problem is further aggravated when the two channels are fully populated with IDE devices.

Etc.

So the basic point stands (fine details aside): it is best in speed-critical situations to run a single drive per IDE channel. A 4-channel card with four separate drives card is preferred over four drives on a 2-channel card. Perhaps better still is two 2-channel cards, each with two drives.

[JB72]

Originally posted by JB72:

The only potential downside is that I think it's only single channel. I love channels. Might have to wait for the next gen FW card. Or maybe not.

Good news, I found a 4 channel FW800 card in the works. Bad news, I haven't heard anything re: the availability of the 4 channel Acard board.

[CIA]

I have 2 PCI IDE cards, one is an ATA/133 RAID card, and the other is just a normal ATA/133 card. One is a SIIG and the other I think is a Sonnet (Acard based?) Either way I know one of them is based on Acard chips and the other is a promise chip.
I have 4 drives running total (2 per card, or, 1 per available BUS)
I've been running a striped RAID as my System drive using 2 WD Special Edition 120GB's. The performance difference is noticable on my Quicksilver Dual800. It's run like a champ 24/7 for 9 months. Yes, I frequently back up.
The firmware update a few months back fixed the sleep issue. There is a new firmware update out now for large drive support and some other things, anyone use it yet?

[macgyvr64]

What about non-RAID cards? I got a 180 GB drive and got stuck with the ATA/100 hardware limitation of my Dual 500. Are there any non-RAID ATA/133 cards that work under OS X flawlessly?

[JB72]

Originally posted by macgyvr64:
Are there any non-RAID ATA/133 cards that work under OS X flawlessly?

I can't speak for their performance or reliability, but Acard, Siig, and Sonnet Make standard ATA cards.

[solaris]

For speed and redundancy (and a hole in the wallet) you could go RAID 0+1 (stripe then mirror, aka: mirror-stripe).

Buy 2 ata raid cards and 4 drives. Create a 2 drive stripe on each card, then use the OS X disk utility to mirror the two stripes together. In theory you could do this with one card that supports 4 drives, but you still have a single point of failure - the card).

[JB72]

What's been people's experience with PCI cards and deep sleep? I just bought a FireWire Depot FW800 PCI card and the bloody thing prevents my Mac from deep sleeping now. WTF?