DSM-G600, DNS-3xx and NSA-220 Hack Forum

bscott · 2008-04-29 15:55:32

Hiya - just wondering about the risk of using mismatched drives in RAID 1...

I have a 320G Seagate and am about to buy a 320G Samsung - same size, RPM, seek time, cache, pretty much identical specs in every way, but they're not actually the same make or model... so am I a fool to consider replacing the aging (but matched pair!) 120gig drives that are currently in my DNS-323?

I would think that any slight difference in write performance would not affect the server's function, given that even using gigabit Ethernet you're only pulling about a quarter of the theoretical throughput of modern drives. I also know that many RAID setups used to demand identical drives, and that suggests that the DNS-323's ability to at least try using mismatched drives is there for a reason.

Anyone have real-world experience with this sort of thing? I know I could buy another Seagate but it's not really about upgrading my server in the first place - I just have a chance to get this Samsung drive virtually for free, and so I'm wondering if I should try putting it into the DLink box (starting to wonder when the 4-year-old drives in there now will give up the ghost...) or just play it safe and leave it in a simple USB enclosure.

fordem · 2008-04-29 18:19:29

As far as I'm aware there is no great risk in using "mismatched" drives in a RAID array - at least as you describe them.

The potential problems are a loss of useable space, and performance (throughput) degradation.

The useable space in a RAID array is dictated by the size of the smallest drive, the actual calculation will vary according to the RAID type, but in your case, the potential loss is going to be no more than a gigabyte or so.

Perfomance degradation can occur on the more sophisticated arrays when rpm, seek time and inteface bandwidth do not match - on a software based array such as the DNS-323 - it does not happen.

On enterprise grade RAID arrays, we routinely replace one make & model of drive with a different make & model, the primary focus being to ensure that the replacement is not a smaller capacity (a larger one may be used - the excess space will not be available), followed by RPM and then interface bus speed.

On my DNS-323 I have built RAID1 arrays with an 80GB Maxtor paired with a 250GB Seagate - I get 80GB of RAID1 storage and approximately 160GB of standard volume - I've also used 250GB Maxtor & Seagate drives, which as you describe are "the same size, RPM, seek time, cache, pretty much identical specs" and I have neither experienced problems, nor do I anticipate any.

Last edited by fordem (2008-04-29 18:20:15)

bscott · 2008-04-30 06:09:11

Well, if there's a more helpful or comprehensive answer to my post, I'm not sure what it would be - thanks!

Bits · 2008-04-30 08:31:05

I'm a bit of a newbie in this but having replaced the NAS and 2 Seagate HDDs...here's my experience.

I experienced the HDD degrading issue that's well documented in this forum. Called DLink up and they replaced the NAS. I went through the pain of re-formatting the HDDs and the pink light came up again.

Called up Seagate and they said I shdn't be running a 11th generation Barracuda and a 10th generation Barracuda (both 750Gb but with different cache buffer) in a RAID1. So they replaced my 11th gen with a 10th gen so I got 2 identical Barracudas.

Went through the re-formatting pain again and 2 weeks later, the NAS is humming along with 2 blue lights (fingers & toes crossed)

bscott · 2008-05-01 06:34:45

Hi Bits - I looked up the problem you referenced, and it does seem to be a puzzler... just as well, perhaps, that the sweet deal on the Samsung drive slipped through my fingers anyway? I can get another Seagate ST3320620AS for $60, so I'll probably just do that since I've gotten used to the idea of replacing my old drives.

But frankly - technical realities aside, I liked Fordem's answer better; it was far more reassuring and upbeat. Maybe you should work on your posting skills? People like "Yeah, sure, go ahead!" more than "well, you may encounter problems..." - just friendly advice! (heh)

mig · 2008-05-01 09:03:56

bscott, Yeah, sure but... whose answer will you like better if you experience the HDD degraded issue?

Last edited by mig (2008-05-01 09:04:17)

fordem · 2008-05-01 15:33:07

Based on D-Link's comments - the hdd degraded issue appears to be an incompatability between the 1.04 firmware on the device and the firmware on specific models of drive

I would not expect a cache size issue, or specifically running a Barracuda 7200.10 along with a 7200.11 to cause a problem - in fact, I'm willing to be that running a pair of 7200.11s would also result in the hdd degraded problem.

It's not my data and therefore not my call - but - on the subject of running "mismatched" drives in a RAID array - as I mentioned earlier, there is very little risk, I do it on a day to day basis - the Lord alone knows how many "mismatched" drives I've installed - don't check to see what brand is in the system (I can't since I'm hotswapping on live systems), but I've been doing it for over twelve years, and not once had a "degraded" array as the result.

Correction - I have had one problem - a RAID1 array on an IBM server - although both drives were nominally the same capacity, the replacement was a few hundred MB smaller than the failed drive, and as a result could not be used (these would have been 36GB SCSI disks) - the fix was to recreate the array using the new drive - however - I don't recall if it was a different brand, it may not have been.

DSM-G600, DNS-3xx and NSA-220 Hack Forum

Announcement

#1 2008-04-29 15:55:32

Mismatched drive risk...?

#2 2008-04-29 18:19:29

Re: Mismatched drive risk...?

#3 2008-04-30 06:09:11

Re: Mismatched drive risk...?

#4 2008-04-30 08:31:05

Re: Mismatched drive risk...?

#5 2008-05-01 06:34:45

Re: Mismatched drive risk...?

#6 2008-05-01 09:03:56

Re: Mismatched drive risk...?

#7 2008-05-01 15:33:07

Re: Mismatched drive risk...?

Board footer