8x 4TB Array -- RAID 5, 6, or 50?

I am using an Areca-1223-8i to run 8 4TB Hitachi 7K4000 Ultrastars in RAID 5. It is used for Blu-ray and DVD iso's so performance isn't necessarily the number one priority. I like that RAID 5 offers the most usable space but from what I read online almost no one uses RAID 5 in large arrays like this--but perhaps this may be because they are usually for enterprise purposes.

The array is backed up with an external JBOD system on a weekly basis, so even in the worst case scenario, failure would not necessarily be catastrophic.

Metadata and artwork is stored on the array, so random access speed is sort of important. Right now it takes a 5-7 seconds to load a screen of two dozen movie covers. I've considered upgrading to a more powerful raid card with dual core to help with this issue but they are extremely expensive.

The entire array is within the media server itself, with the RAID card plugged into the pci-express slot of a X58A motherboard with a Q6600 cpu

raid 10 forever and always
 
Would it make sense to use a small (~64gb) SSD as a cache? That would load the oft-requested artwork and metadata quickly.

It looks like the commercially available products for SSD caching purposes (like the OCZ Synapse) require the OS to be installed on the drive you wish to cache.
 
I also have My Movies installed and find it very slow. I have found the Media Browser is faster now that they've adding caching on the client.
 
I also have My Movies installed and find it very slow. I have found the Media Browser is faster now that they've adding caching on the client.

I tried Media Browser. I like it, but I am too far heavily invested in My Movies at this point to swap.
 
Did you compare the bottom 3 numbers to benchmarks found online for your card (not that I know of any). The top number that measures sequential performance is not at all important for your usage pattern.

John
 
Would it make sense to use a small (~64gb) SSD as a cache? That would load the oft-requested artwork and metadata quickly.

It looks like the commercially available products for SSD caching purposes (like the OCZ Synapse) require the OS to be installed on the drive you wish to cache.

Although I believe the problem is caused by a lot of small reads / writes I am not excactly sure what your software is doing to generate the list. Does it have a database?
 
I tried Media Browser. I like it, but I am too far heavily invested in My Movies at this point to swap.

I'm the same way, I use both now. My Movies on the server for content management, Media Browser for the WMC interface. I also use Media Center Master for get additional back drops and meta data now that My Movies locked theirs down.

Hopefully you find a solution that works that I can use too.
 
Last edited:
If you're hung up on mymovies then unfortunately the speed issue is on his client end and doubtful it ever gets sped up without a completely ground up rewriting of the sotware. Slow now as it was in '06 when I pointed out to the author he was going the wrong way with his design choices and he refused to change. I couldn't deal with the long draw time in MM, a couple hundred titles would take like 15-20 seconds to display at least back in the day. media browser on the other hand will bring up 10000 titles instantly
 
I'm the same way, I use both now. My Movies on the server for content management, Media Browser for the WMC interface. I also use Media Center Master for get additional back drops and meta data now that My Movies locked theirs down.

Hopefully you find a solution that works that I can use too.

I noticed a new feature for MyMovies is to export metadata in MediaBrowser compatible format. You need to have a certain level of contribution points, but if it works correctly, you could continue to use My Movies to manage basic metadata and covers, and it will leave a text file for MediaBrowser to use in the file directory that you can edit yourself to add custom backdrops.

I really like My Movies; my only gripe is that it is slow to load.
 
You dont even need MM for managing meta & covers -- Media Center Master is superior in every way. MediaBrowser + MediaCenterMaster = perfection
 
You dont even need MM for managing meta & covers -- Media Center Master is superior in every way. MediaBrowser + MediaCenterMaster = perfection

Never heard of MCM before but it looks like exactly what I need if I want to move on from My Movies. Thanks odditory
 
You dont even need MM for managing meta & covers -- Media Center Master is superior in every way. MediaBrowser + MediaCenterMaster = perfection

I got to check this out. I use Mymovies for the data and the artwork and the information but it always been slow as poo just to do basic editing too. sigh.

I have a dune, So i use zappiti to run on my player, but when you do individual drives, i still use mymovies. sigh.
 
Never heard of MCM before but it looks like exactly what I need if I want to move on from My Movies. Thanks odditory

Also make sure to you've got a small fast SSD in your HTPC since it makes a big difference in how snappy and fluid the overall experience is. If there's one thing I want the 10' experience w/ a remote in front of the TV to be, its instantaneous, like an appliance.
 
And lose half the space to store movies? He doesn't need the extra speed, so that is a complete waste of space in this instance.

you don't waste space with raid 1, you trade it for better performance! downtime is less frequent and less time consuming. reading and writing is significantly faster.

the guy on the first page had it right about needing raid 6 + hot spare with 8 4tb drives. just to make a parity system work with disks this big you have to bend over backwards. in my opinion more experienced professionals will do raid 10 because they believe anything else is asking for trouble!
 
you don't waste space with raid 1, you trade it for better performance! downtime is less frequent and less time consuming. reading and writing is significantly faster.

the guy on the first page had it right about needing raid 6 + hot spare with 8 4tb drives. just to make a parity system work with disks this big you have to bend over backwards. in my opinion more experienced professionals will do raid 10 because they believe anything else is asking for trouble!


In raid 10 you lose half of your space to parity. Considering the OP is looking to minimize the amount of space wasted, and specifically said he doesn't need the extra performance as it's a media drive..... Raid 10 is a waste.

Honestly, if this is nothing but movies, I would be looking into something like flexraid or another snapshot raid. You would maximize your storage space as it only takes 1 disk for parity.
 
In raid 10 you lose half of your space to parity. Considering the OP is looking to minimize the amount of space wasted, and specifically said he doesn't need the extra performance as it's a media drive..... Raid 10 is a waste.

Honestly, if this is nothing but movies, I would be looking into something like flexraid or another snapshot raid. You would maximize your storage space as it only takes 1 disk for parity.

calling a raid 1 mirror "parity" is not conventional. and it is not a waste, it is the safest raid configuration by far.
 
If you're hung up on mymovies then unfortunately the speed issue is on his client end and doubtful it ever gets sped up without a completely ground up rewriting of the sotware. Slow now as it was in '06 when I pointed out to the author he was going the wrong way with his design choices and he refused to change. I couldn't deal with the long draw time in MM, a couple hundred titles would take like 15-20 seconds to display at least back in the day. media browser on the other hand will bring up 10000 titles instantly

I think I found the solution. I found that if I go into Collection Management, then Tools>Otions and turn off "Update title profiles automatically if update is available" then my cover art displays in less than 1 second.
 
By far? Using what measure?

all of them? when the arrays are degraded, raid 5 takes longer to rebuild than raid 1 or 10 because it needs to process multiple times more data to complete the rebuild. and in that state, raid 1 or 10 can afford to lose some fraction of the remaining drives, while raid 5 can lose none. the math is beyond me now. but the result is raid 5 is multiple times more likely to have data loss than raid 10.
 
You lose half the storage with mirrors. With RAID10 you can only risk losing two drives if they're the safe two. Lose two on the same mirror and you're completely hosed, all data would be lost. More so during a rebuild. With a RAID6 you're capable of losing up to ANY two drives. So you could, in theory, lose another one of the drives during a rebuild. Yes, it does require a lot of reads from the existing drives in order to rebuild replacements. But then it's also holding 33% more data, so some added time is to be expected.

So it's clearly not "all measures". It's a matter of calculated risk. Too many people harp on just their favorite aspect (often one they don't actually understand) of any given storage mechanism.

Whereas a system like ZFS only needs to resilver where there's data present. Unlike some RAID systems requiring a full copy of all blocks on the device, as the RAID controller doesn't know what actually consists of filesystem data.

The weaks point of any redundancy system are going to be stress and risk during rebuilds. It would be enlightening to see an actual study that 'ran the numbers' for equivalent amounts of storage held in each different kind of array. As in, real world rebuild and load numbers based on a more than 60% loaded filesystem.
 
"Literally nothing", except 1-disk-at-a-time online capacity expansion, more data recovery options and available software tools in the event an array becomes corrupted (raidz is a black curtain abstraction layer and if your pool decides it doesnt want to mount one day and your backup isn't current, you're SOL), and most importantly HW raid is independent of host O/S and doesn't require one to run the array.

I gotta say the enthusiasm shared by ZFS fans is usually helpful but gets a bit tiresome when its every thread as it can mislead people just trying to weigh all the options. ZFS is great but not nearly one size fits all for storage.

Also ZFS resilver times get massive under high load situations or on a heavily fragmented array. I have seen resilvers take weeks on ZFS. Never seen a rebuild take that long on hardware raid.

As others have said OP go raid6. Raid6 > Raid10. I deal with raid array failures *all* the time at work as we have >1000 raid arrays. Trust me when I say you are much better off with raid6 rather than raid10/raid5.
 
Last edited:
You lose half the storage with mirrors. With RAID10 you can only risk losing two drives if they're the safe two. Lose two on the same mirror and you're completely hosed, all data would be lost. More so during a rebuild. With a RAID6 you're capable of losing up to ANY two drives. So you could, in theory, lose another one of the drives during a rebuild. Yes, it does require a lot of reads from the existing drives in order to rebuild replacements. But then it's also holding 33% more data, so some added time is to be expected.

So it's clearly not "all measures". It's a matter of calculated risk. Too many people harp on just their favorite aspect (often one they don't actually understand) of any given storage mechanism.

Whereas a system like ZFS only needs to resilver where there's data present. Unlike some RAID systems requiring a full copy of all blocks on the device, as the RAID controller doesn't know what actually consists of filesystem data.

The weaks point of any redundancy system are going to be stress and risk during rebuilds. It would be enlightening to see an actual study that 'ran the numbers' for equivalent amounts of storage held in each different kind of array. As in, real world rebuild and load numbers based on a more than 60% loaded filesystem.

On a full file-system ZFS resilver will likely be slower than a good hardware controller. Also the file-based parity scheme is nice if you have a relatively empty volume but this also (unfortunately) means resilvers are not sequential read/write activity which actually makes a resilver harder on the disks than a hardware controller running raid6 (vs raidz2). The only thing i really like about ZFS is they have raidz3.
 
Why are you using RAID at all for this task? RAID is designed to maximize availability. I dare say that if you're without your movie library for a couple of hours while a backup is being restored, you'll somehow manage to live through the day. At least you're bright enough to keep a backup rather than being under the mistaken impression that RAID functions as some type of backup system.

Is there a requirement for a contiguous 28 GB (or whatever it ends up being) of disk space? If not, go with JBOD and you'll have a simpler system, a less expensive system, and a more operator-proof system. I'd be willing to wager that 50% of the users running RAID on a home network would f**k up any attempt at restoring or rebuilding the array in the case of a disk failure. And that 90% of them would be unable to recover from a failed controller. Then look at the simple fact that even when you do lose a disk in an eight disk JBOD system, 88% of your movies will still be available to you, rather than having a system running in a degraded state until it's rebuilt.
 
Why are you using RAID at all for this task? RAID is designed to maximize availability. I dare say that if you're without your movie library for a couple of hours while a backup is being restored, you'll somehow manage to live through the day. At least you're bright enough to keep a backup rather than being under the mistaken impression that RAID functions as some type of backup system.

Is there a requirement for a contiguous 28 GB (or whatever it ends up being) of disk space? If not, go with JBOD and you'll have a simpler system, a less expensive system, and a more operator-proof system. I'd be willing to wager that 50% of the users running RAID on a home network would f**k up any attempt at restoring or rebuilding the array in the case of a disk failure. And that 90% of them would be unable to recover from a failed controller. Then look at the simple fact that even when you do lose a disk in an eight disk JBOD system, 88% of your movies will still be available to you, rather than having a system running in a degraded state until it's rebuilt.


I think you are inflating those percentages a bit. 50% of the users rune a raid on a rebuild? I find that highly unlikely especially with raid6 where you can pull the wrong drive in a degraded state.

Also most people like to have everything accessible in a single volume and the performance boost is nice as well. If you span the file-system over a single volume then odds are you are going to be in a situation where a single disk failure is going to cause massive issues.

Of course raid is not backup but if all your previous data loss came from a failing hard-drive then its pretty understandable to use raid as a way to not have issues/data loss from a drive randomly failing.

This OP does not sound like a noob nor does it sound like for his usage he should go JBOD.
 
Also most people like to have everything accessible in a single volume and the performance boost is nice as well. If you span the file-system over a single volume then odds are you are going to be in a situation where a single disk failure is going to cause massive issues.

That's the crux of most consumer media library situations. All too often the software being used to access the media is not as flexible as one would like. So having everything descend from one point in the filesystem is necessary. The hassle is trying to juggle just how to get everything "at" that point. Various schemes exist but all have their compromises.

And it's phenomenally naive and perhaps more than a little arrogant to be so dismissive about the backup recovery times. For anyone that's ever had to deal with impatient family members or guests, it's cold comfort saying it'll only take "a couple hours" to recover..
 
Also most people like to have everything accessible in a single volume

Sure, but it's a really weak reason for running RAID. Especially in this use case where I imagine everything is normally accessed through library management software that couldn't care less whether the files are stored in a single volume.

and the performance boost is nice as well.

But is it necessary for the task at hand? Do the alternatives fail to provide sufficient performance?

If you span the file-system over a single volume then odds are you are going to be in a situation where a single disk failure is going to cause massive issues.

Which is why you shouldn't do that.

Of course raid is not backup but if all your previous data loss came from a failing hard-drive then its pretty understandable to use raid as a way to not have issues/data loss from a drive randomly failing.

Again, high availability is the reason to run RAID. Using it for a movie collection that is faithfully backed up is overkill, at best. More complex than it needs to be, more prone to mistakes, more expensive. More everything except reliable.

RAID does not protect against issues with disk failures any better than JBOD. It keeps the data online and available when there is a disk failure. Availability. There are many mission critical systems where high availability is an absolute must, no questions asked. Keeping Saving Private Ryan available 24/7 isn't likely to be one of them.

This OP does not sound like a noob nor does it sound like for his usage he should go JBOD.

Noob or not, the solution should fit the problem.
 
Sure, but it's a really weak reason for running RAID. Especially in this use case where I imagine everything is normally accessed through library management software that couldn't care less whether the files are stored in a single volume.
Clearly you're not using the same software as others out in the real world are using. Yes, it would be nice if the software handled it all. Hand waving won't make that happen.

Again, high availability is the reason to run RAID. Using it for a movie collection that is faithfully backed up is overkill, at best. More complex than it needs to be, more prone to mistakes, more expensive. More everything except reliable.
THE reason? Wow, quite the expert there, aren't you? I guess everyone else should just give up then.

RAID does not protect against issues with disk failures any better than JBOD. It keeps the data online and available when there is a disk failure. Availability. There are many mission critical systems where high availability is an absolute must, no questions asked.
What is it you consider an "issue" then? RAID can save the trouble of recovering data from backups. This is not something to dismiss so readily. Yes, the data on a single spindle mounted individually could be lost and recovered from backup. Or the array could be rebuilt in parallel without any recovery at all. What kind of complexity is it you're arguing is acceptable here? Because recovery is often quite a lot more complex than just swapping in a drive and telling or letting the array rebuild.

Keeping Saving Private Ryan available 24/7 isn't likely to be one of them.
Maybe for you, no. But others may have different desires. Arguing they're stupid is not very helpful. And does more to make you look arrogant rather than knowledgeable.

Noob or not, the solution should fit the problem.
Who's the noob here? Because with the tone of your responses you're not coming across as being very expert. Opinionated, sure, but expert?
 
clearly what the hell. raid 5 is more likely to lose data than raid 10. this is not disputable.

Perhaps, but I think the comparison was 10 vs. 6, where 6 has a much better chance of keeping your data intact during a rebuild.

I'm also perplexed by the notion that RAID is unnecessarily complex even when the data is backed up elsewhere. RAID means you've got a fighting chance at not having to restore from backup in the event of failure, which can take a considerably greater amount of time than a rebuild.
 
I'm also perplexed by the notion that RAID is unnecessarily complex even when the data is backed up elsewhere. RAID means you've got a fighting chance at not having to restore from backup in the event of failure, which can take a considerably greater amount of time than a rebuild.

Again... what kinda of data are we talking about, and what is the operating environment? It's not like the accounting department of the company is going to be standing around with their thumbs up their arses while you bring the database server back to snuff. They're freaking movies on a home network!

Use some common sense.
 
RAID means you've got a fighting chance at not having to restore from backup in the event of failure, which can take a considerably greater amount of time than a rebuild.

The last comment is likely incorrect for most reasonable home data-loss situations. If you have your movies stored on a bunch of HDDs with each HDD having its own filesystem, then if one of your HDDs dies at worst you would have to restore one HDD of data from your backup. Of course, if you had snapshot RAID, you could just restore the one drive from parity.

If you have a conventional distributed-parity RAID 6 and one drive dies, you also only have to restore one drive from parity. As with the bunch of HDDs situation above, the bottleneck is the time to fill one HDD at whatever write speed the HDD is capable of. So the time would be about the same in both cases.

Except with RAID 6, if the operator screws up, or you just have some really bad luck, you could lose the entire array. Then, unless your backup is another RAID 6 array and you can connect to it at very high speed, THAT restore from backup is going to be much slower.

I am perplexed by the notion that anyone would even suggest that distributed parity RAID is a better solution for a home media fileserver than snapshot RAID (snapRAID or FlexRAID), possibly with the addition of a virtual drive pool if desired. The distributed parity RAID is just so obviously the inferior option for a home media fileserver.
 
Again... what kinda of data are we talking about, and what is the operating environment? It's not like the accounting department of the company is going to be standing around with their thumbs up their arses while you bring the database server back to snuff. They're freaking movies on a home network!

Use some common sense.

Have you ever sat around ripping movies to computer? Clearly not, because you wouldn't make this argument then. Even if they're all Bluray rips at 40GB each (unlikely, most are closer to 20GB), you're looking at several days of ripping if a single drive is wiped out. Yes, it's backed up elsewhere, but just because you don't consider it important enough to warrant raid does not mean that everyone else agrees. In fact, I think you'd be pretty amazed at just how many people value their time enough to use raid for "freaking movies on a home network".

If things are only worth doing because money is at stake, then how much do you value your free time? I put a fairly high price on mine, meaning that having to re-rip 2TB worth of movies is worth several hundred dollars to me.
 
The last comment is likely incorrect for most reasonable home data-loss situations. If you have your movies stored on a bunch of HDDs with each HDD having its own filesystem, then if one of your HDDs dies at worst you would have to restore one HDD of data from your backup. Of course, if you had snapshot RAID, you could just restore the one drive from parity.

If you have a conventional distributed-parity RAID 6 and one drive dies, you also only have to restore one drive from parity. As with the bunch of HDDs situation above, the bottleneck is the time to fill one HDD at whatever write speed the HDD is capable of. So the time would be about the same in both cases.

Except with RAID 6, if the operator screws up, or you just have some really bad luck, you could lose the entire array. Then, unless your backup is another RAID 6 array and you can connect to it at very high speed, THAT restore from backup is going to be much slower.

I am perplexed by the notion that anyone would even suggest that distributed parity RAID is a better solution for a home media fileserver than snapshot RAID (snapRAID or FlexRAID), possibly with the addition of a virtual drive pool if desired. The distributed parity RAID is just so obviously the inferior option for a home media fileserver.

For clarity, when most people refer to the "backup" they have of their ripped movies, they are referring to the original DVD/Bluray, not another set of hard drives. The time required to re-rip several TB of movies is not insignificant.

FlexRAID is not magically superior to other solutions--it's the most user friendly and "idiot proof", I suppose, compared to RAID 6, but it's not the end-all be-all for home use. If you're never going to do anything besides stream movies, then the performance is good enough.

However, this is [H]. When the fsck did we start doing things "good enough"??
 
Have you ever sat around ripping movies to computer? Clearly not, because you wouldn't make this argument then. Even if they're all Bluray rips at 40GB each (unlikely, most are closer to 20GB), you're looking at several days of ripping if a single drive is wiped out.

He _has_ a backup. He would not have to rerip anything due to drive failure.

Note that I'm not questioning the importance of the data. It could be 32 TB of home movies or photos of Gramma and the kids at Disneyland. Doesn't matter. What I'm calling into question is the idea that in the case of hardware failure or file corruption, this type of data needs to kept online 24/7, with no consideration for either the cost or the complexity of keeping it all afloat.

I think the big logical disconnect here is that the vast majority of home users with large arrays do not implement any type of backup solution (It gets expensive backing up 20 TB of data, doesn't it?), and instead they rely only on RAID. It's not the smartest approach to protecting important data, but I suppose it's better than nothing.
 
For clarity, when most people refer to the "backup" they have of their ripped movies, they are referring to the original DVD/Bluray, not another set of hard drives. The time required to re-rip several TB of movies is not insignificant.

FlexRAID is not magically superior to other solutions--it's the most user friendly and "idiot proof", I suppose, compared to RAID 6, but it's not the end-all be-all for home use. If you're never going to do anything besides stream movies, then the performance is good enough.

However, this is [H]. When the fsck did we start doing things "good enough"??

Clarity? Actually, you seem confused. The OP said he had a JBOD backup. That is what I was talking about.

But if you want to change the subject to some sort of general home situation, then I agree that most people only have the original DVDs and blu-rays as backup. Which is why, in the general case, snapshot RAID is the way to go for a home media fileserver. You can choose dual-parity snapshot RAID so you will not have to re-rip unless you lose more than two drives at once. So it is equivalent to RAID 6 in that regard. But snapshot RAID is superior to RAID 6 for home media fileservers because with RAID 6 you are more likely to lose your entire RAID volume of data than you are with a snapshot RAID / bunch of HDDs configuration. Since most people set up RAID 6 with a single filesystem across the entire RAID, then if the filesystem metadata becomes corrupt, you could lose all your data. But with snapshot RAID, each drive has its own filesystem and you are usually only writing to one (or two at the most) drives at a time, so if a filesystem becomes corrupt, you only lose one drive (or two at the most) and you can restore from parity. Also, if you have three drives die at once, with RAID 6 you just lost everything. With snapshot RAID, you only lost data from the dead drives.

I am perplexed by the notion that any [H] person would even consider for an instant that distributed parity RAID is "good enough" for a home media fileserver when snapshot RAID is obviously better.
 
Use some common sense.

Common sense? Ah, right, more insulting tone trying to distract from a weak argument. Puh-leeze that nonsense just won't fly. As has been pointed out, this is [H], there are many reasons why people choose overkill.
 
Back
Top