NZFS or Flexraid (Raid - F)? - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #1 of 58 Old 05-09-2013, 01:46 PM - Thread Starter
Senior Member
 
Elpee's Avatar
 
Join Date: Dec 2007
Posts: 379
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 17
I've read thru this thread and this, but still confused a lot. confused.gif Can you tell me if there is any advantages of next generation NZFS over Flexraid?
I was told NZFS was being lauched in the next few weeks. But I've been waiting for several "next few weeks" already. Haven't seen it out yet. mad.gif
My Flexraid 14-days trail will end tomorrow on my home media server (just for storage). Should I wait another "next few weeks" or just go ahead and buy FlexRaid (in fact RAID-F)?
Thank you.
Elpee is offline  
Sponsored Links
Advertisement
 
post #2 of 58 Old 05-09-2013, 02:36 PM
AVS Special Member
 
amarshonarbangla's Avatar
 
Join Date: Mar 2011
Posts: 1,239
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 39
IIRC, only the tRAID portion of NZFS will be coming out soon. tRAID is a byproduct of NZFS as claimed by the developer. It has a better realtime RAID engine than FlexRAID, which is unnecessary IMO for a media server. But the feature I am most excited on tRAID is live RAID reconstruction. If I understand correctly, if a drive was to fail in the pool, it will automatically convert one of the PPUs into the lost DRU, very neat feature. I am confused myself if tRAID is going to be implemented in FlexRAID but there's a major update coming and FlexRAID is supposed to get even better. I say stop waiting and just get FlexRAID. For a media server, I doubt NZFS will provide any significant advantages over FlexRAID.
amarshonarbangla is offline  
post #3 of 58 Old 05-09-2013, 03:53 PM
AVS Special Member
 
Nethawk's Avatar
 
Join Date: Sep 2001
Posts: 2,513
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 114
Live RAID reconstruction is what I want too. I'd rather not be tethered to Newegg and UPS to get my system up and running again. FlexRAID is adequate, but if shorter downtime is important it's worth a wait. I'm not getting a warm and fuzzy feeling about any significant breaks for upgrade costs to the newer technology. I just bought the FlexRAID bundle two weeks ago and they're not answering emails about deals or refund to wait for new product.

Nethawk is offline  
post #4 of 58 Old 05-09-2013, 06:08 PM
AVS Special Member
 
amarshonarbangla's Avatar
 
Join Date: Mar 2011
Posts: 1,239
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 39
Quote:
Originally Posted by Nethawk View Post

Live RAID reconstruction is what I want too. I'd rather not be tethered to Newegg and UPS to get my system up and running again. FlexRAID is adequate, but if shorter downtime is important it's worth a wait. I'm not getting a warm and fuzzy feeling about any significant breaks for upgrade costs to the newer technology. I just bought the FlexRAID bundle two weeks ago and they're not answering emails about deals or refund to wait for new product.

I believe current FlexRAID users will get a discount
Quote:
Originally Posted by spectrumbx View Post

Current FlexRAID users will indeed get a discount on NZFS.
I prefer not to have a few users meet me in a dark alley and put some discount on me otherwise. biggrin.gif

http://www.avsforum.com/t/1457880/a-first-look-at-nzfs-and-replacing-unraid-with-nzfs-s-transparent-raid-traid#post_22957394
amarshonarbangla is offline  
post #5 of 58 Old 05-10-2013, 04:39 AM
AVS Addicted Member
 
Mfusick's Avatar
 
Join Date: Aug 2002
Location: Western MA
Posts: 21,511
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 663
Is there any reason why normal Flexraid isn't good for media server , or that newer version of t raid would be better ?

If answer not clear you prob want normal Flexraid like the hundreds of people in this forum are running.

-

"Too much is almost enough. Anything in life worth doing is worth overdoing. Moderation is for cowards."
Mfusick is online now  
post #6 of 58 Old 05-10-2013, 06:17 AM
AVS Special Member
 
Nethawk's Avatar
 
Join Date: Sep 2001
Posts: 2,513
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 114
Maybe you didn't read beyond the subject line of this thread? It's not a matter of "good enough" but rather "better". FlexRAID is a compromise, and better than nothing at all. tRAID is an altogether better solution.

Nethawk is offline  
post #7 of 58 Old 05-10-2013, 07:21 AM
AVS Addicted Member
 
Mfusick's Avatar
 
Join Date: Aug 2002
Location: Western MA
Posts: 21,511
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 663
I didn't make a judgement I only raised the question.

How or why is it better ?

-

"Too much is almost enough. Anything in life worth doing is worth overdoing. Moderation is for cowards."
Mfusick is online now  
post #8 of 58 Old 05-10-2013, 07:44 AM - Thread Starter
Senior Member
 
Elpee's Avatar
 
Join Date: Dec 2007
Posts: 379
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 17
I have two things to tell you and that's the reasons why I said I'm so confused now.

First, from the blog, it said tRAID is not a replacement for RAID-F, but I find they are almost the same.

Second, the later versions come out the better for sure features they have. But, look at what I found from blog


Datarot detection (RAID-F - Yes ; tRAID - No)
Can RAID include specific folders, specific data set, media drives, and ad-hoc content? (RAID-F - Yes ; tRAID - No)
Ability to exclude certain content from the RAID? (RAID-F - Yes ; tRAID - No)
Recovery of specific files (RAID-F - Yes ; tRAID - No, only recover a whole drive. Has no concept of files)
Supports network mapped drives as part of the RAID? (RAID-F - Yes ; tRAID - No)
Can upgrade from a small drive to a bigger drive by just copying the data over and without affecting parity? (RAID-F - Yes ; tRAID - No)
Elpee is offline  
post #9 of 58 Old 05-10-2013, 09:13 AM
AVS Addicted Member
 
Mfusick's Avatar
 
Join Date: Aug 2002
Location: Western MA
Posts: 21,511
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 663
RAID F is Flexraid. So I'll ask the question again,

Is there some feature lacking on Flexraid or something about T raid/NZFS that makes it clearly better ?

If not I'd say just get Flexraid.

Personally I like the concept of Flexraid and that your files remain as they are and readable by any machine inside or outside of a Flexraid setup.

That's huge for media server

-

"Too much is almost enough. Anything in life worth doing is worth overdoing. Moderation is for cowards."
Mfusick is online now  
post #10 of 58 Old 05-10-2013, 09:34 AM
AVS Special Member
 
Nethawk's Avatar
 
Join Date: Sep 2001
Posts: 2,513
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 114
^^
Quote:
But the feature I am most excited on tRAID is live RAID reconstruction. If I understand correctly, if a drive was to fail in the pool, it will automatically convert one of the PPUs into the lost DRU, very neat feature.

If you have a spare HDD laying around, then it may not make much difference. If you don't, you still have access to your files without the downtime.

That said, I'm now as confused as Elpee. While I only care about whole drives rather than files I'm not sure what effect the upgrade from small drive to bigger drive means. I presume it's as simple as rebuilding parity, which would be required regardless.

Nethawk is offline  
post #11 of 58 Old 05-12-2013, 09:21 AM
AVS Addicted Member
 
Mfusick's Avatar
 
Join Date: Aug 2002
Location: Western MA
Posts: 21,511
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 663
Right.

-

"Too much is almost enough. Anything in life worth doing is worth overdoing. Moderation is for cowards."
Mfusick is online now  
post #12 of 58 Old 05-12-2013, 09:45 AM
AVS Special Member
 
amarshonarbangla's Avatar
 
Join Date: Mar 2011
Posts: 1,239
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 39
Quote:
Originally Posted by Nethawk View Post

^^
If you have a spare HDD laying around, then it may not make much difference.

Automatic reconstruction makes a HUGE difference.
Quote:
Originally Posted by Nethawk View Post

If you don't, you still have access to your files without the downtime.

Are you talking about FlexRAID here? If a DRU was to fail in FlexRAID, the files in that drive will be unavailable until you replace the drive and reconstruct the array.
amarshonarbangla is offline  
post #13 of 58 Old 05-12-2013, 11:03 AM
AVS Addicted Member
 
Mfusick's Avatar
 
Join Date: Aug 2002
Location: Western MA
Posts: 21,511
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 663
Right.

But if the drive is only unreadable in certain areas (like part is not readable) the part that is can still be read (individidual files)

And yes you could also rebuild the entire drive and it's contents via parity.

Bottom line with flexraid is anything that is readable is readable until the drive itself is dead. In which case you rebuild the entire drive. But I believe with flexraid you can recover individual files or items without needing to rebuild the entire drive (even though you can recover and entire drive easily)

-

"Too much is almost enough. Anything in life worth doing is worth overdoing. Moderation is for cowards."
Mfusick is online now  
post #14 of 58 Old 05-12-2013, 03:01 PM
AVS Special Member
 
Nethawk's Avatar
 
Join Date: Sep 2001
Posts: 2,513
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 114
Quote:
Originally Posted by amarshonarbangla View Post

Automatic reconstruction makes a HUGE difference.
Are you talking about FlexRAID here? If a DRU was to fail in FlexRAID, the files in that drive will be unavailable until you replace the drive and reconstruct the array.

Hey, that was as clear as mud to me! tongue.gif I smashed a knuckle last week replacing the muffler on one of my Alfas. Typing on the tablet wasn't the most fun in the world, brevity was called for. I think you got it, but reworded:

If you don't, with tRAID you still have access to your files without the downtime.
amarshonarbangla likes this.

Nethawk is offline  
post #15 of 58 Old 05-12-2013, 06:02 PM
AVS Special Member
 
amarshonarbangla's Avatar
 
Join Date: Mar 2011
Posts: 1,239
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 39
Quote:
Originally Posted by Nethawk View Post

If you don't, with tRAID you still have access to your files without the downtime.

Agreed.
amarshonarbangla is offline  
post #16 of 58 Old 05-13-2013, 08:50 AM - Thread Starter
Senior Member
 
Elpee's Avatar
 
Join Date: Dec 2007
Posts: 379
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 17
Quote:
Originally Posted by amarshonarbangla View Post

Agreed.

Quote:
Originally Posted by Elpee View Post

I have two things to tell you and that's the reasons why I said I'm so confused now.

First, from the blog, it said tRAID is not a replacement for RAID-F, but I find they are almost the same.

Second, the later versions come out the better for sure features they have. But, look at what I found from blog


Datarot detection (RAID-F - Yes ; tRAID - No)
Can RAID include specific folders, specific data set, media drives, and ad-hoc content? (RAID-F - Yes ; tRAID - No)
Ability to exclude certain content from the RAID? (RAID-F - Yes ; tRAID - No)
Recovery of specific files (RAID-F - Yes ; tRAID - No, only recover a whole drive. Has no concept of files)
Supports network mapped drives as part of the RAID? (RAID-F - Yes ; tRAID - No)
Can upgrade from a small drive to a bigger drive by just copying the data over and without affecting parity? (RAID-F - Yes ; tRAID - No)

If so, anybody can explain this?
Elpee is offline  
post #17 of 58 Old 05-13-2013, 09:24 AM
AVS Special Member
 
amarshonarbangla's Avatar
 
Join Date: Mar 2011
Posts: 1,239
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 39
I am confused about your confusion. What's your confusion exactly?

tRAID is not the same as RAID-F. It's most likely going to be the replacement for FlexRAID's current Realtime RAID engine.
amarshonarbangla is offline  
post #18 of 58 Old 05-13-2013, 09:38 AM
Senior Member
 
DavidT99's Avatar
 
Join Date: Jun 2009
Location: United Kingdom
Posts: 226
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 13
I think Elpee is confused as to why the items he/she listed are all supported by RAID-F but not tRAID. Four of them look pretty important to me so given the short subset of features quoted how is tRAID better than RAID-F. Or is it not meant to be better, just different.

David
DavidT99 is offline  
post #19 of 58 Old 05-13-2013, 10:06 AM
AVS Special Member
 
Dark_Slayer's Avatar
 
Join Date: May 2012
Posts: 2,296
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 54 Post(s)
Liked: 228
Quote:
Originally Posted by Elpee View Post

First, from the blog, it said tRAID is not a replacement for RAID-F, but I find they are almost the same.
You find they are almost the same based on what?

raid-f is basically smartmontools, cron, and LVM ported to Windows with a Web UI slapped on top. Linux support as well (obviously). Mac support promised, not delivered

traid looks like nothing other than zfs port, but "may" be host-free. NZFS is a made up name by the dev(s), and it remains to be seen whether or not it will be a true filesystem or just a misuse of the term
Quote:
Originally Posted by Elpee View Post

Datarot detection (RAID-F - Yes ; tRAID - No)
Yeah, I don't know what kind of datarot detection raid-f provides, whatever it "detects" it has no way of healing nor does the "detection" make any useable since if it's regarding the s.m.a.r.t. reports you receive from your array

However, ZFS (and likely NZFS) supports checksums on all data and self healing, so it's definitely a more robust data protection way to go
Quote:
Originally Posted by Elpee View Post

Can RAID include specific folders, specific data set, media drives, and ad-hoc content? (RAID-F - Yes ; tRAID - No)
Ability to exclude certain content from the RAID? (RAID-F - Yes ; tRAID - No)
Recovery of specific files (RAID-F - Yes ; tRAID - No, only recover a whole drive. Has no concept of files)
all essentially the same thing, traid won't allow parity/backup protection of symlinks or any other junction method - all or none in regards to drive
Quote:
Originally Posted by Elpee View Post

Supports network mapped drives as part of the RAID? (RAID-F - Yes ; tRAID - No)
No network mapped drives? I for one would never ever use a network mapped drive anyway. If I have additional drives outside my server in my home that need backing up, they are typically attached to something that is NOT powered on 24/7. Even if the server could wake / write parity / and return to sleep I'd still never use a mapped drive in a server software raid since it introduces too many possibilities I'd prefer to avoid
Quote:
Originally Posted by Elpee View Post

Can upgrade from a small drive to a bigger drive by just copying the data over and without affecting parity? (RAID-F - Yes ; tRAID - No)
Don't know what to say, I didn't even know that was a feature. I'd have to assume most people wouldn't use it. Typically shutting down, replacing a drive, booting up, I'd think most people would just go ahead and rebuild the parity
Dark_Slayer is offline  
post #20 of 58 Old 05-13-2013, 10:30 AM
Advanced Member
 
jim2100's Avatar
 
Join Date: Aug 2011
Posts: 762
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 33
Quote:
Originally Posted by Dark_Slayer View Post

Yeah, I don't know what kind of datarot detection raid-f provides, whatever it "detects" it has no way of healing nor does the "detection" make any useable since if it's regarding the s.m.a.r.t. reports you receive from your array

I believe it maintains hashes (checksums) at the file level. I don't know that for certain, but that is what I have heard. If true, then it could certainly heal a corrupted file -- just do a normal file restore using parity, and then verify the file hash to see if it is fixed.

If RAID-F does use file-level hashes, then this is a case where SnapRAID is significantly better than FlexRAID, because SnapRAID uses block-level hashes. The advantage there (in addition to more fine-grained checking) is that whenever SnapRAID reads a block of data (for example, if it is doing a sync after new files were added, and it has to read some old data in order to compute the new parity) it verifies the hash to see if the data has been corrupted. So with SnapRAID, even if you never run a full check for silent data corruption, you will nevertheless be protected from having your newly-generated parity data corrupted due to silent data corruption.
jim2100 is offline  
post #21 of 58 Old 05-13-2013, 10:35 AM
Advanced Member
 
jim2100's Avatar
 
Join Date: Aug 2011
Posts: 762
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 33
Quote:
Originally Posted by Dark_Slayer View Post

Don't know what to say, I didn't even know that was a feature. I'd have to assume most people wouldn't use it. Typically shutting down, replacing a drive, booting up, I'd think most people would just go ahead and rebuild the parity

Huh? It is certainly a useful feature. Why spend many hours regenerating the parity if you do not have to? SnapRAID is capable of that, too (basically, any system that works at the filesystem level should be able to do it) and I have done it.
jim2100 is offline  
post #22 of 58 Old 05-14-2013, 11:29 AM - Thread Starter
Senior Member
 
Elpee's Avatar
 
Join Date: Dec 2007
Posts: 379
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 17
Quote:
Originally Posted by amarshonarbangla View Post

tRAID is not the same as RAID-F. It's most likely going to be the replacement for FlexRAID's current Realtime RAID engine.

I now can understand what you meant.
I did test the snapshot FlexRaid with 1 parity drive (4TB) and 4 Data drives (1x640MB, 2x250MB and 1x 150MB) a few days ago, I initalized raid and it took me overnight (~ 9 hours) to run. It was still under "running" when I checked in the morning. Cancelled that job and cross my fingers with hope NZFS coming soon.
Imagine just 'raid initialization' of 20TB server, it cost 66 hours to finish. I really don't know about my 100 TB server just for raid initialization, how long?

66 x 5 = 330 hours ~ 14 days eek.gif

That's really crazy and so stupid. Sorry, but it is.

Back to the topic. As you said tRAID is not same as RAID-F, I know. Raid-f is known as RAID over Filesystem and tRAID is known as RAID under filesystem and RAID within filesystem. As customers, who care? All we care is how many reliable and neccessary features beside a cheaper price that RAID software can provide.
I think NZFS, this time, casts away all features those Raid-f is currently carrying and just focuses on RT Raid engine because the example I just said above. Once again snapshot vs RT raid.
Correct me if I'm wrong.

Hi, Jim2100. I know you're running snapRaid on your server. How about that one? Good? Any suggestions in this case? Thanks.
Elpee is offline  
post #23 of 58 Old 05-14-2013, 12:59 PM
AVS Special Member
 
amarshonarbangla's Avatar
 
Join Date: Mar 2011
Posts: 1,239
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 39
No it's won't take 66hours to initialize 20TB of data. If you read further down the thread, you will see it took him 11 hours I believe. Remember, FlexRAID reads from all the HDDs in parallel, not one by one. So parity calculation is much faster than you would think.

http://www.avsforum.com/t/1457151/how-long-should-20tb-of-full-drives-take-to-initialize-flexraid-help-me-with-basic-math/60#post_22957457

http://wiki.flexraid.com/2013/02/12/understanding-the-throughput-reported-by-flexraid/
amarshonarbangla is offline  
post #24 of 58 Old 05-14-2013, 01:16 PM
Advanced Member
 
jim2100's Avatar
 
Join Date: Aug 2011
Posts: 762
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 33
Quote:
Originally Posted by Elpee View Post

Hi, Jim2100. I know you're running snapRaid on your server. How about that one? Good? Any suggestions in this case? Thanks.

I run SnapRAID on my linux server. It works very well. SnapRAID's design is better than FlexRAID's (block- rather than file-based checksums) and SnapRAID is more focused, which I like. But some people like a GUI and want all-in-one programs rather than the linux approach of having a streamlined program for each distinct task. SnapRAID can be run under Windows, there is a third party GUI available for it, and SnapRAID recently added some basic pooling capability, but I think SnapRAID still retains some of its focused do-one-task-and-do-it-well flavor, as compared to FlexRAID which to me seems to be a jack-of-all-trades master-of-none and have a GUI that slows me down (give me a clean text file configuration any day, but that is me, I know others have different opinions). Of course, SnapRAID is free and open-source, unlike FlexRAID.
jim2100 is offline  
post #25 of 58 Old 05-14-2013, 01:23 PM
Advanced Member
 
jim2100's Avatar
 
Join Date: Aug 2011
Posts: 762
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 33
Quote:
Originally Posted by amarshonarbangla View Post

No it's won't take 66hours to initialize 20TB of data. If you read further down the thread, you will see it took him 11 hours I believe. Remember, FlexRAID reads from all the HDDs in parallel, not one by one. So parity calculation is much faster than you would think.

Right. A rough approximation is that the time to generate the parity from scratch is a little longer than the time required to read all the data on your drive that takes longest to read all the data from (generally the one with the most TBs of data, but could be another one if you have a smallish slow drive that is full). The reason I say "a little longer" is that, at any point during the parity computation, the aggregate read speed of all the drives will be limited by the speed of the slowest drive which has not yet been completely read, since an equal amount of data per second must be read from each drive in order to compute the parity. So if you have a very slow drive as well as a large, full faster drive, the faster drive will be read at a slower rate than the maximum speed it is capable of.
jim2100 is offline  
post #26 of 58 Old 05-14-2013, 01:24 PM
AVS Special Member
 
Dark_Slayer's Avatar
 
Join Date: May 2012
Posts: 2,296
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 54 Post(s)
Liked: 228
Quote:
Originally Posted by Elpee View Post

66 x 5 = 330 hours ~ 14 days eek.gif

That's really crazy and so stupid. Sorry, but it is.

Yeah, the developers post quoted above better explains throughput, and there is a similar discussion in the wiki

Unfortunately nothing ever takes it all the way down to basic examples with seconds. My guess at it is in this post and it is pretty close for both of our parity build times
Dark_Slayer is offline  
post #27 of 58 Old 05-14-2013, 01:27 PM
AVS Special Member
 
Dark_Slayer's Avatar
 
Join Date: May 2012
Posts: 2,296
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 54 Post(s)
Liked: 228
Quote:
Originally Posted by jim2100 View Post

Right. A rough approximation is that the time to generate the parity from scratch is a little longer than the time required to read all the data on your drive that takes longest to read all the data from (generally the one with the most TBs of data, but could be another one if you have a smallish slow drive that is full)

Remember it's writing to the parity disk, so it's safe to assume that could be your slowest spot. Also if the slowest drive is substantially smaller than the others it should only affect the throughput for that percentage of the total parity build
Dark_Slayer is offline  
post #28 of 58 Old 05-14-2013, 01:30 PM
Advanced Member
 
jim2100's Avatar
 
Join Date: Aug 2011
Posts: 762
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 33
Quote:
Originally Posted by Dark_Slayer View Post

Remember it's writing to the parity disk, so it's safe to assume that could be your slowest spot.

Not a good assumption in my experience. The parity drive will have capacity equal to or larger than the largest data drive, and that generally means it will be as fast or faster than the largest data drive. Also, the parity drive is written sequentially during the initial sync, while some data drives may be having some data read from the inner portion of the drive (i.e., slower) even at the beginning of the sync.
jim2100 is offline  
post #29 of 58 Old 05-14-2013, 01:33 PM
AVS Special Member
 
Dark_Slayer's Avatar
 
Join Date: May 2012
Posts: 2,296
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 54 Post(s)
Liked: 228
Quote:
Originally Posted by jim2100 View Post

Not a good assumption in my experience. The parity drive will have capacity equal to or larger than the largest data drive, and that generally means it will be as fast or faster than the largest data drive.
Not sure what that is supposed to indicate. Capacity and speed are not inter-related. Write speeds are almost universally slower than read speeds
Dark_Slayer is offline  
post #30 of 58 Old 05-14-2013, 01:41 PM
Advanced Member
 
jim2100's Avatar
 
Join Date: Aug 2011
Posts: 762
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 33
Quote:
Originally Posted by Dark_Slayer View Post

Capacity and speed are not inter-related. Write speeds are almost universally slower than read speeds

Wrong, and wrong.

Capacity and speed are correlated. Since hard drives have at most 5 platters (most are 3 or 4), the main way to increase capacity is to increase the density of data on the platters. Since the rotational speed of most drives can be expected to be the same (either ~5400rpm or 7200rpm) in a given system (most people would not choose a 5400rpm parity drive and 7200rpm data drives), the drive with the higher platter data density will have the higher throughput, which means the higher capacity drive will have higher throughput.

The sequential read and write speeds of hard drives are essentially the same. You can look up reviews that measure it with HD Tune, but really, this is an obvious point. Since the platter speed is constant (or almost so, some of the low-power drives may vary the speed slightly in a low-power mode), the throughput will be the same as the platter spins under the read/write head, regardless of whether the drive is reading or writing.
jim2100 is offline  
Reply Home Theater Computers

User Tag List

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off