May 1, 2012 at 7:01 am
rmechaber (5/1/2012)
So, while I've got your attention: is there any? miniscule? significant? write-performance hit in writing to a 2-drive mirror compared to writing the same data to a single disk (assuming a hardware RAID controller)? That's what I was really inquiring about.
Sure, it depends on the RAID controller or storage processor. If it was just one write in absolute isolation, it's going to be minimal, but as disk activity starts to pile up, you can run into problems with the bandwidth of the storage gear altogether. If the storage gear doesn't have enough connections to the drives, or if it uses a small shared connection (like 2Gb FC) across a whole lot of drives, then you can run into a bottleneck there.
May 1, 2012 at 7:04 am
Brent Ozar (5/1/2012)
rmechaber (5/1/2012)
So, while I've got your attention: is there any? miniscule? significant? write-performance hit in writing to a 2-drive mirror compared to writing the same data to a single disk (assuming a hardware RAID controller)? That's what I was really inquiring about.Sure, it depends on the RAID controller or storage processor. If it was just one write in absolute isolation, it's going to be minimal, but as disk activity starts to pile up, you can run into problems with the bandwidth of the storage gear altogether. If the storage gear doesn't have enough connections to the drives, or if it uses a small shared connection (like 2Gb FC) across a whole lot of drives, then you can run into a bottleneck there.
Thanks Brent. I don't design SANs or NAS at all, but I like to have an understanding of the performance bottlenecks that can arise.
Rich
May 1, 2012 at 7:16 am
Good article altought was corrected by BrentO. I recently attended to at South Florida group meeting where Brent spoke about SSD. It would be interesting to see what should be the impact on RAID configuration on SSD.
Thanks Greg for posting this article and Brent for complementing it.
May 1, 2012 at 7:23 am
In my situation I have a table that is partioned into 12 partiones. So I have different file for each partioned group on the same lun going to the 5 disk stripped set. Would it make more sense to break it up into 12 logical luns. Would SQl server and Windows Server access faster because my SASS channel is 6g and I am only utilizing up to 1g (in monitoring it). This is a read only Data set.
May 1, 2012 at 8:09 am
jonalberghini (5/1/2012)
In my situation I have a table that is partitioned into 12 partitions. So I have different file for each portioned group on the same lun going to the 5 disk stripped set. Would it make more sense to break it up into 12 logical luns. Would SQl server and Windows Server access faster because my SASS channel is 6g and I am only utilizing up to 1g (in monitoring it). This is a read only Data set.
If you have 1 5-disk stripe with no redundancy at all I doubt your partitioning is gaining you anything. You are likely hitting your max reads on the disks as is. The stripe set can only operate 1 read at a time. You will have no parallel reads occurring. I would suggest you have separate stripe sets at that point to allow for parallel access, but that would depend on if the controller and pathing you use would also allow for the parallel access.
May 1, 2012 at 8:51 am
A questoin; out of curiosity:
You quoted some PerfMon stats on your production environment.
Given your analysis, would I be correct in assuming your produciton environment already runs an array of > 100 disks?
In fact given your preference for RAID 1+0, are you running > 200 disks?
No to both questions.
1) We arent using RAID 1+0 or 0+1. We are using a NAS via NFS that implements RAID DP
2) Our production environment Disk Queues during peak periods. Pretty excessively actually......
GAJ
Gregory A Jackson MBA, CSM
May 1, 2012 at 8:52 am
With the proliferation of SSD disk and its costs going down, some RAID levels with low write performances are no longer so low. Put an SSD in your life. I've done it, both home, laptop and work. Is the best money can buy right now in order to increase performance several degrees. We are planning to do it also on our local servers, first on the test server and then on the main one if evertyhing is ok.
SSDs are also a little bit of a controversial topic mostly due to their short lifespan.
It's certainly an option to consider though....I agree.
GAJ
Gregory A Jackson MBA, CSM
May 1, 2012 at 9:09 am
A few small nits to pick:
Raid 1+0 != Raid 0+1
If you have 8 drives in a Raid 1+0, say the 4 pairs are:
AE (mirror)
BF (mirror)
CG (mirror)
DH (mirror)
And then those 4 mirrors are striped.
In a Raid 0+1 set, you would have
ABCD(Striped)
EFGH(Striped)
and then those two stripe sets are mirrored.
You could lose 4 drives, say A,B,C, & D, and still have no data loss in either case.
The difference is, if you lose drive A in 1+0, a 2nd drive failure only has a 1 in 7 chance of causing data loss (if you lose drive E), in 0+1 you have a 4 in 7 chance of losing data (if you lose E,F,G, or H). If you have already lost 2 drives (say A & B) with no data lost, then 1+0 has a 2 in 6 chance of losing data on the 3rd failure, while the 0+1 has 4 in 6 chance.
If you scale it up to an array of 100 drives, the disparity is even worse, a 1 in 99 chance of losing data on your second drive failure, vs a 50 in 99 chance.
Raid 6 is effectively the same as Raid DP (there are some differences as to which blocks are being protected by each set of parity data, but the important part remains the same, you are protected against 2 drive failures).
Also depending on the level of hardware you are using, the write penalty of raid 5 will be largely invisible to the database due to a combination of cache and scheduling of the extra writes during a drives idle time.
Given arrays of 4/10/and 50 drives (with a 1% per year failure rate) you would expect to see the following failure rates for the different raid types over that year:
Raid 0: 4%/9.5%,39%
Raid 1: .000001%/10^-20/10^-100
Raid 5: .12%,.9%,24.5%
Raid 6: .0024%/.072%/11.76%
Raid 1+0: .02%/.05%/.25%
Raid 0+1: .04%/.24%/4.9%
Read Speed (assuming 100 operations per second on a single drive):
Raid 0: 400/1000/5000
Raid 1: 400/1000/5000
Raid 5: 300/900/4900
Raid 6: 200/800/4800
Raid 1+0: 400/1000/5000
Raid 0+1: 400/1000/5000
Write Speed:
Raid 0: 400/1000/5000
Raid 1: 100/100/100
Raid 5: 100/250/1250
Raid 6: 67/166/833
Raid 1+0: 200/500/2500
Raid 0+1: 200/500/2500
So I would say that the fault tolerance on 1 may be Excellent, and 1+0 may be good (and 0+1 worse than that), but depending on the number of disks, Raid 6/DP may no longer be much better than fair.
On read performance, 1+0 will be no better than 1, in fact they should all be pretty equal (given the same number of disks) except for 5/6 which will be slightly worse.
On write performance I would disagree that 1 is "good", the "write penalty" is only 2 in the case of a 2 disk set. So your write performance is the same as a single disk. Once you go above 4 disks (with raid 5) or 6 disks (in a raid 6 array), you will start to see better write performance than with a raid 1 array.
May 1, 2012 at 9:20 am
I agree short-stroking (using only a portion of the fastest area of the disk, leaving the rest entirely unallocated) can significantly improve spinning disk performance, because it reduces the normal (i.e. not calibrating) maximum head travel, which in turn reduces maximum seek times, which can reduce average seek times. This can cause lower RPM disks to perform as if they were high RPM.
I'm also disappointed that average seek was used throughout; even without using actual short-stroking, if you start with fresh drives dedicated to SQL data and/or log files, write properly sized files once (without growing them), it's very likely that all of the disk used will be at the "beginning" (outer edge), and will thus act as if short-stroked, with much lower average seek times.
Even without that, operations where the portions of the data files are close together can have the same effect.
Likewise, the article ignored sequential rates, which do matter - when you have a close enough approximation of a large scan of a large clustered or nonclustered index without fragmentation, in a filegroup whose files aren't (as a whole as well as individually) fragmented, on a contiguous allocation of dedicated spindles, which isn't in the buffer already, then sequential performance can become very important; this can particularly affect index maintenance and very large queries.
Note that a previous poster was incorrect regarding SSD's; RAID level can matter, and if RAID level doesn't matter (perhaps your controller or firmware or throughput path is the bottleneck), then a RAID level that provides more storage space at the same performance and a sufficient level of fault tolerance is clearly best.
The previous poster talking about a SAN with a "massive" cache may have a setup I haven't seen, but I'm hard pressed to find SAN's with more than 64GB of cache*, and even that's pretty rare; our SAN has about 16GB of cache (shared with all SAN clients), while our big SQL servers have over 100GB of buffer space.
For anyone that wants some raw data, here's a few 6 disk 3Gbps SATA local SSD setups on a modern controller in comparison to a few 10 disk 15k 4Gbps spindle SAN setups at various IOs outstanding - if you look closely, you'll see some odd effects. Every individual test was run for 1200 seconds to reduce cache effects.
RndOrSeq IOSizeKB ReadOrWrite IOsOutstanding IOsPerSec MBsPerSec DriveQty DriveRPM RAID RAIDCfg
Rnd 8 Read 2 11934 93 6 0 10 3x2 Stripe 128KB
Rnd 8 Read 2 11871 93 6 0 10 3x2 Stripe 64KB
Rnd 8 Read 2 11715 92 6 0 5 1x5+1 Stripe 64KB
Rnd 8 Read 2 11646 91 6 0 5 1x5+1 stripe 64KB
Rnd 8 Read 2 11630 91 6 0 5 1x5+1 Stripe 256KB
Rnd 8 Read 2 11702 91 6 0 5 1x5+1 Stripe 128KB
Rnd 8 Read 2 11649 91 6 0 0 1x6 Stripe 64KB
Rnd 8 Read 2 8251 64 6 0 10 3x2 Stripe 256KB
Rnd 8 Read 2 1086 8 10 15000 50 2x4+1
Rnd 8 Read 2 923 7 10 15000 5 1x9+1
Rnd 8 Read 2 915 7 10 15000 50 2x4+1
Rnd 8 Read 2 877 7 10 15000 10 2x5
Rnd 8 Read 8 37202 291 6 0 5 1x5+1 Stripe 128KB
Rnd 8 Read 8 37132 290 6 0 5 1x5+1 Stripe 64KB
Rnd 8 Read 8 36855 288 6 0 5 1x5+1 stripe 64KB
Rnd 8 Read 8 36740 287 6 0 0 1x6 Stripe 64KB
Rnd 8 Read 8 36578 286 6 0 5 1x5+1 Stripe 256KB
Rnd 8 Read 8 36315 284 6 0 10 3x2 Stripe 128KB
Rnd 8 Read 8 36257 283 6 0 10 3x2 Stripe 256KB
Rnd 8 Read 8 36104 282 6 0 10 3x2 Stripe 64KB
Rnd 8 Read 8 2578 20 10 15000 50 2x4+1
Rnd 8 Read 8 2079 16 10 15000 5 1x9+1
Rnd 8 Read 8 2061 16 10 15000 50 2x4+1
Rnd 8 Read 8 1886 15 10 15000 10 2x5
Rnd 8 Read 16 59804 467 6 0 5 1x5+1 Stripe 128KB
Rnd 8 Read 16 59372 464 6 0 5 1x5+1 Stripe 64KB
Rnd 8 Read 16 58830 460 6 0 5 1x5+1 stripe 64KB
Rnd 8 Read 16 58865 460 6 0 0 1x6 Stripe 64KB
Rnd 8 Read 16 58551 457 6 0 5 1x5+1 Stripe 256KB
Rnd 8 Read 16 57413 449 6 0 10 3x2 Stripe 256KB
Rnd 8 Read 16 57282 448 6 0 10 3x2 Stripe 128KB
Rnd 8 Read 16 56784 444 6 0 10 3x2 Stripe 64KB
Rnd 8 Read 16 2922 23 10 15000 50 2x4+1
Rnd 8 Read 16 2301 18 10 15000 50 2x4+1
Rnd 8 Read 16 2322 18 10 15000 5 1x9+1
Rnd 8 Read 16 2070 16 10 15000 10 2x5
Rnd 8 Read 32 85818 670 6 0 5 1x5+1 Stripe 128KB
Rnd 8 Read 32 84656 661 6 0 5 1x5+1 Stripe 256KB
Rnd 8 Read 32 82039 641 6 0 0 1x6 Stripe 64KB
Rnd 8 Read 32 81953 640 6 0 5 1x5+1 stripe 64KB
Rnd 8 Read 32 81301 635 6 0 5 1x5+1 Stripe 64KB
Rnd 8 Read 32 81049 633 6 0 10 3x2 Stripe 256KB
Rnd 8 Read 32 80516 629 6 0 10 3x2 Stripe 128KB
Rnd 8 Read 32 76704 599 6 0 10 3x2 Stripe 64KB
Rnd 8 Write 2 17867 140 6 0 0 1x6 Stripe 64KB
Rnd 8 Write 2 15790 123 6 0 10 3x2 Stripe 64KB
Rnd 8 Write 2 15733 123 6 0 10 3x2 Stripe 128KB
Rnd 8 Write 2 15359 120 6 0 10 3x2 Stripe 256KB
Rnd 8 Write 2 8445 66 6 0 5 1x5+1 stripe 64KB
Rnd 8 Write 2 7623 60 6 0 5 1x5+1 Stripe 64KB
Rnd 8 Write 2 7120 56 6 0 5 1x5+1 Stripe 128KB
Rnd 8 Write 2 6879 54 6 0 5 1x5+1 Stripe 256KB
Rnd 8 Write 2 2863 22 10 15000 10 2x5
Rnd 8 Write 2 1758 14 10 15000 5 1x9+1
Rnd 8 Write 2 1612 13 10 15000 50 2x4+1
Rnd 8 Write 2 1585 12 10 15000 50 2x4+1
Rnd 8 Write 8 19143 150 6 0 0 1x6 Stripe 64KB
Rnd 8 Write 8 17059 133 6 0 10 3x2 Stripe 128KB
Rnd 8 Write 8 17018 133 6 0 10 3x2 Stripe 64KB
Rnd 8 Write 8 16478 129 6 0 10 3x2 Stripe 256KB
Rnd 8 Write 8 8987 70 6 0 5 1x5+1 Stripe 128KB
Rnd 8 Write 8 8763 68 6 0 5 1x5+1 Stripe 256KB
Rnd 8 Write 8 8672 68 6 0 5 1x5+1 Stripe 64KB
Rnd 8 Write 8 8641 68 6 0 5 1x5+1 stripe 64KB
Rnd 8 Write 8 2950 23 10 15000 10 2x5
Rnd 8 Write 8 1805 14 10 15000 50 2x4+1
Rnd 8 Write 8 1769 14 10 15000 5 1x9+1
Rnd 8 Write 8 1636 13 10 15000 50 2x4+1
Rnd 8 Write 16 19268 151 6 0 0 1x6 Stripe 64KB
Rnd 8 Write 16 16728 131 6 0 10 3x2 Stripe 128KB
Rnd 8 Write 16 16551 129 6 0 10 3x2 Stripe 64KB
Rnd 8 Write 16 16352 128 6 0 10 3x2 Stripe 256KB
Rnd 8 Write 16 8815 69 6 0 5 1x5+1 Stripe 128KB
Rnd 8 Write 16 8608 67 6 0 5 1x5+1 Stripe 256KB
Rnd 8 Write 16 8519 67 6 0 5 1x5+1 Stripe 64KB
Rnd 8 Write 16 8478 66 6 0 5 1x5+1 stripe 64KB
Rnd 8 Write 16 3003 23 10 15000 10 2x5
Rnd 8 Write 16 1806 14 10 15000 50 2x4+1
Rnd 8 Write 16 1770 14 10 15000 5 1x9+1
Rnd 8 Write 16 1639 13 10 15000 50 2x4+1
Rnd 8 Write 32 18808 147 6 0 0 1x6 Stripe 64KB
Rnd 8 Write 32 16344 128 6 0 10 3x2 Stripe 128KB
Rnd 8 Write 32 16181 126 6 0 10 3x2 Stripe 64KB
Rnd 8 Write 32 15966 125 6 0 10 3x2 Stripe 256KB
Rnd 8 Write 32 8832 69 6 0 5 1x5+1 Stripe 128KB
Rnd 8 Write 32 8624 67 6 0 5 1x5+1 Stripe 256KB
Rnd 8 Write 32 8559 67 6 0 5 1x5+1 Stripe 64KB
Rnd 8 Write 32 8510 66 6 0 5 1x5+1 stripe 64KB
Rnd 64 Read 2 6957 435 6 0 10 3x2 Stripe 128KB
Rnd 64 Read 2 6916 432 6 0 10 3x2 Stripe 64KB
Rnd 64 Read 2 6910 432 6 0 10 3x2 Stripe 256KB
Rnd 64 Read 2 6858 429 6 0 5 1x5+1 Stripe 128KB
Rnd 64 Read 2 6833 427 6 0 5 1x5+1 Stripe 64KB
Rnd 64 Read 2 6806 425 6 0 5 1x5+1 stripe 64KB
Rnd 64 Read 2 6790 424 6 0 5 1x5+1 Stripe 256KB
Rnd 64 Read 2 6762 423 6 0 0 1x6 Stripe 64KB
Rnd 64 Read 2 971 61 10 15000 50 2x4+1
Rnd 64 Read 2 811 51 10 15000 5 1x9+1
Rnd 64 Read 2 812 51 10 15000 50 2x4+1
Rnd 64 Read 2 774 48 10 15000 10 2x5
Rnd 64 Read 8 18101 1131 6 0 5 1x5+1 Stripe 128KB
Rnd 64 Read 8 18072 1130 6 0 5 1x5+1 Stripe 64KB
Rnd 64 Read 8 18028 1127 6 0 10 3x2 Stripe 256KB
Rnd 64 Read 8 18027 1127 6 0 10 3x2 Stripe 64KB
Rnd 64 Read 8 18029 1127 6 0 5 1x5+1 stripe 64KB
Rnd 64 Read 8 18009 1126 6 0 10 3x2 Stripe 128KB
Rnd 64 Read 8 18009 1126 6 0 0 1x6 Stripe 64KB
Rnd 64 Read 8 17985 1124 6 0 5 1x5+1 Stripe 256KB
Rnd 64 Read 8 2233 140 10 15000 50 2x4+1
Rnd 64 Read 8 1831 114 10 15000 5 1x9+1
Rnd 64 Read 8 1809 113 10 15000 50 2x4+1
Rnd 64 Read 8 1694 106 10 15000 10 2x5
Rnd 64 Read 16 21602 1350 6 0 10 3x2 Stripe 256KB
Rnd 64 Read 16 21576 1349 6 0 10 3x2 Stripe 128KB
Rnd 64 Read 16 21590 1349 6 0 10 3x2 Stripe 64KB
Rnd 64 Read 16 21557 1347 6 0 5 1x5+1 Stripe 128KB
Rnd 64 Read 16 21521 1345 6 0 5 1x5+1 Stripe 256KB
Rnd 64 Read 16 21494 1343 6 0 0 1x6 Stripe 64KB
Rnd 64 Read 16 21471 1342 6 0 5 1x5+1 Stripe 64KB
Rnd 64 Read 16 21467 1342 6 0 5 1x5+1 stripe 64KB
Rnd 64 Read 16 2515 157 10 15000 50 2x4+1
Rnd 64 Read 16 2040 127 10 15000 5 1x9+1
Rnd 64 Read 16 2012 126 10 15000 50 2x4+1
Rnd 64 Read 16 1856 116 10 15000 10 2x5
Rnd 64 Read 32 22387 1399 6 0 10 3x2 Stripe 256KB
Rnd 64 Read 32 22382 1399 6 0 10 3x2 Stripe 64KB
Rnd 64 Read 32 22248 1390 6 0 10 3x2 Stripe 128KB
Rnd 64 Read 32 21742 1359 6 0 0 1x6 Stripe 64KB
Rnd 64 Read 32 21700 1356 6 0 5 1x5+1 Stripe 64KB
Rnd 64 Read 32 21676 1355 6 0 5 1x5+1 stripe 64KB
Rnd 64 Read 32 21682 1355 6 0 5 1x5+1 Stripe 256KB
Rnd 64 Read 32 21659 1354 6 0 5 1x5+1 Stripe 128KB
Rnd 64 Write 2 6191 387 6 0 0 1x6 Stripe 64KB
Rnd 64 Write 2 3382 211 6 0 10 3x2 Stripe 64KB
Rnd 64 Write 2 3365 210 6 0 10 3x2 Stripe 256KB
Rnd 64 Write 2 3350 209 6 0 10 3x2 Stripe 128KB
Rnd 64 Write 2 2017 126 6 0 5 1x5+1 Stripe 64KB
Rnd 64 Write 2 1974 123 6 0 5 1x5+1 stripe 64KB
Rnd 64 Write 2 1551 97 6 0 5 1x5+1 Stripe 256KB
Rnd 64 Write 2 1537 96 6 0 5 1x5+1 Stripe 128KB
Rnd 64 Write 2 1452 91 10 15000 10 2x5
Rnd 64 Write 2 832 52 10 15000 50 2x4+1
Rnd 64 Write 2 788 49 10 15000 50 2x4+1
Rnd 64 Write 2 750 47 10 15000 5 1x9+1
Rnd 64 Write 8 6265 392 6 0 0 1x6 Stripe 64KB
Rnd 64 Write 8 3392 212 6 0 10 3x2 Stripe 256KB
Rnd 64 Write 8 3370 211 6 0 10 3x2 Stripe 64KB
Rnd 64 Write 8 3359 210 6 0 10 3x2 Stripe 128KB
Rnd 64 Write 8 2066 129 6 0 5 1x5+1 Stripe 256KB
Rnd 64 Write 8 2052 128 6 0 5 1x5+1 Stripe 128KB
Rnd 64 Write 8 2021 126 6 0 5 1x5+1 Stripe 64KB
Rnd 64 Write 8 1972 123 6 0 5 1x5+1 stripe 64KB
Rnd 64 Write 8 1488 93 10 15000 10 2x5
Rnd 64 Write 8 875 55 10 15000 50 2x4+1
Rnd 64 Write 8 800 50 10 15000 50 2x4+1
Rnd 64 Write 8 759 47 10 15000 5 1x9+1
Rnd 64 Write 16 6235 390 6 0 0 1x6 Stripe 64KB
Rnd 64 Write 16 3398 212 6 0 10 3x2 Stripe 256KB
Rnd 64 Write 16 3387 212 6 0 10 3x2 Stripe 64KB
Rnd 64 Write 16 3350 209 6 0 10 3x2 Stripe 128KB
Rnd 64 Write 16 2063 129 6 0 5 1x5+1 Stripe 128KB
Rnd 64 Write 16 2062 129 6 0 5 1x5+1 Stripe 256KB
Rnd 64 Write 16 2027 127 6 0 5 1x5+1 Stripe 64KB
Rnd 64 Write 16 1975 123 6 0 5 1x5+1 stripe 64KB
Rnd 64 Write 16 1505 94 10 15000 10 2x5
Rnd 64 Write 16 882 55 10 15000 50 2x4+1
Rnd 64 Write 16 805 50 10 15000 50 2x4+1
Rnd 64 Write 16 761 48 10 15000 5 1x9+1
Rnd 64 Write 32 6186 387 6 0 0 1x6 Stripe 64KB
Rnd 64 Write 32 3381 211 6 0 10 3x2 Stripe 256KB
Rnd 64 Write 32 3369 211 6 0 10 3x2 Stripe 64KB
Rnd 64 Write 32 3354 210 6 0 10 3x2 Stripe 128KB
Rnd 64 Write 32 2064 129 6 0 5 1x5+1 Stripe 256KB
Rnd 64 Write 32 2055 128 6 0 5 1x5+1 Stripe 128KB
Rnd 64 Write 32 2006 125 6 0 5 1x5+1 Stripe 64KB
Rnd 64 Write 32 1983 124 6 0 5 1x5+1 stripe 64KB
Seq 8 Read 2 63537 496 6 0 5 1x5+1 Stripe 128KB
Seq 8 Read 2 62511 488 6 0 5 1x5+1 Stripe 256KB
Seq 8 Read 2 55142 431 6 0 5 1x5+1 Stripe 64KB
Seq 8 Read 2 53876 421 6 0 5 1x5+1 stripe 64KB
Seq 8 Read 2 53696 419 6 0 10 3x2 Stripe 256KB
Seq 8 Read 2 51947 406 6 0 0 1x6 Stripe 64KB
Seq 8 Read 2 46403 363 6 0 10 3x2 Stripe 128KB
Seq 8 Read 2 38272 299 6 0 10 3x2 Stripe 64KB
Seq 8 Read 2 18329 143 10 15000 50 2x4+1
Seq 8 Read 2 16474 129 10 15000 5 1x9+1
Seq 8 Read 2 16387 128 10 15000 50 2x4+1
Seq 8 Read 2 14990 117 10 15000 10 2x5
Seq 8 Read 8 83623 653 6 0 5 1x5+1 Stripe 256KB
Seq 8 Read 8 83616 653 6 0 5 1x5+1 Stripe 128KB
Seq 8 Read 8 75914 593 6 0 10 3x2 Stripe 256KB
Seq 8 Read 8 73740 576 6 0 5 1x5+1 Stripe 64KB
Seq 8 Read 8 69832 546 6 0 5 1x5+1 stripe 64KB
Seq 8 Read 8 69000 539 6 0 0 1x6 Stripe 64KB
Seq 8 Read 8 65375 511 6 0 10 3x2 Stripe 128KB
Seq 8 Read 8 53096 415 6 0 10 3x2 Stripe 64KB
Seq 8 Read 8 26483 207 10 15000 50 2x4+1
Seq 8 Read 8 20666 161 10 15000 5 1x9+1
Seq 8 Read 8 20591 161 10 15000 50 2x4+1
Seq 8 Read 8 19571 153 10 15000 10 2x5
Seq 8 Read 16 96430 753 6 0 5 1x5+1 Stripe 256KB
Seq 8 Read 16 90790 709 6 0 5 1x5+1 Stripe 128KB
Seq 8 Read 16 85029 664 6 0 10 3x2 Stripe 256KB
Seq 8 Read 16 78333 612 6 0 0 1x6 Stripe 64KB
Seq 8 Read 16 75840 592 6 0 10 3x2 Stripe 128KB
Seq 8 Read 16 67638 528 6 0 10 3x2 Stripe 64KB
Seq 8 Read 16 49328 385 6 0 5 1x5+1 Stripe 64KB
Seq 8 Read 16 48821 381 6 0 5 1x5+1 stripe 64KB
Seq 8 Read 16 29076 227 10 15000 50 2x4+1
Seq 8 Read 16 24661 193 10 15000 5 1x9+1
Seq 8 Read 16 22651 177 10 15000 50 2x4+1
Seq 8 Read 16 21718 170 10 15000 10 2x5
Seq 8 Read 32 100070 782 6 0 10 3x2 Stripe 64KB
Seq 8 Read 32 93688 732 6 0 5 1x5+1 Stripe 256KB
Seq 8 Read 32 87197 681 6 0 10 3x2 Stripe 256KB
Seq 8 Read 32 84965 664 6 0 10 3x2 Stripe 128KB
Seq 8 Read 32 70597 552 6 0 0 1x6 Stripe 64KB
Seq 8 Read 32 69834 546 6 0 5 1x5+1 Stripe 128KB
Seq 8 Read 32 68631 536 6 0 5 1x5+1 stripe 64KB
Seq 8 Read 32 68400 534 6 0 5 1x5+1 Stripe 64KB
Seq 8 Write 2 58992 461 6 0 5 1x5+1 Stripe 256KB
Seq 8 Write 2 57796 452 6 0 5 1x5+1 Stripe 128KB
Seq 8 Write 2 53162 415 6 0 5 1x5+1 Stripe 64KB
Seq 8 Write 2 51271 401 6 0 5 1x5+1 stripe 64KB
Seq 8 Write 2 49876 390 6 0 0 1x6 Stripe 64KB
Seq 8 Write 2 47943 375 6 0 10 3x2 Stripe 128KB
Seq 8 Write 2 44568 348 6 0 10 3x2 Stripe 256KB
Seq 8 Write 2 41895 327 6 0 10 3x2 Stripe 64KB
Seq 8 Write 2 14098 110 10 15000 50 2x4+1
Seq 8 Write 2 13629 106 10 15000 5 1x9+1
Seq 8 Write 2 11772 92 10 15000 50 2x4+1
Seq 8 Write 2 10680 83 10 15000 10 2x5
Seq 8 Write 8 86867 679 6 0 5 1x5+1 Stripe 256KB
Seq 8 Write 8 81534 637 6 0 5 1x5+1 Stripe 128KB
Seq 8 Write 8 68948 539 6 0 10 3x2 Stripe 256KB
Seq 8 Write 8 68352 534 6 0 5 1x5+1 Stripe 64KB
Seq 8 Write 8 68197 533 6 0 0 1x6 Stripe 64KB
Seq 8 Write 8 66822 522 6 0 5 1x5+1 stripe 64KB
Seq 8 Write 8 65318 510 6 0 10 3x2 Stripe 128KB
Seq 8 Write 8 54262 424 6 0 10 3x2 Stripe 64KB
Seq 8 Write 8 20246 158 10 15000 50 2x4+1
Seq 8 Write 8 20125 157 10 15000 50 2x4+1
Seq 8 Write 8 19850 155 10 15000 10 2x5
Seq 8 Write 8 19716 154 10 15000 5 1x9+1
Seq 8 Write 16 97379 761 6 0 5 1x5+1 Stripe 256KB
Seq 8 Write 16 88641 693 6 0 5 1x5+1 Stripe 128KB
Seq 8 Write 16 77760 608 6 0 0 1x6 Stripe 64KB
Seq 8 Write 16 73964 578 6 0 5 1x5+1 stripe 64KB
Seq 8 Write 16 73981 578 6 0 5 1x5+1 Stripe 64KB
Seq 8 Write 16 70057 547 6 0 10 3x2 Stripe 256KB
Seq 8 Write 16 69929 546 6 0 10 3x2 Stripe 128KB
Seq 8 Write 16 60694 474 6 0 10 3x2 Stripe 64KB
Seq 8 Write 16 22823 178 10 15000 50 2x4+1
Seq 8 Write 16 22470 176 10 15000 50 2x4+1
Seq 8 Write 16 22091 173 10 15000 5 1x9+1
Seq 8 Write 16 22148 173 10 15000 10 2x5
Seq 8 Write 32 98539 770 6 0 5 1x5+1 Stripe 256KB
Seq 8 Write 32 88283 690 6 0 5 1x5+1 Stripe 128KB
Seq 8 Write 32 79104 618 6 0 0 1x6 Stripe 64KB
Seq 8 Write 32 77732 607 6 0 5 1x5+1 Stripe 64KB
Seq 8 Write 32 74770 584 6 0 5 1x5+1 stripe 64KB
Seq 8 Write 32 70591 551 6 0 10 3x2 Stripe 128KB
Seq 8 Write 32 70463 550 6 0 10 3x2 Stripe 256KB
Seq 8 Write 32 63217 494 6 0 10 3x2 Stripe 64KB
Seq 64 Read 2 23762 1485 6 0 5 1x5+1 Stripe 128KB
Seq 64 Read 2 22331 1396 6 0 0 1x6 Stripe 64KB
Seq 64 Read 2 21762 1360 6 0 5 1x5+1 Stripe 256KB
Seq 64 Read 2 19893 1243 6 0 10 3x2 Stripe 256KB
Seq 64 Read 2 18004 1125 6 0 10 3x2 Stripe 128KB
Seq 64 Read 2 9156 572 6 0 10 3x2 Stripe 64KB
Seq 64 Read 2 8872 555 6 0 5 1x5+1 stripe 64KB
Seq 64 Read 2 8453 528 6 0 5 1x5+1 Stripe 64KB
Seq 64 Read 2 5147 322 10 15000 50 2x4+1
Seq 64 Read 2 3805 238 10 15000 50 2x4+1
Seq 64 Read 2 3647 228 10 15000 5 1x9+1
Seq 64 Read 2 3433 215 10 15000 10 2x5
Seq 64 Read 8 22451 1403 6 0 10 3x2 Stripe 64KB
Seq 64 Read 8 21797 1362 6 0 10 3x2 Stripe 128KB
Seq 64 Read 8 21441 1340 6 0 10 3x2 Stripe 256KB
Seq 64 Read 8 20880 1305 6 0 5 1x5+1 Stripe 128KB
Seq 64 Read 8 20798 1300 6 0 5 1x5+1 Stripe 64KB
Seq 64 Read 8 20748 1297 6 0 5 1x5+1 stripe 64KB
Seq 64 Read 8 20738 1296 6 0 0 1x6 Stripe 64KB
Seq 64 Read 8 20145 1259 6 0 5 1x5+1 Stripe 256KB
Seq 64 Read 8 5874 367 10 15000 50 2x4+1
Seq 64 Read 8 4584 286 10 15000 50 2x4+1
Seq 64 Read 8 4480 280 10 15000 5 1x9+1
Seq 64 Read 8 4012 251 10 15000 10 2x5
Seq 64 Read 16 23266 1454 6 0 10 3x2 Stripe 128KB
Seq 64 Read 16 23250 1453 6 0 10 3x2 Stripe 64KB
Seq 64 Read 16 23254 1453 6 0 10 3x2 Stripe 256KB
Seq 64 Read 16 22616 1414 6 0 5 1x5+1 stripe 64KB
Seq 64 Read 16 22621 1414 6 0 5 1x5+1 Stripe 256KB
Seq 64 Read 16 22576 1411 6 0 5 1x5+1 Stripe 128KB
Seq 64 Read 16 22570 1411 6 0 0 1x6 Stripe 64KB
Seq 64 Read 16 22536 1409 6 0 5 1x5+1 Stripe 64KB
Seq 64 Read 16 5901 369 10 15000 50 2x4+1
Seq 64 Read 16 4934 308 10 15000 5 1x9+1
Seq 64 Read 16 4884 305 10 15000 50 2x4+1
Seq 64 Read 16 4029 252 10 15000 10 2x5
Seq 64 Read 32 22749 1422 6 0 10 3x2 Stripe 256KB
Seq 64 Read 32 22733 1421 6 0 10 3x2 Stripe 64KB
Seq 64 Read 32 22708 1419 6 0 10 3x2 Stripe 128KB
Seq 64 Read 32 21771 1361 6 0 5 1x5+1 Stripe 64KB
Seq 64 Read 32 21768 1360 6 0 5 1x5+1 stripe 64KB
Seq 64 Read 32 21766 1360 6 0 0 1x6 Stripe 64KB
Seq 64 Read 32 21765 1360 6 0 5 1x5+1 Stripe 128KB
Seq 64 Read 32 21750 1359 6 0 5 1x5+1 Stripe 256KB
Seq 64 Write 2 17185 1074 6 0 0 1x6 Stripe 64KB
Seq 64 Write 2 14778 924 6 0 5 1x5+1 Stripe 128KB
Seq 64 Write 2 14645 915 6 0 5 1x5+1 Stripe 64KB
Seq 64 Write 2 14591 912 6 0 5 1x5+1 stripe 64KB
Seq 64 Write 2 14456 903 6 0 5 1x5+1 Stripe 256KB
Seq 64 Write 2 8956 560 6 0 10 3x2 Stripe 128KB
Seq 64 Write 2 8873 555 6 0 10 3x2 Stripe 64KB
Seq 64 Write 2 8651 541 6 0 10 3x2 Stripe 256KB
Seq 64 Write 2 3906 244 10 15000 50 2x4+1
Seq 64 Write 2 3481 218 10 15000 50 2x4+1
Seq 64 Write 2 2920 182 10 15000 10 2x5
Seq 64 Write 2 2747 172 10 15000 5 1x9+1
Seq 64 Write 8 17727 1108 6 0 0 1x6 Stripe 64KB
Seq 64 Write 8 14799 925 6 0 5 1x5+1 Stripe 128KB
Seq 64 Write 8 14753 922 6 0 5 1x5+1 Stripe 64KB
Seq 64 Write 8 14671 917 6 0 5 1x5+1 stripe 64KB
Seq 64 Write 8 14418 901 6 0 5 1x5+1 Stripe 256KB
Seq 64 Write 8 8888 556 6 0 10 3x2 Stripe 64KB
Seq 64 Write 8 8902 556 6 0 10 3x2 Stripe 128KB
Seq 64 Write 8 8648 541 6 0 10 3x2 Stripe 256KB
Seq 64 Write 8 4651 291 10 15000 50 2x4+1
Seq 64 Write 8 4564 285 10 15000 50 2x4+1
Seq 64 Write 8 2928 183 10 15000 10 2x5
Seq 64 Write 8 2645 165 10 15000 5 1x9+1
Seq 64 Write 16 17699 1106 6 0 0 1x6 Stripe 64KB
Seq 64 Write 16 14775 923 6 0 5 1x5+1 Stripe 128KB
Seq 64 Write 16 14755 922 6 0 5 1x5+1 Stripe 64KB
Seq 64 Write 16 14702 919 6 0 5 1x5+1 stripe 64KB
Seq 64 Write 16 14398 900 6 0 5 1x5+1 Stripe 256KB
Seq 64 Write 16 8894 556 6 0 10 3x2 Stripe 64KB
Seq 64 Write 16 8874 555 6 0 10 3x2 Stripe 128KB
Seq 64 Write 16 8794 550 6 0 10 3x2 Stripe 256KB
Seq 64 Write 16 4857 304 10 15000 50 2x4+1
Seq 64 Write 16 4778 299 10 15000 50 2x4+1
Seq 64 Write 16 2958 185 10 15000 10 2x5
Seq 64 Write 16 2757 172 10 15000 5 1x9+1
Seq 64 Write 32 17713 1107 6 0 0 1x6 Stripe 64KB
Seq 64 Write 32 14806 925 6 0 5 1x5+1 Stripe 128KB
Seq 64 Write 32 14747 922 6 0 5 1x5+1 Stripe 64KB
Seq 64 Write 32 14710 919 6 0 5 1x5+1 stripe 64KB
Seq 64 Write 32 14488 905 6 0 5 1x5+1 Stripe 256KB
Seq 64 Write 32 8916 557 6 0 10 3x2 Stripe 128KB
Seq 64 Write 32 8873 555 6 0 10 3x2 Stripe 64KB
Seq 64 Write 32 8764 548 6 0 10 3x2 Stripe 256KB
Seq 256 Read 2 5435 1359 6 0 5 1x5+1 Stripe 256KB
Seq 256 Read 2 4991 1248 6 0 10 3x2 Stripe 64KB
Seq 256 Read 2 4841 1210 6 0 5 1x5+1 Stripe 64KB
Seq 256 Read 2 4785 1196 6 0 0 1x6 Stripe 64KB
Seq 256 Read 2 4762 1190 6 0 5 1x5+1 stripe 64KB
Seq 256 Read 2 4283 1071 6 0 10 3x2 Stripe 128KB
Seq 256 Read 2 4095 1024 6 0 10 3x2 Stripe 256KB
Seq 256 Read 2 3862 966 6 0 5 1x5+1 Stripe 128KB
Seq 256 Read 2 1499 375 10 15000 50 2x4+1
Seq 256 Read 2 1405 351 10 15000 50 2x4+1
Seq 256 Read 2 1137 284 10 15000 5 1x9+1
Seq 256 Read 2 1080 270 10 15000 10 2x5
Seq 256 Read 8 5824 1456 6 0 10 3x2 Stripe 64KB
Seq 256 Read 8 5768 1442 6 0 10 3x2 Stripe 128KB
Seq 256 Read 8 5757 1439 6 0 10 3x2 Stripe 256KB
Seq 256 Read 8 5604 1401 6 0 5 1x5+1 stripe 64KB
Seq 256 Read 8 5587 1397 6 0 0 1x6 Stripe 64KB
Seq 256 Read 8 5582 1396 6 0 5 1x5+1 Stripe 64KB
Seq 256 Read 8 5454 1364 6 0 5 1x5+1 Stripe 128KB
Seq 256 Read 8 5449 1362 6 0 5 1x5+1 Stripe 256KB
Seq 256 Read 8 1517 379 10 15000 50 2x4+1
Seq 256 Read 8 1500 375 10 15000 50 2x4+1
Seq 256 Read 8 1364 341 10 15000 5 1x9+1
Seq 256 Read 8 1275 319 10 15000 10 2x5
Seq 256 Read 16 5812 1453 6 0 10 3x2 Stripe 64KB
Seq 256 Read 16 5759 1440 6 0 10 3x2 Stripe 256KB
Seq 256 Read 16 5752 1438 6 0 10 3x2 Stripe 128KB
Seq 256 Read 16 5448 1362 6 0 5 1x5+1 Stripe 256KB
Seq 256 Read 16 5446 1361 6 0 5 1x5+1 Stripe 128KB
Seq 256 Read 16 5437 1359 6 0 5 1x5+1 Stripe 64KB
Seq 256 Read 16 5434 1359 6 0 5 1x5+1 stripe 64KB
Seq 256 Read 16 5432 1358 6 0 0 1x6 Stripe 64KB
Seq 256 Read 16 1518 380 10 15000 50 2x4+1
Seq 256 Read 16 1507 377 10 15000 50 2x4+1
Seq 256 Read 16 1471 368 10 15000 5 1x9+1
Seq 256 Read 16 1375 344 10 15000 10 2x5
Seq 256 Read 32 5765 1441 6 0 10 3x2 Stripe 64KB
Seq 256 Read 32 5755 1439 6 0 10 3x2 Stripe 128KB
Seq 256 Read 32 5632 1408 6 0 10 3x2 Stripe 256KB
Seq 256 Read 32 5448 1362 6 0 5 1x5+1 stripe 64KB
Seq 256 Read 32 5438 1360 6 0 5 1x5+1 Stripe 64KB
Seq 256 Read 32 5439 1360 6 0 0 1x6 Stripe 64KB
Seq 256 Read 32 5441 1360 6 0 5 1x5+1 Stripe 128KB
Seq 256 Read 32 5434 1359 6 0 5 1x5+1 Stripe 256KB
Seq 256 Write 2 4429 1107 6 0 0 1x6 Stripe 64KB
Seq 256 Write 2 3702 925 6 0 5 1x5+1 Stripe 128KB
Seq 256 Write 2 3687 922 6 0 5 1x5+1 Stripe 64KB
Seq 256 Write 2 3668 917 6 0 5 1x5+1 stripe 64KB
Seq 256 Write 2 3621 905 6 0 5 1x5+1 Stripe 256KB
Seq 256 Write 2 2236 559 6 0 10 3x2 Stripe 128KB
Seq 256 Write 2 2220 555 6 0 10 3x2 Stripe 64KB
Seq 256 Write 2 2200 550 6 0 10 3x2 Stripe 256KB
Seq 256 Write 2 1151 288 10 15000 50 2x4+1
Seq 256 Write 2 1054 264 10 15000 50 2x4+1
Seq 256 Write 2 732 183 10 15000 10 2x5
Seq 256 Write 2 642 160 10 15000 5 1x9+1
Seq 256 Write 8 4429 1107 6 0 0 1x6 Stripe 64KB
Seq 256 Write 8 3705 926 6 0 5 1x5+1 Stripe 128KB
Seq 256 Write 8 3690 923 6 0 5 1x5+1 Stripe 64KB
Seq 256 Write 8 3675 919 6 0 5 1x5+1 stripe 64KB
Seq 256 Write 8 3608 902 6 0 5 1x5+1 Stripe 256KB
Seq 256 Write 8 2234 559 6 0 10 3x2 Stripe 128KB
Seq 256 Write 8 2217 554 6 0 10 3x2 Stripe 64KB
Seq 256 Write 8 2210 552 6 0 10 3x2 Stripe 256KB
Seq 256 Write 8 1312 328 10 15000 50 2x4+1
Seq 256 Write 8 1194 299 10 15000 50 2x4+1
Seq 256 Write 8 744 186 10 15000 5 1x9+1
Seq 256 Write 8 738 184 10 15000 10 2x5
Seq 256 Write 16 4423 1106 6 0 0 1x6 Stripe 64KB
Seq 256 Write 16 3697 924 6 0 5 1x5+1 Stripe 128KB
Seq 256 Write 16 3688 922 6 0 5 1x5+1 Stripe 64KB
Seq 256 Write 16 3674 919 6 0 5 1x5+1 stripe 64KB
Seq 256 Write 16 3605 901 6 0 5 1x5+1 Stripe 256KB
Seq 256 Write 16 2232 558 6 0 10 3x2 Stripe 128KB
Seq 256 Write 16 2218 554 6 0 10 3x2 Stripe 64KB
Seq 256 Write 16 2214 553 6 0 10 3x2 Stripe 256KB
Seq 256 Write 16 1313 328 10 15000 50 2x4+1
Seq 256 Write 16 1191 298 10 15000 50 2x4+1
Seq 256 Write 16 745 186 10 15000 5 1x9+1
Seq 256 Write 16 740 185 10 15000 10 2x5
Seq 256 Write 32 4423 1106 6 0 0 1x6 Stripe 64KB
Seq 256 Write 32 3712 928 6 0 5 1x5+1 Stripe 128KB
Seq 256 Write 32 3686 922 6 0 5 1x5+1 Stripe 64KB
Seq 256 Write 32 3683 921 6 0 5 1x5+1 stripe 64KB
Seq 256 Write 32 3630 907 6 0 5 1x5+1 Stripe 256KB
Seq 256 Write 32 2237 559 6 0 10 3x2 Stripe 128KB
Seq 256 Write 32 2216 554 6 0 10 3x2 Stripe 64KB
Seq 256 Write 32 2191 548 6 0 10 3x2 Stripe 256KB
Seq 1024 Read 2 1484 1484 6 0 5 1x5+1 Stripe 128KB
Seq 1024 Read 2 1481 1481 6 0 5 1x5+1 Stripe 256KB
Seq 1024 Read 2 1468 1468 6 0 10 3x2 Stripe 128KB
Seq 1024 Read 2 1433 1433 6 0 10 3x2 Stripe 256KB
Seq 1024 Read 2 1409 1409 6 0 5 1x5+1 stripe 64KB
Seq 1024 Read 2 1385 1385 6 0 5 1x5+1 Stripe 64KB
Seq 1024 Read 2 1367 1367 6 0 0 1x6 Stripe 64KB
Seq 1024 Read 2 1323 1323 6 0 10 3x2 Stripe 64KB
Seq 1024 Read 2 379 379 10 15000 50 2x4+1
Seq 1024 Read 2 373 373 10 15000 50 2x4+1
Seq 1024 Read 2 355 355 10 15000 5 1x9+1
Seq 1024 Read 2 331 331 10 15000 10 2x5
Seq 1024 Read 8 1448 1448 6 0 10 3x2 Stripe 128KB
Seq 1024 Read 8 1444 1444 6 0 10 3x2 Stripe 256KB
Seq 1024 Read 8 1433 1433 6 0 5 1x5+1 Stripe 64KB
Seq 1024 Read 8 1432 1432 6 0 5 1x5+1 stripe 64KB
Seq 1024 Read 8 1397 1397 6 0 0 1x6 Stripe 64KB
Seq 1024 Read 8 1378 1378 6 0 10 3x2 Stripe 64KB
Seq 1024 Read 8 1376 1376 6 0 5 1x5+1 Stripe 128KB
Seq 1024 Read 8 1360 1360 6 0 5 1x5+1 Stripe 256KB
Seq 1024 Read 8 381 381 10 15000 50 2x4+1
Seq 1024 Read 8 309 309 10 15000 5 1x9+1
Seq 1024 Read 8 305 305 10 15000 50 2x4+1
Seq 1024 Read 8 266 266 10 15000 10 2x5
Seq 1024 Read 16 1433 1433 6 0 5 1x5+1 stripe 64KB
Seq 1024 Read 16 1430 1430 6 0 5 1x5+1 Stripe 64KB
Seq 1024 Read 16 1424 1424 6 0 0 1x6 Stripe 64KB
Seq 1024 Read 16 1392 1392 6 0 5 1x5+1 Stripe 128KB
Seq 1024 Read 16 1370 1370 6 0 10 3x2 Stripe 256KB
Seq 1024 Read 16 1367 1367 6 0 10 3x2 Stripe 128KB
Seq 1024 Read 16 1361 1361 6 0 5 1x5+1 Stripe 256KB
Seq 1024 Read 16 1342 1342 6 0 10 3x2 Stripe 64KB
Seq 1024 Read 16 381 381 10 15000 50 2x4+1
Seq 1024 Read 16 308 308 10 15000 50 2x4+1
Seq 1024 Read 16 297 297 10 15000 5 1x9+1
Seq 1024 Read 16 264 264 10 15000 10 2x5
Seq 1024 Read 32 1431 1431 6 0 0 1x6 Stripe 64KB
Seq 1024 Read 32 1427 1427 6 0 5 1x5+1 stripe 64KB
Seq 1024 Read 32 1414 1414 6 0 5 1x5+1 Stripe 64KB
Seq 1024 Read 32 1393 1393 6 0 5 1x5+1 Stripe 128KB
Seq 1024 Read 32 1372 1372 6 0 10 3x2 Stripe 256KB
Seq 1024 Read 32 1369 1369 6 0 10 3x2 Stripe 128KB
Seq 1024 Read 32 1359 1359 6 0 5 1x5+1 Stripe 256KB
Seq 1024 Read 32 1339 1339 6 0 10 3x2 Stripe 64KB
Seq 1024 Write 2 1103 1103 6 0 0 1x6 Stripe 64KB
Seq 1024 Write 2 925 925 6 0 5 1x5+1 Stripe 128KB
Seq 1024 Write 2 922 922 6 0 5 1x5+1 Stripe 64KB
Seq 1024 Write 2 918 918 6 0 5 1x5+1 stripe 64KB
Seq 1024 Write 2 918 918 6 0 5 1x5+1 Stripe 256KB
Seq 1024 Write 2 559 559 6 0 10 3x2 Stripe 128KB
Seq 1024 Write 2 554 554 6 0 10 3x2 Stripe 64KB
Seq 1024 Write 2 544 544 6 0 10 3x2 Stripe 256KB
Seq 1024 Write 2 298 298 10 15000 50 2x4+1
Seq 1024 Write 2 284 284 10 15000 50 2x4+1
Seq 1024 Write 2 184 184 10 15000 10 2x5
Seq 1024 Write 2 179 179 10 15000 5 1x9+1
Seq 1024 Write 8 1105 1105 6 0 0 1x6 Stripe 64KB
Seq 1024 Write 8 927 927 6 0 5 1x5+1 Stripe 128KB
Seq 1024 Write 8 921 921 6 0 5 1x5+1 Stripe 64KB
Seq 1024 Write 8 919 919 6 0 5 1x5+1 stripe 64KB
Seq 1024 Write 8 913 913 6 0 5 1x5+1 Stripe 256KB
Seq 1024 Write 8 558 558 6 0 10 3x2 Stripe 128KB
Seq 1024 Write 8 555 555 6 0 10 3x2 Stripe 256KB
Seq 1024 Write 8 554 554 6 0 10 3x2 Stripe 64KB
Seq 1024 Write 8 321 321 10 15000 50 2x4+1
Seq 1024 Write 8 298 298 10 15000 50 2x4+1
Seq 1024 Write 8 185 185 10 15000 10 2x5
Seq 1024 Write 8 182 182 10 15000 5 1x9+1
Seq 1024 Write 16 1103 1103 6 0 0 1x6 Stripe 64KB
Seq 1024 Write 16 927 927 6 0 5 1x5+1 Stripe 128KB
Seq 1024 Write 16 921 921 6 0 5 1x5+1 Stripe 64KB
Seq 1024 Write 16 919 919 6 0 5 1x5+1 stripe 64KB
Seq 1024 Write 16 903 903 6 0 5 1x5+1 Stripe 256KB
Seq 1024 Write 16 559 559 6 0 10 3x2 Stripe 128KB
Seq 1024 Write 16 554 554 6 0 10 3x2 Stripe 64KB
Seq 1024 Write 16 554 554 6 0 10 3x2 Stripe 256KB
Seq 1024 Write 16 320 320 10 15000 50 2x4+1
Seq 1024 Write 16 298 298 10 15000 50 2x4+1
Seq 1024 Write 16 185 185 10 15000 10 2x5
Seq 1024 Write 16 182 182 10 15000 5 1x9+1
Seq 1024 Write 32 1107 1107 6 0 0 1x6 Stripe 64KB
Seq 1024 Write 32 928 928 6 0 5 1x5+1 Stripe 128KB
Seq 1024 Write 32 921 921 6 0 5 1x5+1 Stripe 64KB
Seq 1024 Write 32 919 919 6 0 5 1x5+1 stripe 64KB
Seq 1024 Write 32 914 914 6 0 5 1x5+1 Stripe 256KB
Seq 1024 Write 32 560 560 6 0 10 3x2 Stripe 128KB
Seq 1024 Write 32 555 555 6 0 10 3x2 Stripe 64KB
Seq 1024 Write 32 552 552 6 0 10 3x2 Stripe 256KB
*I mean real RAM, not SSD's used as cache.
ETA: The two spindle RAID 50's results are because the faster ones are from 300GB disks, and the slower ones are from 146GB disks.
May 1, 2012 at 9:25 am
yazalpizar_ (5/1/2012)
With the proliferation of SSD disk and its costs going down, some RAID levels with low write performances are no longer so low. Put an SSD in your life. I've done it, both home, laptop and work. Is the best money can buy right now in order to increase performance several degrees. We are planning to do it also on our local servers, first on the test server and then on the main one if evertyhing is ok.
Make sure you have good backups and be careful here. When SSDs die, they die ,and you may or may not get good notice.
A quick read: http://www.codinghorror.com/blog/2011/05/the-hot-crazy-solid-state-drive-scale.html
Follow me on Twitter: http://www.twitter.com/way0utwest
Forum Etiquette: How to post data/code on a forum to get the best help
My Blog: www.voiceofthedba.com
May 1, 2012 at 9:34 am
Good article!
I would caution against using the term LUN when referring to anything other than SAN or NAS storage. I've never heard that term used for locally attached disks.
As mentioned in a previous post or two RAID 0+1 and RAID 1+0 are quite different and I would caution any SQL DBA against using RAID 0+1. The risk of data loss is too great and you increase the chance that you will get a 2:00AM call.
I have several clustered instances utilizing a 123 TB 3Par SAN with 4 GB HBA dual port cards. In working with 3Par they claim to be consistently able to achieve > 53,000 IOPS. Of course this is a $3 million SAN but it gives an idea of what level of service can be provided. And that is with fiber channel drives. It gets even faster if you use SSD or a combination of fiber channel and SSD. For the most critical data it can be stored on the outer edge of the disks allowing for even faster seek times.
May 1, 2012 at 9:43 am
I don't see ever getting on board with having a production database running on a VM. Of course, things change but I cringe at the thought of this.
Cheers
May 1, 2012 at 9:52 am
Steve Jones - SSC Editor (5/1/2012)
yazalpizar_ (5/1/2012)
With the proliferation of SSD disk and its costs going down, some RAID levels with low write performances are no longer so low. Put an SSD in your life. I've done it, both home, laptop and work. Is the best money can buy right now in order to increase performance several degrees. We are planning to do it also on our local servers, first on the test server and then on the main one if evertyhing is ok.Make sure you have good backups and be careful here. When SSDs die, they die ,and you may or may not get good notice.
A quick read: http://www.codinghorror.com/blog/2011/05/the-hot-crazy-solid-state-drive-scale.html
Ew. That is unacceptable. I can see taking the risk with a personal machine but not a prod server. For my Thinkpad I met in the middle and bought a Hybrid drive. I have seen marked improvement in speeds but no SSD like speeds. However, I don't want to have to restore my laptop once a month so I'll stick with it.
Cheers
May 1, 2012 at 9:53 am
Steve Jones - SSC Editor (5/1/2012)
Make sure you have good backups and be careful here. When SSDs die, they die ,and you may or may not get good notice.A quick read: http://www.codinghorror.com/blog/2011/05/the-hot-crazy-solid-state-drive-scale.html
Agreed; SSD's can go down very quickly. Then again, a head crash on a spindle drive is pretty quick also. We put both spinning disks and SSD's in a mix of RAID 5, RAID 1, RAID 10, and RAID 50, depending on what's required.
In all cases, you should have a level of fault tolerance that matches your organization's risk tolerance and budget. Having backups that you've actually restored somewhere to prove they really do work is also critical.
FYI: I anticipate starting to use RAID 6 in the near future, as the two drives worth of parity should help protect against the _lovely_ case of 1 drive with data corruption followed by a different drive failing catastrophically, or the even more wonderful case of multiple drives having data corruption on differing sectors, and then any of them failing catastrophically.
May 1, 2012 at 9:55 am
As a warning: run consistency checks on your RAID sets; see if you've got one or more drives that has corrupt data that your RAID level is hiding from you. Better, more modern controllers do this automatically, old ones don't.
Viewing 15 posts - 16 through 30 (of 95 total)
You must be logged in to reply to this topic. Login to reply