On 20.01.2012 04:39, Stan Hoeppner wrote: [] >>> But that >>> alone isn't going to fix a 10x performance deficit. You've probably got >>> multiple factors degrading performance. >> >> Yes, you have right. But I found recently, that disk mounted on my >> server are slow 5.9K. My tests on in shows that they do fsync 1.5x-2x >> slower than 7.2K with quite often 5x-10x slower peak. Together with >> raid10, lvm, ext4, and much heavier load during delivery it may give >> effect I'm observing. > > 5.9K RPM? Bingo. There's the problem. Those are "green" drives, from > one manufacturer or another, probably Western Digital EARS 2TB drives.
Most likely seagate. WD usually either does not tell the speed (saying it varies) or they're 5.4K RPM. > They are not suitable for RAID use, nor server use, nor any high > performance use whatsoever. As you've seen first hand their performance > is low, unpredictable, and unreliable. > > In addition, they are "Advanced Format" drives, meaning 4KB hardware > sectors but reported to the host as 512 byte sectors. This can cause > stripe alignment problems with RAID and the filesystem, which will > exacerbate fsync delays. Normally this misalignment is only an issue > with parity arrays but it can also affect non parity striped arrays in > certain configurations. > Needless to say, you're not going to get decent queue spooling > performance with these green drives, ever. If you can't wholesale swap > these 8 drives for units suitable for mail server duty, consider > sticking two small inexpensive 7.2k SATA drives in the box and mirroring > them. Move the Postfix spool directory onto them--and any other > applications you're running that need higher random IOPS performance > than you're going to get from these green drives. Please excuse me for the somewhat harsh words, but except of the alignment issues which should be solved for once when partitioning and creating filesystem, the rest is a complete bullshit collected from various forums where people does not understand what they're doing and blame bad drives. I understand this is not a proper place to discuss this too: it is postfix, it is not a disk comparison forum. But I can't resist. These drives are excellent when set up and used properly (the misalignment issue you mentioned is real indeed, and MUST be taken into account: everything should be aligned to 4Kb, lots of especially old tools don't do that or even don't LET you to do that - eg cfdisk in linux). Very reliable, fast and predictable. I'll give just one note about speed, which may look completely wrong at first. The reason they're speedy is that at their low rotational speed, they also have much more data density, -- ie, basically, they can transfer much more data during single rotate. This way, their linear speed (sequentional read or write) often goes FASTER than enterprize-class 15KRPM drives which are of much less volumes (300-600Gb as opposed to 1 or 2Tb or more for these "green" drives). Yes, due to slower rotational speed, they take more time to position platter to the right place. But that's, again, not whole story: the seek time is about the same as for their "elder" brothers. Now, use just first 300 or 600Gb out of this 2Tb drive, to have more fair comparison with 15KRPM drives, and you realize that the seek time improves greatly, since we now have to seek less! That's about speed. As you can see, the picture is FAR from definitive. Their speed is reduced _dramatically_ when the drive have to resort to read-modify-write cycle when the host sends data to write in 512-byte units instead of in 4kb units, or when the blocks are not aligned to 4kb. This is where 99% of hysteria comes from in various forums where people blame these drives. And this one can't be solved 100% by doing right partitioning and filesystem alignment: in addition to right alignment, _size_ of each block being transferred matters too, it should not be 512 but 4096. This is where some operating systems (especially windowsXP which is still used alot) cant guarantee the right size. Linux, especially recent kernel versions, is much better, but here, again, it largely depends on the filesystem. Note that it is very difficult (lacking proper tools) to align msdos-style partitions properly, since traditionally, all partitions in extended partition are aligned to (chs)+512, and all older tools will force the +512 shift. That's enough for now about speeed. As for reliability, it is not better and not worse than any other drives. One can argue that "enterprise-grade" drives are more reliable, but that's very questionable still. And these "consumer-grade" drives have an advantage: they cost alot less, so for the same money you can get several of them to use them in higher-redundrancy raid array, to back the theorerical lack of high reliability. > This is likely the least expensive way, in both $$ and effort, to solve > your problem. I never tried running mailserver tests on an array of such drives because I don't have many of them. But a single 2Tb WD20EARS drive outperforms single 500Gb Hitachi enterprise-class _sata_ (not sas) 7200RPM (both are of the same "generation", ie, bought at about the same time and both were current) for about 10% for postfix smtp-sink workload. Go figure. Thanks, /mjt
