Re: FreeBSD + ZFS on a production server?
Wojciech Puchar wrote: ZFS on FreeBSD is GEOM-ified. While I believe what Wojciech said about needing a full disk is correct under Solaris, it's not the case in i never said it requires full disk. but it will work very slow sharing a disk with non-ZFS things. Well, of course if you are loading your disk with too many seeks it will be slow. This has nothing to do with ZFS. to say more: zfs set copies could be usable to selectively mirror given data while not mirroring other (using unprotected storage for ZFS). but it's broken. it writes N copies under write, but don't remake copies in case of failure! which make it almost unusable. in case of any failure you have to copy and delete every file to make it actually repaired. Eh? It happens automatically. Kris ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
Wojciech Puchar wrote: ZFS can be installed on partitions and share disks with other things, but the performance will be bad. There is nothing ZFS-specific about this statement. ZFS - contrary to every other filesystem that use FreeBSD disk I/O scheduler - does it's own I/O scheduling, so it assumes it's the only user of physical drive. This is false. ZFS uses GEOM along with everything else in FreeBSD, and GEOM is the thing that eventually talks to the disk driver to perform I/O. Kris ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
I'm behind on my mailing list reading and don't really want to prolong/resurrect this thread unduly, but I do want to respond to this point: On Thursday 12 June 2008 07:37:06 am Wojciech Puchar wrote: you must have disks dedicated for raidz, disks dedicated for mirrored storage and disks dedicated for unprotected storage. it's inflexible and not much usable. actually - much less usable than legacy gmirror/gstripe/gconcat+bsdlabel. ZFS on FreeBSD is GEOM-ified. While I believe what Wojciech said about needing a full disk is correct under Solaris, it's not the case in FreeBSD. Any GEOM provider can be added to a zpool--disk, slice, partition, gmirror, gstripe, md device, etc. I just added some storage to a personal server and re-did the layout using ZFS. My zpool (raidz) is made up of two partitions and one gstripe, spanning a total of four disks. I haven't had any issues with it at all (7-STABLE i386, 1.5GB RAM, no tuning other than kmem size and MAXPAGES). All of the disks also have other small partitions--two for a gmirrored root and three for swap. I think FreeBSD is a great storage/fileserver platform exactly _because_ there are so many options. UFS is great, gmirror and gstripe and friends are fantastic, and ZFS is yet another powerful tool in the arsenal. In my case ZFS was the best meeting point for space vs redundancy vs performance. Not having real RAID hardware my other candidates were graid3, graid5 and gvinum. ZFS is much easier to configure than gvinum, much more proven and stable than graid5 (which isn't even in the tree yet), and ought to perform better than graid3. I didn't do any testing to verify the last assertion since this is just a personal box, but I don't have any complaints about performance. JN one of my systems have 8 disks. 80% of data doesn't need any protection, it's just a need for a lot of space, other 20 needs to be mirrored. this 80% of data is used in high bandwidth/low seeks style (only big files). i simply partitioned every disk on 2 partitions, every first is used to make gmirror+gstripe device, every second is used to make gconcat device, and i have what i need WITH BALANCED LOAD. with ZFS i would have to make first 2 drives as mirror, another 6 for unprotected storage, having LOTS of seeks on first 2 drives and very little seeks on other 6 drives. the system would be unable to support the load. to say more: zfs set copies could be usable to selectively mirror given data while not mirroring other (using unprotected storage for ZFS). but it's broken. it writes N copies under write, but don't remake copies in case of failure! ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED] ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
ZFS on FreeBSD is GEOM-ified. While I believe what Wojciech said about needing a full disk is correct under Solaris, it's not the case in i never said it requires full disk. but it will work very slow sharing a disk with non-ZFS things. to say more: zfs set copies could be usable to selectively mirror given data while not mirroring other (using unprotected storage for ZFS). but it's broken. it writes N copies under write, but don't remake copies in case of failure! which make it almost unusable. in case of any failure you have to copy and delete every file to make it actually repaired. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
point: On Thursday 12 June 2008 07:37:06 am Wojciech Puchar wrote: you must have disks dedicated for raidz, disks dedicated for mirrored storage and disks dedicated for unprotected storage. it's inflexible and not much usable. actually - much less usable than legacy gmirror/gstripe/gconcat+bsdlabel. looks like my mistake - or simply a shortcut that made statement imprecise. it should be added: ZFS can be installed on partitions and share disks with other things, but the performance will be bad. ZFS - contrary to every other filesystem that use FreeBSD disk I/O scheduler - does it's own I/O scheduling, so it assumes it's the only user of physical drive. if both non-ZFS and ZFS filesystem will share the same disk AT THE SAME TIME - there will be a lots of thrashing. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
Just a small hint: You should configure your MUA to produce proper attribution lines. Wojciech Puchar wrote: Oliver Fromme wrote: A broken processor usually results in random crashes, not silent data corruption. result in both in my practice. with broken companion chips (chipset) it's silent data corruption is common, while crashes can be under specific cases. that's from what i've got. I've never had a broken processor that did not result in crashes, but maybe I've been just lucky. :-) or even calculate checksum right of wrong data generated by badly operating programs. What do you mean, wrong data generated by programs? If wrong data generated by program because of hardware problem. In that case the input to the program would have to be bad already. A broken disk (or controller) doesn't cause a program to produce wrong output, unless it feeds bad input to the program. And ZFS would catch that. You usually notice it when it's too late and the last good backup media was already recycled. not that bad, but of course - i make backups. But you don't keep every backup forever, do you? (I.e. it would rather be an archive instead of a backup. That would cost a lot of space.) In my case it was a disk with media surface errors, and the disk failed to report the error properly to the OS. Instead it just returned bad data. so i am just happy to never having it, while normal disk failures are quite common.. Yes, fortunately normal disk failures (i.e. reported to the OS so they are clearly noticed) are more common than silent corruption. ZFS may help detect it, or it may not. if it helped for you. Please stop spreading FUD. There is no may or may not. If a disk returns bad data, ZFS _will_ detect it. please read more carefully. i didn't say it. You did. I quoted it. i just say that disk returning bad data is very rare case, Yes, fortunately it is rare. But it does happen. And when it happens, ou are in very serious trouble. For example, on the -stable list Goran Lowkrantz reported on Saturday a corruption on one of his file systems due to a flipped bit in a directory node. He didn't use ZFS, but was lucky to notice the problem because of strange size entries in that directory. He had to use fsdb(8) surgery to fix it. Personally I would recommend to not use that disk anymore, because you never know in what other files bits could be flipped, without you noticing so easily. Well, or use ZFS on that disk -- then you're guaranteed to notice. lots of other - more frequent - hardware problems will not be detected. That's speculative. Personally I don't think so. if you like to give lots of CPU power and disk bandwidth for calculation You're spreading FUD again. The cpu time required for generating and verifying the checksums is very low, and the disk bandwidth is almost zero. i just say it doesn't make lot of protection against bad hardware, not worth the expense. Well, if the integrity of your files isn't important to you ... Best regards Oliver -- Oliver Fromme, secnetix GmbH Co. KG, Marktplatz 29, 85567 Grafing b. M. Handelsregister: Registergericht Muenchen, HRA 74606, Geschäftsfuehrung: secnetix Verwaltungsgesellsch. mbH, Handelsregister: Registergericht Mün- chen, HRB 125758, Geschäftsführer: Maik Bachmann, Olaf Erb, Ralf Gebhart FreeBSD-Dienstleistungen, -Produkte und mehr: http://www.secnetix.de/bsd That's what I love about GUIs: They make simple tasks easier, and complex tasks impossible. -- John William Chambless ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
On Mon, 9 Jun 2008 23:31:35 +0200 (CEST), Wojciech Puchar [EMAIL PROTECTED] said: W but why you need [a filesystem for linux that do checksum on the fly]?! all W PATA/SATA drives do checksumming on every read. in hardware, no CPU load. These days, hardware isn't just hardware. A disk drive can have around 300,000 lines of low-level firmware, and who wants to bet that it's completely bug-free? Silent-write errors are actually a big problem: http://www.usenix.org/publications/login/2008-06/openpdfs/bairavasundaram.pdf An Analysis of Data Corruption in the Storage Stack In this paper, we present the first large-scale study of data corruption. We analyze corruption instances recorded in production storage systems containing a total of 1.53 million disk drives, over a period of 41 months. We study three classes of corruption: checksum mismatches, identity discrepancies, and parity inconsistencies. We focus on checksum mismatches since they occur the most; more than 400,000 instances of checksum mismatches over the 41-month period. -- Karl Vogel I don't speak for the USAF or my company Mangled song lyric: Looks like tomatoes Actual lyric: Looks like we made it. (Barry Mannilow) ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
RE: FreeBSD + ZFS on a production server?
Anders Häggström wrote: I plan to install a web server for production use and ZFS looks very interesting, especially since it has built-in support for RAID and checksum. ZFS is very nice, but slightly over-hyped imho. However, some of the hype is warranted and for some use cases ZFS is a much better fit than UFS. Despite what Wojciech Puchar says, ZFS checksumming can be very useful. I recently had two drives in a hardware RAID-5 array (8 x 1 TB on a Highpoint RocketRAID 2340) develop unreadable sectors seemingly at the same time. I'm not sure what caused it but the end result was a broken/unavailable array. To make a long story short I managed to get the drives to remap the bad sectors and bring the array back online. Since I had ZFS on the array I didn't have to wait for fsck to run (takes a very long time on a 7 TB array and requires a LOT of memory to even work), and after the pool had been scrubbed I had a list of files with bad checksums that I could restore from backup. With UFS I would have had silent data corruption. Beware, there have been reports of mmap not working properly together with ZFS. I'm not sure if this is still a problem and if it would affect a typical web server. It does not seem to affect any of my fileservers (exporting NFS). /Daniel Eriksson ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
RE: FreeBSD + ZFS on a production server?
ZFS is very nice, but slightly over-hyped imho. not slightly and not only over-hyped. it's definitely far from being for storage as VM is for memory. for example you can't select per file (or at least - per pseudo-filesystem) if you want no protection, mirrored or raidz. you must have disks dedicated for raidz, disks dedicated for mirrored storage and disks dedicated for unprotected storage. it's inflexible and not much usable. actually - much less usable than legacy gmirror/gstripe/gconcat+bsdlabel. one of my systems have 8 disks. 80% of data doesn't need any protection, it's just a need for a lot of space, other 20 needs to be mirrored. this 80% of data is used in high bandwidth/low seeks style (only big files). i simply partitioned every disk on 2 partitions, every first is used to make gmirror+gstripe device, every second is used to make gconcat device, and i have what i need WITH BALANCED LOAD. with ZFS i would have to make first 2 drives as mirror, another 6 for unprotected storage, having LOTS of seeks on first 2 drives and very little seeks on other 6 drives. the system would be unable to support the load. to say more: zfs set copies could be usable to selectively mirror given data while not mirroring other (using unprotected storage for ZFS). but it's broken. it writes N copies under write, but don't remake copies in case of failure! ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
RE: FreeBSD + ZFS on a production server?
On Thu, 2008-06-12 at 13:37 +0200, Wojciech Puchar wrote: for example you can't select per file (or at least - per pseudo-filesystem) if you want no protection, mirrored or raidz. Isn't it a pity that the fbsd implementation of ZFS lacks such a feature. Your anti stories of ZFS often show these aspects. Almost none of your comments on zfs are valid in Solaris. But hey, what the heck, use what you want. I don't write for SUN, I use ZFS on all systems and it never disappointed me. It's blazingly fast, very flexible, configurable. Stripes, mirrors, it's all so easy. It will be even better in time. Conclusion: Wojciech Puchar is against zfs. so what. I belong to the pro's fwiw. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
RE: FreeBSD + ZFS on a production server?
pseudo-filesystem) if you want no protection, mirrored or raidz. Isn't it a pity that the fbsd implementation of ZFS lacks such a feature. Your anti stories of ZFS often show these aspects. Almost none of your comments on zfs are valid in Solaris. AFAIK on solaris set copies= and what i told before is the same. am i wrong? ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
Wojciech Puchar wrote: 3) a CPU,cache and memory bandwidth hogging feature of checksumming all blocks. thing that are already done in disk hardware. fortunately you can turn this off Obviously you have been lucky to never be a victim of silent disk corruption (or you just haven't noticed). There are other people who didn't have that much luck, including me. ZFS' checksumming and self-healing is a blessing. If you don't know how it works and call it marketing blah, then I suggest you read up on it a bit. And by the way, it doesn't take any significant amount of CPU power on hardware that is not ancient. I agree that ZFS is not suitable to run on ancient hardware. It isn't designed for that. You're free to use UFS, of course, and keep suffering from its shortcomings. Best regards Oliver -- Oliver Fromme, secnetix GmbH Co. KG, Marktplatz 29, 85567 Grafing b. M. Handelsregister: Registergericht Muenchen, HRA 74606, Geschäftsfuehrung: secnetix Verwaltungsgesellsch. mbH, Handelsregister: Registergericht Mün- chen, HRB 125758, Geschäftsführer: Maik Bachmann, Olaf Erb, Ralf Gebhart FreeBSD-Dienstleistungen, -Produkte und mehr: http://www.secnetix.de/bsd Python is an experiment in how much freedom programmers need. Too much freedom and nobody can read another's code; too little and expressiveness is endangered. -- Guido van Rossum ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
Wojciech Puchar wrote: 3) a CPU,cache and memory bandwidth hogging feature of checksumming all blocks. thing that are already done in disk hardware. fortunately you can turn this off Obviously you have been lucky to never be a victim of silent disk corruption (or you just haven't noticed). what you mean. that disk wrote the data wrong and doesn't detect it on read? i would mean broken disk processor, it's memory etc. possible - as much as broken main processor, main memory, some of chips on motherboard etc. - which will make ZFS calculate checksum wrong on write, or even calculate checksum right of wrong data generated by badly operating programs. given the complexity of motherboard+CPU etc. to complexity of disk hardware, i don't think silent disk failure happens often. i think all your cases wasn't disk, but general hardware problems. ZFS may help detect it, or it may not. if it helped for you. even without ZFS it WOULD cause problems with programs like random crashes. personally i often got disk failing the way that it was unable to read or write giving an error, but never things like that. You're free to use UFS, of course, and keep suffering from its shortcomings. i have to start suffering at first ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
[attribution fixed] Wojciech Puchar wrote: Oliver Fromme wrote: Wojciech Puchar wrote: 3) a CPU,cache and memory bandwidth hogging feature of checksumming all blocks. thing that are already done in disk hardware. fortunately you can turn this off Obviously you have been lucky to never be a victim of silent disk corruption (or you just haven't noticed). what you mean. that disk wrote the data wrong and doesn't detect it on read? i would mean broken disk processor, it's memory etc. Correct. It does happen. possible - as much as broken main processor, main memory, some of chips on motherboard etc. - A broken processor usually results in random crashes, not silent data corruption. Broken memory will be noticed if it supports ECC, otherwise it will also result in crashes, most probably. which will make ZFS calculate checksum wrong on write, Even if that happens (without crashes or other things that you'll notice immediately), the error will be detected by ZFS and fixed (healed) if possible, i.e. when running with redundancy and at least one copy has a good checksum. (GELI can only detect, but not fix. ZFS can fix it, too. I assume in theory it would be possible to make geli co- operate with gmirror so it could fix bad blocks, too, but that's just theory. ZFS is reality.) or even calculate checksum right of wrong data generated by badly operating programs. What do you mean, wrong data generated by programs? If a program generates wrong output, there's nothing any file system could do about that. That's not the file system's job at all. The file systems job is to ensure the integrity of data written to the disk, and ZFS does exactly that. given the complexity of motherboard+CPU etc. to complexity of disk hardware, i don't think silent disk failure happens often. Fortunately it doesn't happen often, but it does happen. And when it happens, you are in really serious trouble. You usually notice it when it's too late and the last good backup media was already recycled. i think all your cases wasn't disk, but general hardware problems. In my case it was a disk with media surface errors, and the disk failed to report the error properly to the OS. Instead it just returned bad data. ZFS may help detect it, or it may not. if it helped for you. Please stop spreading FUD. There is no may or may not. If a disk returns bad data, ZFS _will_ detect it. Silent corruption _cannot_ happen with ZFS, except if you disable the checksumming feature intentionally. even without ZFS it WOULD cause problems with programs like random crashes. Please elaborate what the problem is, if you think there is one. personally i often got disk failing the way that it was unable to read or write giving an error, but never things like that. As I said: You were lucky. You're free to use UFS, of course, and keep suffering from its shortcomings. i have to start suffering at first Many people suffer without knowing. :-) I do suffer from UFS' shortcomings on many machines on which I can't use ZFS (or other file systems) for various reasons. Best regards Oliver -- Oliver Fromme, secnetix GmbH Co. KG, Marktplatz 29, 85567 Grafing b. M. Handelsregister: Registergericht Muenchen, HRA 74606, Geschäftsfuehrung: secnetix Verwaltungsgesellsch. mbH, Handelsregister: Registergericht Mün- chen, HRB 125758, Geschäftsführer: Maik Bachmann, Olaf Erb, Ralf Gebhart FreeBSD-Dienstleistungen, -Produkte und mehr: http://www.secnetix.de/bsd With Perl you can manipulate text, interact with programs, talk over networks, drive Web pages, perform arbitrary precision arithmetic, and write programs that look like Snoopy swearing. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
A broken processor usually results in random crashes, not silent data corruption. result in both in my practice. with broken companion chips (chipset) it's silent data corruption is common, while crashes can be under specific cases. that's from what i've got. or even calculate checksum right of wrong data generated by badly operating programs. What do you mean, wrong data generated by programs? If wrong data generated by program because of hardware problem. You usually notice it when it's too late and the last good backup media was already recycled. not that bad, but of course - i make backups. i think all your cases wasn't disk, but general hardware problems. In my case it was a disk with media surface errors, and the disk failed to report the error properly to the OS. Instead it just returned bad data. so i am just happy to never having it, while normal disk failures are quite common.. ZFS may help detect it, or it may not. if it helped for you. Please stop spreading FUD. There is no may or may not. If a disk returns bad data, ZFS _will_ detect it. please read more carefully. i didn't say it. i just say that disk returning bad data is very rare case, lots of other - more frequent - hardware problems will not be detected. if you like to give lots of CPU power and disk bandwidth for calculation of checksums on each read/write - then OK. if you think you are secured this way - then OK. i just say it doesn't make lot of protection against bad hardware, not worth the expense. i probably shouldn't type that point as it can be turned off in ZFS. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
On Tue, 10 Jun 2008 00:29:01 +0200 (CEST) Wojciech Puchar [EMAIL PROTECTED] wrote: Because the ZFS checksumming makes the FS selfhealing. Chance for selfhealing WHAT?! could you please instead of repeating sun marketing text like all others tell something clearer? Do your own homework, please. -- Dick Hoogendijk -- PGP/GnuPG key: 01D2433D ++ http://nagual.nl/ + SunOS sxde 01/08 ++ ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
Because the ZFS checksumming makes the FS selfhealing. Chance for selfhealing WHAT?! could you please instead of repeating sun marketing text like all others tell something clearer? Do your own homework, please. i actually did. instead of repeating marketing blah blah. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
Wojciech Puchar wrote: 2) ZFS RAID-z turns your X drives to single drive performance both on read and write. every normal RAID-5 implementation will give you random read speed of X-1 times single drive speed, while slow random write speeds (but still at least half of single drive). but this is advertised as a feature Is this because of checksum verification (the need to read all components) or nothing to checksum. sit down for a while and think. you have say 100 MB file linearly placed on position A on disk. and your program requests 100kB writes to possitions like 200kB*(between 0 and 499) randomly. ZFS will cache all it then blow all this as 50MB linear write at position B on disk. then after some time (data not in cache) you like to read file linearly. what you get: read 100kB from position B, seek to position A to read 100kB, then to position B reading 100kB, then to position A etc. etc. LOTS of seeking. while reads are more common than writes on most cases ZFS make things worse. of course i told about good case where ZFS could find large continous space. if your drive is well filled it's unlikely. with UFS disk is divided for cylinder groups. so too - it's unlikely you will find large continuous space BUT there are very likely you will find large chunk of fragments withing same cylinder group which requires much shorter head movement. while really big blocks are forcibly splitted to different cylinder group, as having long seek every few megabytes isn't a problem. that's what UFS does for 20 years. the major improvement then was soft updates, now it is really fast even with small files. i just skipped talking about memory and CPU usage, because there is nothing to talk much. it's just make ZFS crap and nothing else. Today most people's problem are TOO FAST CPU and TOO MUCH MEMORY.Sun found the solution, just like windows. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
random write speeds (but still at least half of single drive). but this is advertised as a feature Is this because of checksum verification (the need to read all components) or something else? Any documentation/references? RAID-Z stores a single checksum over the whole stripe, instead of checksumming each disks's section separately, so it has to read from all disks to validate the stripe. Only random reads are penalized, though. random reads are most common read on unix, unless you process linearly huge files, but that's fast on UFS too. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
Thank you all for discussing this with me. I really like to here your opinions. I wont answer to all of your posts, because half of them is off-topic, but still interesting to read. I haven't heard/read about any huge CPU consumptions from ZFS, not yet at least. If you have links to benchmarks and comparisons with other fses (UFS2 in particular) it would be grate! For the memory I've read that ZFS use up to approximately 700MB of ram for caching, which is quite much, but not too much compared to my 4GB that is available. However there doesn't seem to be an upper limit for ZFS, which I think is very bad. 2008/6/8 Wojciech Puchar [EMAIL PROTECTED]: The choice is probably between Debian 4.0r3, FreeBSD 7.0 and OpenSolaris 2008.05. All of them have their pros and cons. could you tell any pros for opensolaris? OpenSolaris 2008.05 didn't boot on my hardware, so it's out of this project anyway. The live-cd hangs on device detect. One of the pros for OpenSolaris I've noticed is the support for a virtual host and at the same time able to use ZFS, but that doesn't matter anymore because I can not boot it and ZFS will probably eat my memory if I can set an upper limit. I think Debian / Linux, almost falls off because it lacks support for native ZFS and I have not found any alternative filesystem that offer checksums on the fly. agree I have now found a filesystem for linux that do checksum on the fly, btrfs. But it is still very experimental, so I wont try it for this project. http://btrfs.wiki.kernel.org/ // Anders ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
On Mon, 9 Jun 2008 20:58:10 +0200 Anders Häggström [EMAIL PROTECTED] wrote: For the memory I've read that ZFS use up to approximately 700MB of ram for caching, which is quite much, but not too much compared to my 4GB that is available. However there doesn't seem to be an upper limit for ZFS, which I think is very bad. This limit can be tuned. At least on solaris. Also, ZFS definitely prefers a 64 bit kernel. -- Dick Hoogendijk -- PGP/GnuPG key: 01D2433D ++ http://nagual.nl/ + SunOS sxde 01/08 ++ ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
2008/6/9 Dick Hoogendijk [EMAIL PROTECTED]: On Mon, 9 Jun 2008 20:58:10 +0200 Anders Häggström [EMAIL PROTECTED] wrote: For the memory I've read that ZFS use up to approximately 700MB of ram for caching, which is quite much, but not too much compared to my 4GB that is available. However there doesn't seem to be an upper limit for ZFS, which I think is very bad. This limit can be tuned. At least on solaris. Also, ZFS definitely prefers a 64 bit kernel. That's good to know, thanks! Do you have any reference/link that describes how to manage that? It's good to know for the future. I always run my server on amd64-software, as far as I can, because I see better performance and I do not have the trouble with 4GB memory limit. Thanks for the info! -- Dick Hoogendijk -- PGP/GnuPG key: 01D2433D ++ http://nagual.nl/ + SunOS sxde 01/08 ++ ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED] ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
I haven't heard/read about any huge CPU consumptions from ZFS, not yet as i already said. most people today have problems as they have too fast CPU and too much RAM ;) they don't see high CPU load on quad core machine with 16GB RAM having not big load :) For the memory I've read that ZFS use up to approximately 700MB of ram UFS use what's unused. works on 16MB and 16GB. project anyway. The live-cd hangs on device detect. One of the pros for OpenSolaris I've noticed is the support for a virtual host man jail while maybe not with resource control like on solaris, but i use it with success. it's really excellent. but use nullfs with it to be able to share binaries. I have now found a filesystem for linux that do checksum on the fly, btrfs. but why you need it?! all PATA/SATA drives do checksumming on every read. in hardware, no CPU load. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
This limit can be tuned. At least on solaris. Also, ZFS definitely prefers a 64 bit kernel. That's good to know, thanks! Do you have any reference/link that describes how to manage that? It's good to know for the future. when i tested it i was able to run it on 256MB machine stable after reading about tuning. but sorry i don't remember what options it were exactly. I always run my server on amd64-software, as far as I can, because I me too, while not having 4GB anywhere. it just runs faster ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
On Mon, 9 Jun 2008 23:31:35 +0200 (CEST) Wojciech Puchar [EMAIL PROTECTED] wrote: UFS use what's unused. works on 16MB and 16GB. It's difficult to tell about consumed memory in ZFS vs UFS since UFS can be quite agressive at caching as well. -(although this caching is often hidden by system tools and reported as unused memory)- project anyway. The live-cd hangs on device detect. One of the pros for OpenSolaris I've noticed is the support for a virtual host man jail Jails are just slightly comparable to solaris zones. It's much more then resource control! They're really like independent machines with almost no memory footprint. It's quite common to run different zones for a mailserver, webserver and i.e. users. Sparse zones use little space, because lots of code is shared. It's all very tunable. Running 40 sparse zones is hardly noticable. Try that with 40 jails;-) I have now found a filesystem for linux that do checksum on the fly, btrfs. but why you need it?! all PATA/SATA drives do checksumming on every read. in hardware, no CPU load. Because the ZFS checksumming makes the FS selfhealing. Chance for errors are almost nill. No fsck. Yes, it consumes memory, but memory is cheap, very cheap! CPU load is hardly noticable. -- Dick Hoogendijk -- PGP/GnuPG key: 01D2433D ++ http://nagual.nl/ + SunOS sxde 01/08 ++ ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
Running 40 sparse zones is hardly noticable. Try that with 40 jails;-) you probably don't have your jails configured right. my 1GB pentium 4 machine runs 20 jails, and it is hardly noticable. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
FreeBSD + ZFS on a production server?
Hello list! I plan to install a web server for production use and ZFS looks very interesting, especially since it has built-in support for RAID and checksum. The hardware is already purchased, a 1U-casis with a PhemonX4 9550 CPU, 4GB ECC RAM @ 800MHz and 2x500GB SATA disks and I am about to select the operating system to use. The choice is probably between Debian 4.0r3, FreeBSD 7.0 and OpenSolaris 2008.05. All of them have their pros and cons. I think Debian / Linux, almost falls off because it lacks support for native ZFS and I have not found any alternative filesystem that offer checksums on the fly. My main question is: How is the support for ZFS on FreeBSD? Is it sufficiently stable and fast enough to be used in production yet? If not, is there any alternative filesystem that offers checksums on the fly or other similar technology to reduce the risk of a corrupt filesystem that at the same time plays well with software RAID (RAID-1 in particular)? Thanks in advance! Anders ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
Anders Häggström [EMAIL PROTECTED] wrote: I plan to install a web server for production use and ZFS looks very interesting, especially since it has built-in support for RAID and checksum. The hardware is already purchased, a 1U-casis with a PhemonX4 9550 CPU, 4GB ECC RAM @ 800MHz and 2x500GB SATA disks and I am about to select the operating system to use. The choice is probably between Debian 4.0r3, FreeBSD 7.0 and OpenSolaris 2008.05. All of them have their pros and cons. Just in case you assume that ZFS on OpenSolaris 2008.05 would be superior to ZFS on FreeBSD, this hasn't been my experience. On a system with an Athlon 1700+ and only 512 MB of RAM, receiving snapshots on OpenSolaris renders the GUI pretty much useless. On FreeBSD ZFS operations can cause delays as well, but it's significantly better than on OpenSolaris, even though FreeBSD's ZFS pool lies on a geli-encrypted gmirror while OpenSolaris uses the disk directly. Note that the system is below Sun's recommended specifications for ZFS, though. Things may look differently on more powerful systems. I think Debian / Linux, almost falls off because it lacks support for native ZFS and I have not found any alternative filesystem that offer checksums on the fly. My main question is: How is the support for ZFS on FreeBSD? Is it sufficiently stable and fast enough to be used in production yet? It probably depends on your workload, you'll find several complaints in the archives. It works fine for me, but I haven't tried it on web servers yet. If I were to install a web server today, though, I'd definitely go with ZFS (on FreeBSD). If not, is there any alternative filesystem that offers checksums on the fly or other similar technology to reduce the risk of a corrupt filesystem that at the same time plays well with software RAID (RAID-1 in particular)? You can use geli(8) for checksumming, it can be combined with gmirror but unless with ZFS, you don't get automatic self-healing. Fabian signature.asc Description: PGP signature
Re: FreeBSD + ZFS on a production server?
On Sun, 8 Jun 2008 16:24:56 +0200 Fabian Keil [EMAIL PROTECTED] wrote: On a system with an Athlon 1700+ and only 512 MB of RAM, receiving snapshots on OpenSolaris renders the GUI pretty much useless. Note that the system is below Sun's recommended specifications for ZFS, though. Things may look differently on more powerful systems. The -bare- minimum for OpenSolaris is 512MB. That's not only for ZFS. In my experience ZFS on solaris is rock solid. OpenSolaris is not yet ready for production servers though i.m.h.o. nor is nevada_b90 with the ability to boot off ZFS root. Production servers need to be well (no thoroughly) tested ;-) The best stable (production) server with ZFS is solaris-10u5 If you want to boot off ZFS, S10u6 will support that. But these versions too need lots of ram. I think fbsd has a lighter footprint. -- Dick Hoogendijk -- PGP/GnuPG key: 01D2433D ++ http://nagual.nl/ + SunOS sxde 01/08 ++ ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
Thanks for the quick answers! 2008/6/8 Fabian Keil [EMAIL PROTECTED]: Just in case you assume that ZFS on OpenSolaris 2008.05 would be superior to ZFS on FreeBSD, this hasn't been my experience. Yes, I assumed that because Sun can implement and optimize ZFS to fit OpenSolaris, while we run the risk with FreeBSD to implement bugs while we implement ZFS. 2008/6/8 Dick Hoogendijk [EMAIL PROTECTED]: The best stable (production) server with ZFS is solaris-10u5 If you want to boot off ZFS, S10u6 will support that. But these versions too need lots of ram. I think fbsd has a lighter footprint. As you might have read, I have quite a lot of RAM available on this server (4GB), but ofcource I want the operating system to take as little as possible so that I have as much RAM as possible over for the server processes to work with (mostly web-server and mysql-server). According to a page I have found it says that some basic (FreeBSD-specific) functions are not ready, how does that affect ZFS in general? http://wiki.freebsd.org/ZFS Regards Anders ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
2008/6/8 Fabian Keil [EMAIL PROTECTED]: Anders Häggström [EMAIL PROTECTED] wrote: [...] Just in case you assume that ZFS on OpenSolaris 2008.05 would be superior to ZFS on FreeBSD, this hasn't been my experience. On a system with an Athlon 1700+ and only 512 MB of RAM, receiving snapshots on OpenSolaris renders the GUI pretty much useless. On FreeBSD ZFS operations can cause delays as well, but it's significantly better than on OpenSolaris, even though FreeBSD's ZFS pool lies on a geli-encrypted gmirror while OpenSolaris uses the disk directly. This hasn't anything to do with ZFS but on the different configuration of the clockrate. FreeBSD uses 1000 ticks, while it's 500 on Solaris. This means that FreeBSD can switch to different tasks twice as fast than Solaris. For a server a high tick rate isn't necessary, so it doesn't matter really. And Solaris still is a server OS. If you're running a desktop it makes quite a difference, of course. Interesstingly enough PC BSD configures kern.clockrate to 2000. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
The choice is probably between Debian 4.0r3, FreeBSD 7.0 and OpenSolaris 2008.05. All of them have their pros and cons. could you tell any pros for opensolaris? I think Debian / Linux, almost falls off because it lacks support for native ZFS and I have not found any alternative filesystem that offer checksums on the fly. agree My main question is: How is the support for ZFS on FreeBSD? Is it while i don't use it, it works rather as in manual. no crashes if configured right. sufficiently stable and fast enough to be used in production yet? that's just my opinion about ZFS that it isn't very useful at all. it's just memory and CPU eater. If not, is there any alternative filesystem that offers checksums on the fly or other similar technology to reduce the risk of a corrupt filesystem that at the same time plays well with software RAID (RAID-1 in particular)? while i use RAID-1 for a long time be it linux or netbsd or freebsd, there is no need for checksumming. there are sector's checksums on disks, checked on every read. in SATA protocol there is error checking during transmission too. there is already well done things in hardware to do disk transfers without CPU overhead, but ZFS introduces overhead and advertises it as feature. quick advice - gmirror this 2 drives and then use UFS. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
On a system with an Athlon 1700+ and only 512 MB of RAM, receiving snapshots on OpenSolaris renders the GUI pretty much useless. looks like very bad CPU and I/O scheduling on Solaris. maybe that's their 32-64 hardware threads capable chip is advertised so much? :) On FreeBSD ZFS operations can cause delays as well, but it's significantly better than on OpenSolaris, even though FreeBSD's ZFS pool lies on a geli-encrypted gmirror while OpenSolaris uses the disk directly. there is quite big difference with geli. it is CPU eater and produces delays noticable on machines that like P3 or less. but at least - it does something useful unlike these ZFS checksumming and other things. Note that the system is below Sun's recommended specifications for ZFS, though. Things may look differently on more powerful systems. but comparision probably the same, or difference less noticable on stronger systems. You can use geli(8) for checksumming, it can be combined with gmirror but unless with ZFS, you don't get automatic self-healing. whatever it means ;) ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
On a system with an Athlon 1700+ and only 512 MB of RAM, receiving snapshots on OpenSolaris renders the GUI pretty much useless. On FreeBSD ZFS operations can cause delays as well, but it's significantly better than on OpenSolaris, even though FreeBSD's ZFS pool lies on a geli-encrypted gmirror while OpenSolaris uses the disk directly. anyway something get changed between FreeBSD 6 and 7. as i changed this on quite loaded server from 6.3 to 7 - general throughput increased well. tasks are done much faster. on on my laptop it's the same, but interactive delays was much lower on 6.3 is it possible to turn old mode on FreeBSD 7. on my laptop interactive performance is more important. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
As you might have read, I have quite a lot of RAM available on this server (4GB), but ofcource I want the operating system to take as little as possible so that I have as much RAM as possible over for the server processes to work with (mostly web-server and mysql-server). ZFS is memory and CPU eater. prepare that very few will be left for actual work ;) ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
If you're running a desktop it makes quite a difference, of course. Interesstingly enough PC BSD configures kern.clockrate to 2000. human can't notice delays below 10ms. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
Christian Walther [EMAIL PROTECTED] wrote: 2008/6/8 Fabian Keil [EMAIL PROTECTED]: Anders Häggström [EMAIL PROTECTED] wrote: [...] Just in case you assume that ZFS on OpenSolaris 2008.05 would be superior to ZFS on FreeBSD, this hasn't been my experience. On a system with an Athlon 1700+ and only 512 MB of RAM, receiving snapshots on OpenSolaris renders the GUI pretty much useless. On FreeBSD ZFS operations can cause delays as well, but it's significantly better than on OpenSolaris, even though FreeBSD's ZFS pool lies on a geli-encrypted gmirror while OpenSolaris uses the disk directly. This hasn't anything to do with ZFS but on the different configuration of the clockrate. FreeBSD uses 1000 ticks, while it's 500 on Solaris. With OpenSolaris 2008.05 the GUI becomes unresponsive for multiple-seconds on my system, and it's not clear to me how the clock rate difference would explain that. Fabian signature.asc Description: PGP signature
Re: FreeBSD + ZFS on a production server?
On Sun, 8 Jun 2008 22:08:59 +0200 (CEST) Wojciech Puchar [EMAIL PROTECTED] wrote: As you might have read, I have quite a lot of RAM available on this server (4GB), but ofcource I want the operating system to take as little as possible so that I have as much RAM as possible over for the server processes to work with (mostly web-server and mysql-server). ZFS is memory and CPU eater. prepare that very few will be left for actual work ;) Bollocks. It consumes memory. The more seperate filesystems, the more memory. But don't execurate. For a webserver on zfs 4GB is more than enough. -- Dick Hoogendijk -- PGP/GnuPG key: 01D2433D ++ http://nagual.nl/ + SunOS sxde 01/08 ++ ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
On Sun, 8 Jun 2008 22:05:08 +0200 (CEST) Wojciech Puchar [EMAIL PROTECTED] wrote: On a system with an Athlon 1700+ and only 512 MB of RAM, receiving snapshots on OpenSolaris renders the GUI pretty much useless. looks like very bad CPU and I/O scheduling on Solaris. maybe that's their 32-64 hardware threads capable chip is advertised so much? :) Don't write about things you don't know. *Maybe's* don't help. You don't have to like solaris but don't troll about it, please. Both systems have their pro's and cons. -- Dick Hoogendijk -- PGP/GnuPG key: 01D2433D ++ http://nagual.nl/ + SunOS sxde 01/08 ++ ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
On Sun, 8 Jun 2008 22:01:23 +0200 (CEST) Wojciech Puchar [EMAIL PROTECTED] wrote: while i don't use it, it works rather as in manual. no crashes if configured right. that's just my opinion about ZFS that it isn't very useful at all. it's just memory and CPU eater. Your entitled to your opinion, but please try to base it on some facts. ZFS is herre to stay. You better get used to it. at least you could try to work with it before you make up an opinion. Have you -any- idea at all what this FS is capable off? -- Dick Hoogendijk -- PGP/GnuPG key: 01D2433D ++ http://nagual.nl/ + SunOS sxde 01/08 ++ ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
This hasn't anything to do with ZFS but on the different configuration of the clockrate. FreeBSD uses 1000 ticks, while it's 500 on Solaris. With OpenSolaris 2008.05 the GUI becomes unresponsive for multiple-seconds on my system, and it's not clear to me how the clock rate difference would explain that. no. it's just because of scheduling and I/O algorithms used in solaris. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
ZFS is memory and CPU eater. prepare that very few will be left for actual work ;) Bollocks. It consumes memory. The more seperate filesystems, the more memory. But don't execurate. For a webserver on zfs 4GB is more than enough. still enough for UFS with softupdates - which is REALLY fast. you may set kern.maxvnodes much higher to speed it up even more. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
configured right. that's just my opinion about ZFS that it isn't very useful at all. it's just memory and CPU eater. Your entitled to your opinion, but please try to base it on some facts. ZFS is herre to stay. You better get used to it. at least you could try AFAIK there are no plans to FORCE using ZFS instead of UFS in FreeBSD. or there are? if so - fine time to check something else. to work with it before you make up an opinion. Have you -any- idea at yes i am. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
ZFS is herre to stay. You better get used to it. at least you could try to work with it before you make up an opinion. Have you -any- idea at all what this FS is capable off? if you like - quick summary 1) ZFS turns random writes into sequential writes as they say. yes that's true. they just forgot to say that it then turns sequential reads into random reads. simple for anyone that still can use his/her brain. 2) ZFS RAID-z turns your X drives to single drive performance both on read and write. every normal RAID-5 implementation will give you random read speed of X-1 times single drive speed, while slow random write speeds (but still at least half of single drive). but this is advertised as a feature 3) a CPU,cache and memory bandwidth hogging feature of checksumming all blocks. thing that are already done in disk hardware. fortunately you can turn this off 4) write anywhere style of writing, just with large buffers it could get large blocks to be written at once if only large continous space are found. quite good (but not that much better than UFS) as long as your drive is mostly empty. 5) incredibly high memory consumption. very high CPU consumption compared to UFS. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: FreeBSD + ZFS on a production server?
Wojciech Puchar wrote: 2) ZFS RAID-z turns your X drives to single drive performance both on read and write. every normal RAID-5 implementation will give you random read speed of X-1 times single drive speed, while slow random write speeds (but still at least half of single drive). but this is advertised as a feature Is this because of checksum verification (the need to read all components) or something else? Any documentation/references? signature.asc Description: OpenPGP digital signature
Re: FreeBSD + ZFS on a production server?
In the last episode (Jun 09), Ivan Voras said: Wojciech Puchar wrote: 2) ZFS RAID-z turns your X drives to single drive performance both on read and write. every normal RAID-5 implementation will give you random read speed of X-1 times single drive speed, while slow random write speeds (but still at least half of single drive). but this is advertised as a feature Is this because of checksum verification (the need to read all components) or something else? Any documentation/references? RAID-Z stores a single checksum over the whole stripe, instead of checksumming each disks's section separately, so it has to read from all disks to validate the stripe. Only random reads are penalized, though. http://blogs.sun.com/roch/entry/when_to_and_not_to -- Dan Nelson [EMAIL PROTECTED] ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
