Re: ZFS vs EXT4 vs XFS
There's been discussion on the ZFS bloghttp://www.opensolaris.org/jive/thread.jspa?messageID=150818tstart=0#150818 . Someone else did a benchmark comparing ZFS on hardware raid vs software raid. Software was faster on this system, plus you get the ECC type stuff with ZFS. On 8/30/07, Dan Miller [EMAIL PROTECTED] wrote: Interesting benchmarks: http://tastic.brillig.org/%7Ejwb/zfs-xfs-ext4.html I would also like to see Reiserfs4 in those results to see how it compares, but those are not included. Dan ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/ ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: ZFS vs EXT4 vs XFS
On 8/31/07, Tom Buskey [EMAIL PROTECTED] wrote: Someone else did a benchmark comparing ZFS on hardware raid vs software raid. Software was faster on this system, plus you get the ECC type stuff with ZFS. I regard most such storage-related benchmarks with a great deal of suspicion. They always seem to assume the computer won't be doing anything else when the filesystem is being used. -- Ben ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: ZFS vs EXT4 vs XFS
On 8/31/07, Ben Scott [EMAIL PROTECTED] wrote: On 8/31/07, Tom Buskey [EMAIL PROTECTED] wrote: Someone else did a benchmark comparing ZFS on hardware raid vs software raid. Software was faster on this system, plus you get the ECC type stuff with ZFS. I regard most such storage-related benchmarks with a great deal of suspicion. They always seem to assume the computer won't be doing anything else when the filesystem is being used. The test was on a Sun x4200(?) aka Thumper. Multiple Opeterons with dual cores. IMHO we're going to see more more cores in a system by default and most software will not be keeping up. I think the OS will be able to take advantage of the multiple cores and apps will not. Unless you're running lots of apps, cores will be idle. I'd rather have one of those cores consumed by software raid instead of being idle. Maybe that won't be quite as fast as a dedicated hardware raid but it's in my system already so I can buy a $20 multi SATA port card instead of a $xxx multi SATA RAID card. The bottleneck will be, IMO, I/O. Disk data will go though that no matter if you have hardware or software raid. ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: ZFS vs EXT4 vs XFS
On 8/31/07, Tom Buskey [EMAIL PROTECTED] wrote: IMHO we're going to see more more cores in a system by default ... Sure, if you've actually got a surplus of cores. Going forward, for most small systems, that's going to be true. But it's not a given for everything today. That's all I'm saying. -- Ben ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: ZFS vs EXT4 vs XFS
I regard most such storage-related benchmarks with a great deal of suspicion. They always seem to assume the computer won't be doing anything else when the filesystem is being used. Well said. Amplifying ... ALL benchmarks are at best hints of reality, since they're ALL over-simplifications. It takes an actual workload simulation to properly benchmark a balanced system design (like IBM Power Series P) in comparison to systems where each subsystem was tuned to a simplistic benchmark. -- Bill [EMAIL PROTECTED] [EMAIL PROTECTED] ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: ZFS vs EXT4 vs XFS
On 8/31/07, Ben Scott [EMAIL PROTECTED] wrote: On 8/31/07, Tom Buskey [EMAIL PROTECTED] wrote: IMHO we're going to see more more cores in a system by default ... Sure, if you've actually got a surplus of cores. Going forward, for most small systems, that's going to be true. But it's not a given for everything today. That's all I'm saying. Hence my saying we're going to see in above. Most new computers and CPUs I see advertised today are dual core or more. ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: ZFS vs EXT4 vs XFS
On 8/31/07, Tom Buskey [EMAIL PROTECTED] wrote: ... Going forward, for most small systems, that's going to be true. ... ... Hence my saying *we're going to see* in above. ... Hence my saying that's going to be true in above. ;-) -- Ben ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: ZFS vs EXT4 vs XFS
The bottleneck will be, IMO, I/O. Disk data will go though that no matter if you have hardware or software raid. The bottle neck has and always will be I/O. I doubt the day will ever come that a fetch out to the disk will take the same time as a fetch out to memory. Some of the latest flash drives make it faster, but still no where near the time it takes to go out to memory. Dan ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: ZFS vs EXT4 vs XFS
On 8/31/07, Dan Miller [EMAIL PROTECTED] wrote: The bottle neck has and always will be I/O. I doubt the day will ever come that a fetch out to the disk will take the same time as a fetch out to memory. That assumes the constraints are not cumulative, and that workload is a fungible thing, equally affected by both I/O and CPU wait. A process which is I/O bound and light on the CPU will likely behave in the manner you describe. Say, copying a file. The CPU isn't doing anything hard anyway. The system is just pushing bytes through buffers. If the CPU is put to work doing storage management, so much the better. On the other hand, something which is keeping CPU busy while also doing some I/O (say, processing of a dataset), may well find that latency stacks up, as the throughput is delayed first by an I/O wait, and then a processor wait. -- Ben ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: ZFS vs EXT4 vs XFS
On 8/31/07, Ben Scott [EMAIL PROTECTED] wrote: On the other hand, something which is keeping CPU busy while also doing some I/O (say, processing of a dataset), may well find that latency stacks up, as the throughput is delayed first by an I/O wait, and then a processor wait. It may be worth pointing out that I don't necessarily discount the software approach to storage management. If I've got a workload that is CPU-bound, and it is being dragged down by I/O wait due to storage management being done in software, I've got two options: Buy dedicated storage controllers, or buy more general-purpose cores. The GP cores may well be the more effective route. In addition to often being cheaper, scaling better, and distributing better, I can use GP cores for other things when I/O isn't an issue. -- Ben ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
ZFS vs EXT4 vs XFS
Interesting benchmarks: http://tastic.brillig.org/%7Ejwb/zfs-xfs-ext4.html I would also like to see Reiserfs4 in those results to see how it compares, but those are not included. Dan ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
