Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?)
I've seen quite a few folks touting the Opteron as 2.5x faster with postgres than a Xeon box. What makes the Opteron so quick? Is it that Postgres really prefers to run in 64-bit mode? I don't know about 2.5x faster (perhaps on specific types of loads), but the reason Opterons rock for database applications is their insanely good memory bandwidth and latency that scales much better than the Xeon. Opterons also have a ccNUMA-esque I/O fabric and two dedicated on-die memory channels *per processor* -- no shared bus there, closer to real UNIX server iron than a glorified PC. We run a large Postgres database on a dual Opteron in 32-bit mode that crushes Xeons running at higher clock speeds. It has little to do with bitness or theoretical instruction dispatch, and everything to do with the superior memory controller and I/O fabric. Databases are all about moving chunks of data around and the Opteron systems were engineered to do this very well and in a very scalable fashion. For the money, it is hard to argue with the price/performance of Opteron based servers. We started with one dual Opteron postgres server just over a year ago (with an equivalent uptime) and have considered nothing but Opterons for database servers since. Opterons really are clearly superior to Xeons for this application. I don't work for AMD, just a satisfied customer. :-) re: 6 disks. Unless you are tight on disk space, a hot spare might be nice as well depending on your needs. Cheers, J. Andrew Rogers ---(end of broadcast)--- TIP 1: subscribe and unsubscribe commands go to [EMAIL PROTECTED]
Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?)
On Tue, 19 Apr 2005, J. Andrew Rogers wrote: I don't know about 2.5x faster (perhaps on specific types of loads), but the reason Opterons rock for database applications is their insanely good memory bandwidth and latency that scales much better than the Xeon. Opterons also have a ccNUMA-esque I/O fabric and two dedicated on-die memory channels *per processor* -- no shared bus there, closer to real UNIX server iron than a glorified PC. Thanks J! That's exactly what I was suspecting it might be. Actually, I found an anandtech benchmark that shows the Opteron coming in at close to 2.0x performance: http://www.anandtech.com/linux/showdoc.aspx?i=2163p=2 It's an Opteron 150 (2.4ghz) vs. Xeon 3.6ghz from August. I wonder if the differences are more pronounced with the newer Opterons. -Jeff ---(end of broadcast)--- TIP 9: the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match
Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?)
I posted this link a few months ago and there was some surprise over the difference in postgresql compared to other DBs. (Not much surprise in Opteron stomping on Xeon in pgsql as most people here have had that experience -- the surprise was in how much smaller the difference was in other DBs.) If it was across the board +100% in MS-SQL, MySQL, etc -- you can chalk in up to overall better CPU architecture. Most of the time though, the numbers I've seen show +0-30% for [insert DB here] and a huge whopping + for pgsql. Why the pronounced preference for postgresql, I'm not sure if it was explained fully. BTW, the Anandtech test compares single CPU systems w/ 1GB of RAM. Go to dual/quad and SMP Xeon will suffer even more since it has to share a fixed amount of FSB/memory bandwidth amongst all CPUs. Xeons also seem to suffer more from context-switch storms. Go 4GB of RAM and the Xeon suffers another hit due to the lack of a 64-bit IOMMU. Devices cannot map to addresses 4GB which means the OS has to do extra work in copying data from/to 4GB anytime you have IO. (Although this penalty might exist all the time in 64-bit mode for Xeon if Linux/Windows took the expedient and less-buggy route of using a single method versus checking whether target addresses are or 4GB.) Jeff Frost wrote: On Tue, 19 Apr 2005, J. Andrew Rogers wrote: I don't know about 2.5x faster (perhaps on specific types of loads), but the reason Opterons rock for database applications is their insanely good memory bandwidth and latency that scales much better than the Xeon. Opterons also have a ccNUMA-esque I/O fabric and two dedicated on-die memory channels *per processor* -- no shared bus there, closer to real UNIX server iron than a glorified PC. Thanks J! That's exactly what I was suspecting it might be. Actually, I found an anandtech benchmark that shows the Opteron coming in at close to 2.0x performance: http://www.anandtech.com/linux/showdoc.aspx?i=2163p=2 It's an Opteron 150 (2.4ghz) vs. Xeon 3.6ghz from August. I wonder if the differences are more pronounced with the newer Opterons. -Jeff ---(end of broadcast)--- TIP 9: the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match ---(end of broadcast)--- TIP 6: Have you searched our list archives? http://archives.postgresql.org
Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?)
On Apr 20, 2005, at 12:40 AM, Jeff Frost wrote: I've seen quite a few folks touting the Opteron as 2.5x faster with postgres than a Xeon box. What makes the Opteron so quick? Is it that Postgres really prefers to run in 64-bit mode? The I/O path on the opterons seems to be much faster, and having 64-bit all the way to the disk controller helps... just be sure to run a 64-bit version of your OS. Vivek Khera, Ph.D. +1-301-869-4449 x806 smime.p7s Description: S/MIME cryptographic signature
Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?)
In terms of vendor specific models - Does anyone have any good/bad experiences/recommendations for a 4-way Opteron from Sun (v40z, 6 internal drives) or HP (DL585 5 internal drives) models? This is in comparison with the new Dell 6850 (it has PCIexpress, faster FSB 667MHz, which doesn't match up with AMD's total IO bandwidth, but much better than previous 6650s). Thanks, Anjan -Original Message- From: William Yu [mailto:[EMAIL PROTECTED] Sent: Wednesday, April 20, 2005 11:10 AM To: pgsql-performance@postgresql.org Subject: Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?) I posted this link a few months ago and there was some surprise over the difference in postgresql compared to other DBs. (Not much surprise in Opteron stomping on Xeon in pgsql as most people here have had that experience -- the surprise was in how much smaller the difference was in other DBs.) If it was across the board +100% in MS-SQL, MySQL, etc -- you can chalk in up to overall better CPU architecture. Most of the time though, the numbers I've seen show +0-30% for [insert DB here] and a huge whopping + for pgsql. Why the pronounced preference for postgresql, I'm not sure if it was explained fully. BTW, the Anandtech test compares single CPU systems w/ 1GB of RAM. Go to dual/quad and SMP Xeon will suffer even more since it has to share a fixed amount of FSB/memory bandwidth amongst all CPUs. Xeons also seem to suffer more from context-switch storms. Go 4GB of RAM and the Xeon suffers another hit due to the lack of a 64-bit IOMMU. Devices cannot map to addresses 4GB which means the OS has to do extra work in copying data from/to 4GB anytime you have IO. (Although this penalty might exist all the time in 64-bit mode for Xeon if Linux/Windows took the expedient and less-buggy route of using a single method versus checking whether target addresses are or 4GB.) Jeff Frost wrote: On Tue, 19 Apr 2005, J. Andrew Rogers wrote: I don't know about 2.5x faster (perhaps on specific types of loads), but the reason Opterons rock for database applications is their insanely good memory bandwidth and latency that scales much better than the Xeon. Opterons also have a ccNUMA-esque I/O fabric and two dedicated on-die memory channels *per processor* -- no shared bus there, closer to real UNIX server iron than a glorified PC. Thanks J! That's exactly what I was suspecting it might be. Actually, I found an anandtech benchmark that shows the Opteron coming in at close to 2.0x performance: http://www.anandtech.com/linux/showdoc.aspx?i=2163p=2 It's an Opteron 150 (2.4ghz) vs. Xeon 3.6ghz from August. I wonder if the differences are more pronounced with the newer Opterons. -Jeff ---(end of broadcast)--- TIP 9: the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match ---(end of broadcast)--- TIP 6: Have you searched our list archives? http://archives.postgresql.org ---(end of broadcast)--- TIP 7: don't forget to increase your free space map settings
Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?)
Anjan Dave wrote: In terms of vendor specific models - Does anyone have any good/bad experiences/recommendations for a 4-way Opteron from Sun (v40z, 6 internal drives) or HP (DL585 5 internal drives) models? This is in comparison with the new Dell 6850 (it has PCIexpress, faster FSB 667MHz, which doesn't match up with AMD's total IO bandwidth, but much better than previous 6650s). Dell cuts too many corners to be a good server. -- Bruce Momjian| http://candle.pha.pa.us pgman@candle.pha.pa.us | (610) 359-1001 + If your life is a hard drive, | 13 Roberts Road + Christ can be your backup.| Newtown Square, Pennsylvania 19073 ---(end of broadcast)--- TIP 2: you can get off all lists at once with the unregister command (send unregister YourEmailAddressHere to [EMAIL PROTECTED])
Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?)
On Wednesday 20 April 2005 17:50, Bruce Momjian wrote: Anjan Dave wrote: In terms of vendor specific models - Does anyone have any good/bad experiences/recommendations for a 4-way Opteron from Sun (v40z, 6 internal drives) or HP (DL585 5 internal drives) models? This is in comparison with the new Dell 6850 (it has PCIexpress, faster FSB 667MHz, which doesn't match up with AMD's total IO bandwidth, but much better than previous 6650s). Dell cuts too many corners to be a good server. Hi Which corners do Dell cut compared to the competition ? Thanks Christian ---(end of broadcast)--- TIP 2: you can get off all lists at once with the unregister command (send unregister YourEmailAddressHere to [EMAIL PROTECTED])
Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?)
There have been some discussions on this list and others in general about Dell's version of RAID cards, and server support, mainly linux support. Before I venture into having another vendor in the shop I want to know if there are any dos/don't's about 4-way Opteron offerings from Sun and HP. Don't want to put the topic on a different tangent, but I would be interested in the discussion of AMD Vs. XEON in terms of actual products available today. Thanks, Anjan -Original Message- From: Christian Sander Røsnes [mailto:[EMAIL PROTECTED] Sent: Wednesday, April 20, 2005 12:14 PM To: Bruce Momjian Cc: pgsql-performance@postgresql.org Subject: Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?) On Wednesday 20 April 2005 17:50, Bruce Momjian wrote: Anjan Dave wrote: In terms of vendor specific models - Does anyone have any good/bad experiences/recommendations for a 4-way Opteron from Sun (v40z, 6 internal drives) or HP (DL585 5 internal drives) models? This is in comparison with the new Dell 6850 (it has PCIexpress, faster FSB 667MHz, which doesn't match up with AMD's total IO bandwidth, but much better than previous 6650s). Dell cuts too many corners to be a good server. Hi Which corners do Dell cut compared to the competition ? Thanks Christian ---(end of broadcast)--- TIP 2: you can get off all lists at once with the unregister command (send unregister YourEmailAddressHere to [EMAIL PROTECTED]) ---(end of broadcast)--- TIP 8: explain analyze is your friend
Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?)
Anjan, Does anyone have any good/bad experiences/recommendations for a 4-way Opteron from Sun (v40z, 6 internal drives) or HP (DL585 5 internal drives) models? Last I checked, the v40z only takes 5 drives, unless you yank the cd-rom and get an extra disk tray. That's the main defect of the model, the second being its truly phenominal noise level. Other than that (and price) and excellent Opteron machine. The HPs are at root pretty good machines -- and take 6 drives, so I expect you're mixed up there. However, they use HP's proprietary RAID controller which is seriously defective. So you need to factor replacing the RAID controller into the cost. This is in comparison with the new Dell 6850 (it has PCIexpress, faster FSB 667MHz, which doesn't match up with AMD's total IO bandwidth, but much better than previous 6650s). Yes, but you can still expect the 6650 to have 1/2 the performance ... or less ... of the above-name models. It: 1) is Xeon 32-bit 2) uses a cheap northbridge which makes the Xeon's cache contention even worse 3) depending on the model and options, may ship with a cheap Adaptec raid card instead of an LSI or other good card If all you *need* is 1/2 the performance of an Opteron box, and you can get a good deal, then go for it. But don't be under the illusion that Dell is competitive with Sun, IBM, HP, Penguin or Microway on servers. -- --Josh Josh Berkus Aglio Database Solutions San Francisco ---(end of broadcast)--- TIP 1: subscribe and unsubscribe commands go to [EMAIL PROTECTED]
Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?)
The HPs are at root pretty good machines -- and take 6 drives, so I expect you're mixed up there. However, they use HP's proprietary RAID controller which is seriously defective. So you need to factor replacing the RAID controller into the cost. Do you have any additional materials on what is defective with their raid controllers? -- ---(end of broadcast)--- TIP 7: don't forget to increase your free space map settings
Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?)
On 4/20/05, Anjan Dave [EMAIL PROTECTED] wrote: In terms of vendor specific models - Does anyone have any good/bad experiences/recommendations for a 4-way Opteron from Sun (v40z, 6 internal drives) or HP (DL585 5 internal drives) models? We are going with the 90nm HPs for production. They feel like beefier boxes than the Suns, but the Suns cost a LOT less, IIRC. We're only using the internal drives for the OS. PG gets access to a fibre-channel array, HP StorageWorks 3000. I _can't wait_ to get this in. Our dev box is a 130nm DL585 with 16G of RAM and an HP SCSI array, and I have absolutely zero complaints. :) -- Mike Rylander [EMAIL PROTECTED] GPLS -- PINES Development Database Developer http://open-ils.org ---(end of broadcast)--- TIP 7: don't forget to increase your free space map settings
Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?)
There have been some discussions on this list and others in general about Dell's version of RAID cards, and server support, mainly linux support. I was pretty impressed with the Dell guy. He spent the day with me remotely and went through my system 6650 with powervault. Changed my drives from ext3 to ext2 with no journaling checked all the drivers and such. I did not see any marked improvement, but I dont think my issues are related to the hardware. I am giving up on postgres and three developers two months of work and trying MYSQL. I have posted several items and not got a response (not that I expect folks to drop everything). I want to thank everyone who has been of help and there are several. It just is running way slow on several of my views. I tried them today in MYSQL and found that the MYSQL was beating out my MSSQL. On certain items I could get PG to work ok, but it never was faster the MSSQL. On certain items it is taking several minutes compared to a few seconds on MYSQL. I really like the environment and feel I have learned a lot in the past few months, but bottom line for me is speed. We bought a 30K Dell 6650 to get better performance. I chose PG because MSSQL was 70K to license. I believe the MYSQL will be 250.00 to license for us, but I may choose the 4k platinum support just to feel safe about having some one to touch base with in the event of an issue. Again thanks to everyone who has answered my newb questions and helped me get it on the 3 spindles and tweek the install. Commandpromt.com was a big help and if I wanted to budget a bunch more $ and mostly if I was at liberty to share my database with them they may of helped me get through all the issues. I am not sure I am walking away feeling real good about postgres, because it just should not take a rocket scientist to get it to work, and I used to think I was fairly smart and could figure stuff out and I hate admitting defeat (especially since we have everything working with postgres now). ---(end of broadcast)--- TIP 6: Have you searched our list archives? http://archives.postgresql.org
Re: [PERFORM] Opteron vs Xeon (Was: What to do with 6 disks?)
Josh Berkus josh@agliodbs.com writes: Last I checked, the v40z only takes 5 drives, unless you yank the cd-rom and get an extra disk tray. That's the main defect of the model, the second being its truly phenominal noise level. Other than that (and price) and excellent Opteron machine. Incidentally, Sun sells a bunch of v20z and v40z machines on Ebay as some kind of marketing strategy. You can pick one up for only a slightly absurd price if you're happy with the configurations listed there. (And if you're in the US). -- greg ---(end of broadcast)--- TIP 3: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to [EMAIL PROTECTED] so that your message can get through to the mailing list cleanly
