Re: [squid-users] Recommend for hardware configurations
Well, I based my argument from the 10 instances of reverse proxies I'm running. It has 266,268,230 objects and 3.7 TB of space. CPU usage is always around 0.2 according to ganglia. So unless you have some other statistics to prove CPU is that important, I'm stick w/ my argument that disk and RAM is way more important that CPU. ok, reverse proxy does not so very much, so sure it depends on what you do with the machine mike At 03:41 AM 7/6/2008, Michel wrote: The cpu doesn't do any IO, it's WAITING for the disk most of the time. If you want fast squid performance, CPU speed/count is irrelevant; get more disks and ram. When I mean more disk, I mean more spindles. eg: 2x 100GB will is better than a 200GB disk. well well, get prepared ... take your cpu out and then you'll see who is waiting forever :) even if IO wait is an issue it is or better WAS one on old giant lock systems where the cpu was waiting until getting the lock on a busy thread because there was only ONE CPU and even on multi-cpu-systems there was only one core a time bound to the kernel to get around this issue good old posix aio_*calls where used in order not to wait for a new lock what I believe is squid's aufs cache_dir model which is still very good and even better on modern smp machines and even with squid's not-smp-optimized code - you really can drain disks to their physical limits - but that is not all SMP (modern) works around the global giant lock, the kernel is not anymore limited to get one core a time SMP sistems are going to work with spin locks (Linux) and sleep locks (freebsd) where the linux way is focusing thread synchronizing which is going to be outperformanced by the sleep lock mechanism. Spin locks certainly still waste cpu while spinning what sleeplocks do not, cpu is free to do other work. This was kind of benefit for Linux last couple of years when freebsd was in deep developing of it's new threading model which is now on top I think, especially in shared memory environments. basicly is it not important if you use one or ten disks, this you should consider later as fine tuning but the threading model works the same, for one or two disks, or for 2 our 32Gigs of memory - so you certainly do NOT get araound your IO-Wait with more memory or more disk when the cpu(s) can not handle it waiting for locks as you say ... So IMO your statement is not so very true anymore, with a modern SMP-OS on modern smp hardware of course. michel Tecnologia Internet Matik http://info.matik.com.br Sistemas Wireless para o Provedor Banda Larga Hospedagem e Email personalizado - e claro, no Brasil. A mensagem foi scaneada pelo sistema de e-mail e pode ser considerada segura. Service fornecido pelo Datacenter Matik https://datacenter.matik.com.br michel Tecnologia Internet Matik http://info.matik.com.br Sistemas Wireless para o Provedor Banda Larga Hospedagem e Email personalizado - e claro, no Brasil.
Re: [squid-users] Recommend for hardware configurations
On mån, 2008-07-07 at 07:34 -0300, Michel wrote: ok, reverse proxy does not so very much, so sure it depends on what you do with the machine The known configurations which can easily push Squid to CPU bound limits is a) reverse proxy setups with a reasonably small but very frequently accessed set of objects. b) reverse proxy acting as an SSL frontend. c) forward proxies without any cache. It's quite hard to push a caching forward proxy to the CPU limit. You usually run into other limits first.. Regards Henrik signature.asc Description: This is a digitally signed message part
Re: [squid-users] Recommend for hardware configurations
On lör, 2008-07-05 at 12:44 -0300, Michel wrote: I am not understanding why you keep suggesting single core as preferred cpu Did I? Not what I can tell. I said Squid uses only one core. even if squid's core is actually not multi-thread capable a faster cpu is better - there are also other things running on a machine so a smp machine ever is a benefit to overall performance Both yes and no. For an application like Squid you will find that nearly all OS:es gets bound to a single core running both networking and the application, leaving the other cores to run various tiny other stuff.. modern OS also should give squid's aufs threading benefits but I am not totally aufs isn't very cpu hungry. It's main purpose is to be able to exploit the parallellism there is in the harddrive hardware. The Squid cache function gets quite seek intensive so there is a huge benefit of being able to have multiple concurrent I/O operations (especially open/create calls...). diskd also isn't very cpu hungry. In fact probably a bit less than aufs. But diskd can not push the drives as far as aufs, and is still plauged some instability issues.. Why I recommend dual core instead of quad core is simply because you get a faster core speed in dual core than quad core for the same price (and often availability as well..) which will directly benefit Squid in high performance. Yes, Squid quite easily gets CPU bound, and is then limited to the core speed of your CPU, and the faster the core speed is the better in that situation. Selecting a slower core speed to fit more cores hurts performance for Squid when the server is mainly for Squid. sure about your design here but at least diskd when running several diskd processes is getting benefits from multicore cpus - and a lot and if you do not believe it set up squid/disk on a 8-core machine and compare with 1|2|4|8 or more diskds to your single-core-cpu-thing and measure it, in fact you do not even measure it, you can see it and smell it ... You are welcome to give numbers proving that for Squid a 4 core system outperforms a 2 core system with the exact same configuration in all other aspects. Don't forget to include price in the matrix.. The most interesting test configurations is - no disk cache - single drive for disk cache - 4 drives for disk cache Until I see any numbers indicating quad core gives a significant increase outperforming what the same price configuration using dual core I will continue propagating that quad core is not beneficial to Squid. Similarly for dual core vs single core, but it's not as clear cut as there is not a big per core performance difference between single and dual core compared to prices.. Regards Henrik signature.asc Description: This is a digitally signed message part
Re: [squid-users] Recommend for hardware configurations
The cpu doesn't do any IO, it's WAITING for the disk most of the time. If you want fast squid performance, CPU speed/count is irrelevant; get more disks and ram. When I mean more disk, I mean more spindles. eg: 2x 100GB will is better than a 200GB disk. well well, get prepared ... take your cpu out and then you'll see who is waiting forever :) even if IO wait is an issue it is or better WAS one on old giant lock systems where the cpu was waiting until getting the lock on a busy thread because there was only ONE CPU and even on multi-cpu-systems there was only one core a time bound to the kernel to get around this issue good old posix aio_*calls where used in order not to wait for a new lock what I believe is squid's aufs cache_dir model which is still very good and even better on modern smp machines and even with squid's not-smp-optimized code - you really can drain disks to their physical limits - but that is not all SMP (modern) works around the global giant lock, the kernel is not anymore limited to get one core a time SMP sistems are going to work with spin locks (Linux) and sleep locks (freebsd) where the linux way is focusing thread synchronizing which is going to be outperformanced by the sleep lock mechanism. Spin locks certainly still waste cpu while spinning what sleeplocks do not, cpu is free to do other work. This was kind of benefit for Linux last couple of years when freebsd was in deep developing of it's new threading model which is now on top I think, especially in shared memory environments. basicly is it not important if you use one or ten disks, this you should consider later as fine tuning but the threading model works the same, for one or two disks, or for 2 our 32Gigs of memory - so you certainly do NOT get araound your IO-Wait with more memory or more disk when the cpu(s) can not handle it waiting for locks as you say ... So IMO your statement is not so very true anymore, with a modern SMP-OS on modern smp hardware of course. michel Tecnologia Internet Matik http://info.matik.com.br Sistemas Wireless para o Provedor Banda Larga Hospedagem e Email personalizado - e claro, no Brasil.
Re: [squid-users] Recommend for hardware configurations
On lör, 2008-07-05 at 12:44 -0300, Michel wrote: I am not understanding why you keep suggesting single core as preferred cpu Did I? Not what I can tell. I said Squid uses only one core. :) good answer ... but often it does not matter what we say but what is beeing understood, what I meant is that it comes over as if you are suggesting single core computers even if squid's core is actually not multi-thread capable a faster cpu is better - there are also other things running on a machine so a smp machine ever is a benefit to overall performance Both yes and no. For an application like Squid you will find that nearly all OS:es gets bound to a single core running both networking and the application, leaving the other cores to run various tiny other stuff.. nope, not at all. probably on Linux's spin lock model it might be so , but I do not know, on freebsd you can watch the squid process and it's threads, either aufs or diskd related and see that they are handled by all cpus all the time 35867 squid4 -19 3921M 3868M kqread 3 200:28 0.00% squid0 1481 squid4 -19 601M 581M kqread 0 86:03 0.00% squid1 1482 squid4 -19 598M 579M kqread 0 84:49 0.00% squid2 1495 squid -4 -19 8300K 1376K msgrcv 1 20:19 0.00% diskd-daemon 1496 squid -4 -19 8300K 1372K msgrcv 3 20:11 0.00% diskd-daemon 1497 squid -4 -19 8300K 1324K msgrcv 3 5:42 0.00% diskd-daemon 1498 squid -4 -19 8300K 1224K msgrcv 2 5:31 0.00% diskd-daemon 35867 squid4 -19 3921M 3868M kqread 1 200:28 0.00% squid0 1481 squid4 -19 601M 581M kqread 1 86:03 0.00% squid1 1482 squid4 -19 598M 579M kqread 1 84:49 0.00% squid2 1495 squid -4 -19 8300K 1376K msgrcv 0 20:19 0.00% diskd-daemon 1496 squid -4 -19 8300K 1372K msgrcv 0 20:11 0.00% diskd-daemon 1497 squid -4 -19 8300K 1324K msgrcv 2 5:42 0.00% diskd-daemon 1498 squid -4 -19 8300K 1224K msgrcv 2 5:31 0.00% diskd-daemon 35867 squid4 -19 3921M 3868M kqread 1 200:29 0.00% squid0 1481 squid4 -19 601M 581M kqread 2 86:03 0.00% squid1 1482 squid4 -19 598M 579M kqread 3 84:50 0.00% squid2 1495 squid -4 -19 8300K 1376K msgrcv 1 20:19 0.00% diskd-daemon 1496 squid -4 -19 8300K 1372K msgrcv 1 20:11 0.00% diskd-daemon 1497 squid -4 -19 8300K 1324K msgrcv 2 5:42 0.00% diskd-daemon 1498 squid -4 -19 8300K 1224K msgrcv 2 5:31 0.00% diskd-daemon three tops in 3 different seconds, 8'th column show on which cpu it runs, observing threads it still is more fun Why I recommend dual core instead of quad core is simply because you get a faster core speed in dual core than quad core for the same price (and often availability as well..) which will directly benefit Squid in high performance. I understood you recommend single core ... not dual Yes, Squid quite easily gets CPU bound, and is then limited to the core speed of your CPU, and the faster the core speed is the better in that situation. Selecting a slower core speed to fit more cores hurts performance for Squid when the server is mainly for Squid. I am not so sure if the core speed does matter so much as long as there IS CPU left ... then there is CPU left for any other work... You are welcome to give numbers proving that for Squid a 4 core system outperforms a 2 core system with the exact same configuration in all other aspects. Don't forget to include price in the matrix.. The most interesting test configurations is - no disk cache - single drive for disk cache - 4 drives for disk cache Until I see any numbers indicating quad core gives a significant increase outperforming what the same price configuration using dual core I will continue propagating that quad core is not beneficial to Squid. two/three years ago I said next year there are no single cores to buy anymore and everyone is running at least dualcore if not quad and was shot by almost all freebsd 4.x and dragon-fly lovers or should I say by people which didn't saw where the modern threading model was going and were hanging on to the global giant lock because at *THAT* time network and disk performance was still better? then to be honest I do not believe that you ever will be convinced by any test *I* post here :), so do it yourself and get your own conclusions ... the test is easy, get yourself an AM2 MB and a X2 and a X4 and nuke a fixed rate of http requests over a certain time into each CPU and monitor CPU time and disk io (on FREEBSD amd64 7STABLE ) and compare it and I say you show me that X2 is losing and then I get myself a linux box and shut my mouth :) Similarly for dual core vs single core, but it's not as clear cut as there is not a big per core performance difference between single and dual core compared to prices.. as I said soon there are no single cores
Re: [squid-users] Recommend for hardware configurations
* Henrik Nordstrom [EMAIL PROTECTED]: diskd also isn't very cpu hungry. In fact probably a bit less than aufs. But diskd can not push the drives as far as aufs, and is still plauged some instability issues.. So you say aufs is defacto FASTER than diskd? -- Ralf Hildebrandt (i.A. des IT-Zentrums) [EMAIL PROTECTED] Charite - Universitätsmedizin BerlinTel. +49 (0)30-450 570-155 Gemeinsame Einrichtung von FU- und HU-BerlinFax. +49 (0)30-450 570-962 IT-Zentrum Standort CBF send no mail to [EMAIL PROTECTED]
Re: [squid-users] Recommend for hardware configurations
2008/7/6 Henrik Nordstrom [EMAIL PROTECTED]: Until I see any numbers indicating quad core gives a significant increase outperforming what the same price configuration using dual core I will continue propagating that quad core is not beneficial to Squid. My numbers here say, if your Squid workload is memory-locked, you won't benefit from quad-core. If your Squid workload is CPU-bound, you will benefit from Quad-Core. This applies for intel with their single style memory bus. AMD's are a bit different. Similarly for dual core vs single core, but it's not as clear cut as there is not a big per core performance difference between single and dual core compared to prices.. Well, the benefit comes from being able to run the network TX/RX threads on the other CPU, freeing up the first CPU for running Squid. You'd be surprised how much that will give you. Of course, if Squid wasn't so inefficient when it came to memory copying (ie, it does way too much), it would probably be more CPU rather than memory-bus taxing, and quad-core may start to matter. YMMV. Adrian (Who has single and dual-core AMD/Intel, and Dual dual-core intel xeon in testing, and understands what the hell is going on here..)
Re: [squid-users] Recommend for hardware configurations
Well, I based my argument from the 10 instances of reverse proxies I'm running. It has 266,268,230 objects and 3.7 TB of space. CPU usage is always around 0.2 according to ganglia. So unless you have some other statistics to prove CPU is that important, I'm stick w/ my argument that disk and RAM is way more important that CPU. mike At 03:41 AM 7/6/2008, Michel wrote: The cpu doesn't do any IO, it's WAITING for the disk most of the time. If you want fast squid performance, CPU speed/count is irrelevant; get more disks and ram. When I mean more disk, I mean more spindles. eg: 2x 100GB will is better than a 200GB disk. well well, get prepared ... take your cpu out and then you'll see who is waiting forever :) even if IO wait is an issue it is or better WAS one on old giant lock systems where the cpu was waiting until getting the lock on a busy thread because there was only ONE CPU and even on multi-cpu-systems there was only one core a time bound to the kernel to get around this issue good old posix aio_*calls where used in order not to wait for a new lock what I believe is squid's aufs cache_dir model which is still very good and even better on modern smp machines and even with squid's not-smp-optimized code - you really can drain disks to their physical limits - but that is not all SMP (modern) works around the global giant lock, the kernel is not anymore limited to get one core a time SMP sistems are going to work with spin locks (Linux) and sleep locks (freebsd) where the linux way is focusing thread synchronizing which is going to be outperformanced by the sleep lock mechanism. Spin locks certainly still waste cpu while spinning what sleeplocks do not, cpu is free to do other work. This was kind of benefit for Linux last couple of years when freebsd was in deep developing of it's new threading model which is now on top I think, especially in shared memory environments. basicly is it not important if you use one or ten disks, this you should consider later as fine tuning but the threading model works the same, for one or two disks, or for 2 our 32Gigs of memory - so you certainly do NOT get araound your IO-Wait with more memory or more disk when the cpu(s) can not handle it waiting for locks as you say ... So IMO your statement is not so very true anymore, with a modern SMP-OS on modern smp hardware of course. michel Tecnologia Internet Matik http://info.matik.com.br Sistemas Wireless para o Provedor Banda Larga Hospedagem e Email personalizado - e claro, no Brasil.
Re: [squid-users] Recommend for hardware configurations
On tor, 2008-07-03 at 12:04 +0800, Roy M. wrote: We are planning to replace this testing server with two or three cheaper 1U servers (sort of redundancy!) Intel Dual Core or Quad Core CPU x1 (no SMP) Squid uses only one core, so rather Dual core than Quad... I am not understanding why you keep suggesting single core as preferred cpu even if squid's core is actually not multi-thread capable a faster cpu is better - there are also other things running on a machine so a smp machine ever is a benefit to overall performance modern OS also should give squid's aufs threading benefits but I am not totally sure about your design here but at least diskd when running several diskd processes is getting benefits from multicore cpus - and a lot and if you do not believe it set up squid/disk on a 8-core machine and compare with 1|2|4|8 or more diskds to your single-core-cpu-thing and measure it, in fact you do not even measure it, you can see it and smell it ... and at the buttom line more power more performance so there is no way a single core runs faster than a dual or quad core on a modern OS, not even get close to it michel Tecnologia Internet Matik http://info.matik.com.br Sistemas Wireless para o Provedor Banda Larga Hospedagem e Email personalizado - e claro, no Brasil.
Re: [squid-users] Recommend for hardware configurations
Michel wrote: On tor, 2008-07-03 at 12:04 +0800, Roy M. wrote: We are planning to replace this testing server with two or three cheaper 1U servers (sort of redundancy!) Intel Dual Core or Quad Core CPU x1 (no SMP) Squid uses only one core, so rather Dual core than Quad... I am not understanding why you keep suggesting single core as preferred cpu even if squid's core is actually not multi-thread capable a faster cpu is better - there are also other things running on a machine so a smp machine ever is a benefit to overall performance modern OS also should give squid's aufs threading benefits but I am not totally sure about your design here but at least diskd when running several diskd processes is getting benefits from multicore cpus - and a lot and if you do not believe it set up squid/disk on a 8-core machine and compare with 1|2|4|8 or more diskds to your single-core-cpu-thing and measure it, in fact you do not even measure it, you can see it and smell it ... and at the buttom line more power more performance so there is no way a single core runs faster than a dual or quad core on a modern OS, not even get close to it I agree. To give a little bit more specific advise: Squid is mostly single-threaded and in large environments needs a lot of CPU power. So I recommend to buy a system with at least 2 cores and maximize CPU cache and memory speed (use 2 or 4 DIMMs) to get the most of the raw CPU power. For example, go for a Dell R300 with Quad-core 2.66 GHz, 6 MB cache for a good price (EUR 749 ex VAT base price). Dell is just an example, I am not affiliated. -Marcus michel Tecnologia Internet Matik http://info.matik.com.br Sistemas Wireless para o Provedor Banda Larga Hospedagem e Email personalizado - e claro, no Brasil.
Re: [squid-users] Recommend for hardware configurations
At 12:09 PM 7/5/2008, Marcus Kool wrote: Michel wrote: On tor, 2008-07-03 at 12:04 +0800, Roy M. wrote: We are planning to replace this testing server with two or three cheaper 1U servers (sort of redundancy!) Intel Dual Core or Quad Core CPU x1 (no SMP) Squid uses only one core, so rather Dual core than Quad... I am not understanding why you keep suggesting single core as preferred cpu even if squid's core is actually not multi-thread capable a faster cpu is better - there are also other things running on a machine so a smp machine ever is a benefit to overall performance modern OS also should give squid's aufs threading benefits but I am not totally sure about your design here but at least diskd when running several diskd processes is getting benefits from multicore cpus - and a lot and if you do not believe it set up squid/disk on a 8-core machine and compare with 1|2|4|8 or more diskds to your single-core-cpu-thing and measure it, in fact you do not even measure it, you can see it and smell it ... and at the buttom line more power more performance so there is no way a single core runs faster than a dual or quad core on a modern OS, not even get close to it I agree. To give a little bit more specific advise: Squid is mostly single-threaded and in large environments needs a lot of CPU power. So I recommend to buy a system with at least 2 cores and maximize CPU cache and memory speed (use 2 or 4 DIMMs) to get the most of the raw CPU power. For example, go for a Dell R300 with Quad-core 2.66 GHz, 6 MB cache for a good price (EUR 749 ex VAT base price). Dell is just an example, I am not affiliated. -Marcus michel Tecnologia Internet Matik http://info.matik.com.br Sistemas Wireless para o Provedor Banda Larga Hospedagem e Email personalizado - e claro, no Brasil.
Re: [squid-users] Recommend for hardware configurations
Squid is IO and memory bounded, not cpu bounded. Use the CPU money to buy more RAM/disks mike At 12:09 PM 7/5/2008, Marcus Kool wrote: Michel wrote: On tor, 2008-07-03 at 12:04 +0800, Roy M. wrote: We are planning to replace this testing server with two or three cheaper 1U servers (sort of redundancy!) Intel Dual Core or Quad Core CPU x1 (no SMP) Squid uses only one core, so rather Dual core than Quad... I am not understanding why you keep suggesting single core as preferred cpu even if squid's core is actually not multi-thread capable a faster cpu is better - there are also other things running on a machine so a smp machine ever is a benefit to overall performance modern OS also should give squid's aufs threading benefits but I am not totally sure about your design here but at least diskd when running several diskd processes is getting benefits from multicore cpus - and a lot and if you do not believe it set up squid/disk on a 8-core machine and compare with 1|2|4|8 or more diskds to your single-core-cpu-thing and measure it, in fact you do not even measure it, you can see it and smell it ... and at the buttom line more power more performance so there is no way a single core runs faster than a dual or quad core on a modern OS, not even get close to it I agree. To give a little bit more specific advise: Squid is mostly single-threaded and in large environments needs a lot of CPU power. So I recommend to buy a system with at least 2 cores and maximize CPU cache and memory speed (use 2 or 4 DIMMs) to get the most of the raw CPU power. For example, go for a Dell R300 with Quad-core 2.66 GHz, 6 MB cache for a good price (EUR 749 ex VAT base price). Dell is just an example, I am not affiliated. -Marcus michel Tecnologia Internet Matik http://info.matik.com.br Sistemas Wireless para o Provedor Banda Larga Hospedagem e Email personalizado - e claro, no Brasil.
Re: [squid-users] Recommend for hardware configurations
Squid is IO and memory bounded, not cpu bounded. Use the CPU money to buy more RAM/disks guess who is doing the IO ... :) and get yourself a pricelist, the diff between X2 and phenom is irrelevant and whatever, hardware is cheap michel Tecnologia Internet Matik http://info.matik.com.br Sistemas Wireless para o Provedor Banda Larga Hospedagem e Email personalizado - e claro, no Brasil.
Re: [squid-users] Recommend for hardware configurations
The cpu doesn't do any IO, it's WAITING for the disk most of the time. If you want fast squid performance, CPU speed/count is irrelevant; get more disks and ram. When I mean more disk, I mean more spindles. eg: 2x 100GB will is better than a 200GB disk. mike At 01:51 PM 7/5/2008, Michel wrote: Squid is IO and memory bounded, not cpu bounded. Use the CPU money to buy more RAM/disks guess who is doing the IO ... :) and get yourself a pricelist, the diff between X2 and phenom is irrelevant and whatever, hardware is cheap michel Tecnologia Internet Matik http://info.matik.com.br Sistemas Wireless para o Provedor Banda Larga Hospedagem e Email personalizado - e claro, no Brasil.
Re: [squid-users] Recommend for hardware configurations
lör 2008-07-05 klockan 17:51 -0300 skrev Michel: and get yourself a pricelist, the diff between X2 and phenom is irrelevant and whatever, hardware is cheap True, but the wattage difference is not irrelevant. For the lifetime of the server that adds up quite significantly in power and cooling. Also if you want the highest clock rates (which is what Squid benefits most from) then you won't find these quad core yet for the same reasons.. The fastes Phenom X4 you can get is the Phenom X4 9950, where each core runs at 2.6GHz. The fastest Athlon X2 you can get is the Athlon X2 6400+ where each core runs at 3.2GHz. The fastest quad core Opteron you can get is the Opteron 2360 SE at 2.5GHz. The fastest dual core Opteron is the Opteron 2224 SE at 3.2GHz. Regards Henrik
Re: [squid-users] Recommend for hardware configurations
Hi all, On Sun, Jul 6, 2008 at 7:22 AM, Henrik Nordstrom [EMAIL PROTECTED] wrote: ... The fastest quad core Opteron you can get is the Opteron 2360 SE at 2.5GHz. The fastest dual core Opteron is the Opteron 2224 SE at 3.2GHz. Regards Henrik Why you guy need such a fast Squid? I believe squid is easy to saturate your network interface if serving cache from memory, so consider buying more memory...
Re: [squid-users] Recommend for hardware configurations
On tor, 2008-07-03 at 12:04 +0800, Roy M. wrote: We are planning to replace this testing server with two or three cheaper 1U servers (sort of redundancy!) Intel Dual Core or Quad Core CPU x1 (no SMP) Squid uses only one core, so rather Dual core than Quad... 4GB DDR2 800 RAM 500GB or 750GB SATA (Raid 0) For Squid it's easier with JBOD than RAID0. Performance is the same. Regards Henrik signature.asc Description: This is a digitally signed message part
Re: [squid-users] Recommend for hardware configurations
Hi, On 7/3/08, Henrik Nordstrom [EMAIL PROTECTED] wrote: On tor, 2008-07-03 at 12:04 +0800, Roy M. wrote: Squid uses only one core, so rather Dual core than Quad... But will it help if I am using external redirector? (Currently I don't but maybe later) For Squid it's easier with JBOD than RAID0. Performance is the same. If I only have 2 slots of disks, should I use both disks as cache, or use one for system, and the other one for cache? (E.g. reduce read/write for system disk = improve realiablitiy) THanks.
Re: [squid-users] Recommend for hardware configurations
On tor, 2008-07-03 at 22:11 +0800, Roy M. wrote: Squid uses only one core, so rather Dual core than Quad... But will it help if I am using external redirector? (Currently I don't but maybe later) url rewriters generally do not use a lot of CPU. If I only have 2 slots of disks, should I use both disks as cache, or use one for system, and the other one for cache? (E.g. reduce read/write for system disk = improve realiablitiy) Depends on the I/O load you expect. But yes, it's often a good idea to have the cache separate from OS + logs.. Regards Henrik signature.asc Description: This is a digitally signed message part
