always create the locations that need frequent I/O /home, swapon the outer tracks ?
in [Setup]
|
Prev: A newbie question - trying to resize a given 15GB vdi file to a bigger one
Next: always create the locations that need frequent I/O /home, swapon the outer tracks ?
From: The Natural Philosopher on 17 Jul 2010 07:49 annalissa wrote: > Hi all, > > The following is what i have read in a magazine named "linux for > you" , To what extent this is true ? > > The general rule is that you should always create the locations that > need frequent I/O -/home, swap on the outer tracks , the easiest way > to achieve this is to create these partitions first when partitioning > your hard disk ? > > Reason all modern H.D.D's use a concept called ZCAV(zonal constant > angular velocity). this takes advantage of the fact that more linear > space is available on the outer tracks of the disk platter rather than > on the inside tracks. now since the disk spins at a constant rate , > which is also known as CAV (constant angular velocity) the read/write > I/O speed will be greater at the outer tracks as compared to the inner > tracks Correct, but almost totally irrelevant. since if disk I/O speeds are a problem, you have a real problem anyway. And I would challenge that /home needs the fastest access. in terms of program loading /usr is far more relevant. If /swap is being used extensively, you have a serious memory problem as well. An /tmp is far more likely to be a frequent candidate. And /var. So its a bit of irrelevant truth, a few bad or at least questionable assumptions all cobbled together by someone with nothing better to do to make a 'rule' that is at best worthless, and at worst, highly misleading. And which totally ignores disc caching.
From: Matt Giwer on 18 Jul 2010 00:05 On 07/17/2010 06:38 AM, annalissa wrote: > Hi all, > > The following is what i have read in a magazine named "linux for > you" , To what extent this is true ? > > The general rule is that you should always create the locations that > need frequent I/O -/home, swap on the outer tracks , the easiest way > to achieve this is to create these partitions first when partitioning > your hard disk ? > > Reason all modern H.D.D's use a concept called ZCAV(zonal constant > angular velocity). this takes advantage of the fact that more linear > space is available on the outer tracks of the disk platter rather than > on the inside tracks. now since the disk spins at a constant rate , > which is also known as CAV (constant angular velocity) the read/write > I/O speed will be greater at the outer tracks as compared to the inner > tracks It is unlikely you will notice any improvement. 5400 is 90/second. Not much time no matter how you look at it. But unless you are a sysadm with lots of users calling their files common files many times an hour, think of an office of insurance adjusters, your largest files are in /usr. As folks are always interested in bragging rights on availability (and even a newbie can get 0.999 availability) the boot directory is going to be first and /usr second on a normal install. If you are having response time problems, increase RAM. FWIW, this was a sort of geeky think to do on Windows in the 90s. I did it. Never saw a difference. -- If you could reason with holohuggers there would be no holohuggers. -- The Iron Webmaster, 4281 http://www.giwersworld.org/holo2/ a11 Sat Jul 17 23:41:04 EDT 2010
From: Jean-David Beyer on 18 Jul 2010 08:30 Matt Giwer wrote: > On 07/17/2010 06:38 AM, annalissa wrote: >> Hi all, >> >> The following is what i have read in a magazine named "linux for >> you" , To what extent this is true ? >> >> The general rule is that you should always create the locations that >> need frequent I/O -/home, swap on the outer tracks , the easiest way >> to achieve this is to create these partitions first when partitioning >> your hard disk ? If you need swap to run faster, you need to do less swapping, not faster swapping. In this case, more RAM can help. >> >> Reason all modern H.D.D's use a concept called ZCAV(zonal constant >> angular velocity). this takes advantage of the fact that more linear >> space is available on the outer tracks of the disk platter rather than >> on the inside tracks. now since the disk spins at a constant rate , >> which is also known as CAV (constant angular velocity) the read/write >> I/O speed will be greater at the outer tracks as compared to the inner >> tracks > > It is unlikely you will notice any improvement. 5400 is 90/second. Not much > time no matter how you look at it. Do people put 5400 rpm hard drives in new machines these days? I have a 7200 rpm hard drive in my 10 year old computer, but it also has two 10,000 rpm SCSI hard drives in it. On my "new" 7 year old computer, I have six 10,000 rpm SCSI hard drives. 15,000 rpm hard drives have been available for years. Now the rotational latency of the higher speed drives is less than the slower ones. It so happens that the seek times are less too. And the faster drives tend to have much larger caches in them too. > > But unless you are a sysadm with lots of users calling their files common > files many times an hour, think of an office of insurance adjusters, your > largest files are in /usr. As folks are always interested in bragging rights > on availability (and even a newbie can get 0.999 availability) the boot > directory is going to be first and /usr second on a normal install. > > If you are having response time problems, increase RAM. I think you should always measure your machine and see where the bottlenecks are before you change anything. If seek time on your hard drives is a problem, raising the RAM size will do you no good. If your processes are compute-limited, likewise increasing RAM size will do not good. If your machine is bogged down by swapping, then increasing the RAM may help. I had an old machine once that was evenly balanced, but as time went on, it got too slow. It was swapping too much so I doubled the RAM size and that ended the swapping problem. But I doubled it again (I had two RAM modules), and it made little difference. By then it was clear it was the processor that was too samll. But the motherboard would not take a faster processor, so I gave that machine away and it was used for parts. I doubled the RAM size in my present (2004) machine to 8GBytes. 4GBytes were enough, and doubling the RAM size made little difference. I did it because I figured when I actually needed more RAM, the kind I needed would no longer be available. Also, in Linux (that I run), more RAM can always be used for cache so when I need a program (e.g., Firefox), it may load extremely fast because it is in the cache (RAM) and need not be loaded from disk at all. > > FWIW, this was a sort of geeky think to do on Windows in the 90s. I did it. > Never saw a difference. > -- .~. Jean-David Beyer Registered Linux User 85642. /V\ PGP-Key: 9A2FC99A Registered Machine 241939. /( )\ Shrewsbury, New Jersey http://counter.li.org ^^-^^ 08:15:02 up 10 days, 17:01, 3 users, load average: 4.58, 4.67, 4.78
From: Matt Giwer on 19 Jul 2010 03:01
On 07/18/2010 08:30 AM, Jean-David Beyer wrote: > Matt Giwer wrote: .... >> It is unlikely you will notice any improvement. 5400 is 90/second. Not much >> time no matter how you look at it. > > Do people put 5400 rpm hard drives in new machines these days? It is called worst case. -- Vox Populi Vox Sunac -- The Iron Webmaster, 4284 http://www.giwersworld.org/israel/is-seg.phtml a14 Mon Jul 19 03:00:58 EDT 2010 |