A computer components & hardware forum. HardwareBanter

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

Go Back   Home » HardwareBanter forum » General Hardware & Peripherals » Storage (alternative)
Site Map Home Register Authors List Search Today's Posts Mark Forums Read Web Partners

Poor raid 1 performance?



 
 
Thread Tools Display Modes
  #21  
Old December 1st 05, 10:28 PM posted to comp.sys.ibm.pc.hardware.storage
external usenet poster
 
Posts: n/a
Default Poor raid 1 performance?

Peter wrote:
Yes, but also _any_ read is done twice, on both channels and both


mechanisms

(and furthermore they are then checked for equality between, which adds a
step, which could be a reason for misperformance).

I do not see why you could see improvements, any other things being equal


of

course.



http://www.9to5computer.com/atto/Com...igurations.htm
http://www.storagereview.com/guide20...gleLevel1.html

With a good controller, sequential read performance equals that of a single
drive, random read performance scales up with number of drives in a RAID.



Thanks. These were the sorts of descriptions that led me to hope that
raid 1 would lead to improved read performance. Eg, one site says that
"Read performance is faster than a single disk; (if the array controller
is capable of performing simultaneous reads from both devices of a
mirrored pair)."

This is why I was then suspecting that the motherboard implementations
are poor. On the other hand, it might be just that the test was more
reflective of sequential than random read performance.

Can anyone suggest a raid 1 array that has improved performance?
  #22  
Old December 1st 05, 10:31 PM posted to comp.sys.ibm.pc.hardware.storage
external usenet poster
 
Posts: n/a
Default Poor raid 1 performance?

Antoine Leca wrote:
In , Bob Willard wrote:

since RAID1 has twice as many on-disk read channels
and twice as many seek mechanisms.



Yes, but also _any_ read is done twice, on both channels and both mechanisms
(and furthermore they are then checked for equality between, which adds a
step, which could be a reason for misperformance).

I do not see why you could see improvements, any other things being equal of
course.

What am I missing here?


Antoine


As another poster pointed out, it is suggested that raid 1 will improve
performance. It seems to very much depend on how raid 1 is implemented.
For maximum data integrity, you can read the same sector from both
drives and ensure that they do agree.

However, this really isn't a requirement for me. A single drive does
usually return the correct result. I am much more concerned with having
a duplicate drive in case one of the drives crashes. Assuming the drives
are okay (which is the usual case and should be detectable), just
reading from one of the drives is sufficient. Hence, why can't a
mirrored pair be accessed like a striped pair (ie, raid 0) which does
result in increased performance?
  #23  
Old December 1st 05, 10:35 PM posted to comp.sys.ibm.pc.hardware.storage
external usenet poster
 
Posts: n/a
Default Poor raid 1 performance?

Gerhard Fiedler wrote
Rod Speed wrote


Really, IMO there's little reason to run any computer that's used
to store critical data without RAID1 (or one of its derivatives).


Corse there is if the data activity isnt high and
normal backup will be completely adequate.


That's why I wrote "critical data".


Critical data aint necessarily volatile data.

I consider the result of my work "critical", as it is what
I get paid for. If it takes me half a day to redo the work
of a day, and I do daily backups (both probably a common
situation for everybody who works with a computer), a
RAID1 array is a quite nifty thing, as it should give me
approximately 0 downtime in case of a disk problem.


The obvious approach is a higher frequency of backup, like hourly.

Otherwise, I could easily have a day downtime (need to get drive,
restore last image, redo all the work that happened afterwards).


That just means you need more than one drive, not that you need RAID.

And if your system gets infected, you may have a
problem that RAID wont do a damned thing about.

The only computer-caused downtime I had so far was
either a harddisk crash or a Windows reinstall -- and
RAID1 is supposed to crack down on the first one.
(I haven't had a crash since I installed my arrays...)


And if the failure is in the raid hardware, not the drives,
you may well have much more of a problem than you
would have with duplicated PCs and no RAID at all.

Normal backup has a number of advantages over any RAID,


RAID1 is no substitution for backup, of course, it's
an enhancement for an adequate backup strategy.


Not necessarily a very useful enhancement
tho over just duplicating the PC, even if the
spare is the previously replaced system.

Thats much more likely to see you
up much more quickly on any failure.

Yes, you can certainly have both, but most
dont actually need the advantages of RAID.


Most don't, but then most don't create critical data. I'd
recommend it for everybody who creates critical data.


I wouldnt, duplicating much more than just the
hard drive makes a hell of a lot more sense.


  #24  
Old December 1st 05, 10:38 PM posted to comp.sys.ibm.pc.hardware.storage
external usenet poster
 
Posts: n/a
Default Poor raid 1 performance?

Mark wrote
Rod Speed wrote
Peter wrote


Why do you actually need improved read performance ?


He wants his computer to boot up faster


The obvious way to make it boot much faster again is
to minimise the number of boots and to hibernate instead
of shutdown when you do need to shut the system down.


My computer hasn't always recovered from hibernation okay, something
to do with the power management. Hence, I am hesitant to use it.


Makes more sense to fix that than to try to find
a RAID that improves the read performance.

Also, I do like the idea of rebooting regularly to stop the system
becoming bloated with memory resident programs that I don't need.


No need with a decent modern OS like XP.

and load games faster.


The obvious way to fix that problem is to keep
them loaded so they are an instant switch away.


Okay, you obviously don't play games. First, having them stay in the
background quite often reduces the computer to a crawl.


Bull**** with a properly configured system.

Second, even if you are in the game, level loads (and saving and loading your
position) can be very slow.


RAID aint gunna fix that.

To some people an extra 10-20 seconds is eternity.


Improved read performance isnt the only way to fix that. And
its perfectly possible to do much better than any RAID config
can do on that. Not a shred of rocket science required at all.


I'd be interested in other techniques. What do you suggest?


More physical ram.


  #25  
Old December 1st 05, 11:26 PM posted to comp.sys.ibm.pc.hardware.storage
external usenet poster
 
Posts: n/a
Default Poor raid 1 performance?

Rod Speed wrote:

I wouldnt, duplicating much more than just the hard drive makes a hell of
a lot more sense.


You definitely have a point here. But the price point is also quite
different.

Gerhard
  #26  
Old December 1st 05, 11:29 PM posted to comp.sys.ibm.pc.hardware.storage
external usenet poster
 
Posts: n/a
Default Poor raid 1 performance?

Antoine Leca wrote:

In , Bob Willard wrote:


since RAID1 has twice as many on-disk read channels
and twice as many seek mechanisms.



Yes, but also _any_ read is done twice, on both channels and both mechanisms
(and furthermore they are then checked for equality between, which adds a
step, which could be a reason for misperformance).

I do not see why you could see improvements, any other things being equal of
course.

What am I missing here?


Antoine



It is a designer's choice whether a read is or is not done twice and
compared;
RAID1 does not require that. One alternative is to issue the read to the
least busy HD or to that HD which will have the minimum seek to complete
the read.

Reading from both HDs and then comparing probably adds little data integrity
when compared with simply reading from one HD and counting on the ECC check
to catch a bad read or a bad sector.

--
Cheers, Bob
  #27  
Old December 1st 05, 11:40 PM posted to comp.sys.ibm.pc.hardware.storage
external usenet poster
 
Posts: n/a
Default Poor raid 1 performance?

More physical ram.

Looots of RAM. 2, 4 or more GB.
Install a good ramdisk software.
http://www.superspeed.com/desktop/ramdisk.php
Preload ramdisk with your most often used apps and data.
Run them from ramdisk.


  #28  
Old December 1st 05, 11:50 PM posted to comp.sys.ibm.pc.hardware.storage
external usenet poster
 
Posts: n/a
Default Poor raid 1 performance?

Gerhard Fiedler wrote
Rod Speed wrote


I wouldnt, duplicating much more than just the
hard drive makes a hell of a lot more sense.


You definitely have a point here. But
the price point is also quite different.


Not if you use the previous main system as the backup.

And many of us have more than one system for various reasons as well.


  #29  
Old December 2nd 05, 12:14 AM posted to comp.sys.ibm.pc.hardware.storage
external usenet poster
 
Posts: n/a
Default Poor raid 1 performance?

Rod Speed wrote:
Mark wrote

Rod Speed wrote

Peter wrote



Why do you actually need improved read performance ?



He wants his computer to boot up faster



The obvious way to make it boot much faster again is
to minimise the number of boots and to hibernate instead
of shutdown when you do need to shut the system down.



My computer hasn't always recovered from hibernation okay, something
to do with the power management. Hence, I am hesitant to use it.



Makes more sense to fix that than to try to find
a RAID that improves the read performance.


Also, I do like the idea of rebooting regularly to stop the system
becoming bloated with memory resident programs that I don't need.



No need with a decent modern OS like XP.


and load games faster.



The obvious way to fix that problem is to keep
them loaded so they are an instant switch away.



Okay, you obviously don't play games. First, having them stay in the
background quite often reduces the computer to a crawl.



Bull**** with a properly configured system.



Mate, you have no idea. The minimum requirements of some current games
are 512 megabytes of ram, and the preferred configuration is 1 gig! How
many of these do you think you can have loaded at once before the system
is reduced to a crawl? In fact, due to games directly accessing
hardware, there are issues with running more than one game
simultaneously. Hence, you need to quit one game before loading up
another. Alt-tabbing also sometimes doesn't work well due to problems
with display drivers. Again, if you played games, you would know how
tempermental the display drivers can be.


Second, even if you are in the game, level loads (and saving and loading your
position) can be very slow.



RAID aint gunna fix that.


Why not? Raid can increase read performance, so why would this not lead
to faster load times?


To some people an extra 10-20 seconds is eternity.



Improved read performance isnt the only way to fix that. And
its perfectly possible to do much better than any RAID config
can do on that. Not a shred of rocket science required at all.



I'd be interested in other techniques. What do you suggest?



More physical ram.



I am planning on having 2 gigs of ram, which is the most I can afford.
However, how much ram do you have in your system? My system is currently
using about 200 gigs of storage. It is not possible to fill a
motherboard with 200 gigs of ram, so obviously the HD still needs to be
used to load information.
  #30  
Old December 2nd 05, 12:39 AM posted to comp.sys.ibm.pc.hardware.storage
external usenet poster
 
Posts: n/a
Default Poor raid 1 performance?

Mark wrote
Rod Speed wrote
Mark wrote
Rod Speed wrote
Peter wrote


Why do you actually need improved read performance ?


He wants his computer to boot up faster


The obvious way to make it boot much faster again is
to minimise the number of boots and to hibernate instead
of shutdown when you do need to shut the system down.


My computer hasn't always recovered from hibernation okay, something
to do with the power management. Hence, I am hesitant to use it.


Makes more sense to fix that than to try to find
a RAID that improves the read performance.


Also, I do like the idea of rebooting regularly to stop the system
becoming bloated with memory resident programs that I don't need.


No need with a decent modern OS like XP.


and load games faster.


The obvious way to fix that problem is to keep
them loaded so they are an instant switch away.


Okay, you obviously don't play games. First, having them stay in the
background quite often reduces the computer to a crawl.


Bull**** with a properly configured system.


Mate, you have no idea.


We'll see...

The minimum requirements of some current games are 512 megabytes of ram, and
the preferred configuration is 1 gig!
How many of these do you think you can have loaded at once before the system
is reduced to a crawl?


YOU have no idea. It aint the ram use that slows systems
to a crawl, and anyone with a clue can have 4G of ram anyway.

In fact, due to games directly accessing hardware, there are issues with
running more than one game simultaneously.


Then keep those in a ram drive, stupid.

Hence, you need to quit one game before loading up another. Alt-tabbing also
sometimes doesn't work well due to problems
with display drivers. Again, if you played games, you would know how
tempermental the display drivers can be.


You're wildly exaggerating what problems there
are, and they're fixable anyway, most obviously
by loading the games from a ram drive.

Second, even if you are in the game, level loads (and saving and loading
your position) can be very slow.


RAID aint gunna fix that.


Why not?


Essentially because no RAID can make enough
difference to the drive read performance if you
have decent high performance drives in the first place.

Raid can increase read performance,


Not by enough to do anything about VERY SLOW.

so why would this not lead to faster load times?


Basically if it is actually VERY SLOW, no RAID will be
able to make enough of a difference to the read speed.

A ram drive will with very badly written games.

To some people an extra 10-20 seconds is eternity.


Improved read performance isnt the only way to fix that. And
its perfectly possible to do much better than any RAID config
can do on that. Not a shred of rocket science required at all.


I'd be interested in other techniques. What do you suggest?


More physical ram.


I am planning on having 2 gigs of ram, which is the most I can afford.


Is that with or without RAID and the extra drive ?

However, how much ram do you have in your system?


Irrelevant, I use my systems quite differently.

My system is currently using about 200 gigs of storage. It is not possible to
fill a motherboard with 200 gigs of ram, so obviously the HD still needs to be
used to load information.


Sure, but if you only do that rarely, the speed of doing that doesnt matter.

And you dont need every bit of what is on the hard drive in a ram drive.


 




Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Poll (please): Time-shifting Performance Bryan Hoover Ati Videocards 1 December 15th 04 11:56 PM
Question about performance The Berzerker Ati Videocards 1 September 27th 04 09:25 PM
G400 & G-series RR performance question. Kevin Lawton Matrox Videocards 6 May 20th 04 09:51 PM
Maximum System Bus Speed David Maynard Overclocking 41 April 14th 04 10:47 PM
Geforce 4 2D/desktop performance in WinXP zmike6 Nvidia Videocards 2 August 29th 03 07:41 AM


All times are GMT +1. The time now is 04:19 PM.


Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2004-2024 HardwareBanter.
The comments are property of their posters.