Sign in with
Sign up | Sign in
Your question

Benchmarking and maintaining your Raid array in Vista for rookies

Last response: in Storage
Share
February 18, 2009 7:54:13 AM

I have heard of many paths, utilities, etc in order to benchmark the read/write speed of a RAID array and have seen or heard no consensus from which it is to be conducted. I have personally noticed that of all the 3rd party benchmarking utilities and admin tools within Vista that summarize harddrive performance, the most accurate way I have found to benchmark your write speed on any given partition with a solid "number" other than sequential writes, is to monitor the copy speed after dragging and dropping files from one partition to another. This can be done by dropping the 'more details' link at the bottom of the 'transfer progress' box. There you will see the speed at which the new files are being copied to the new location.

Try doing this with a large complex file...(my largest game is 'The Witcher' like 15 gigs). Also, it is worth adding that the 'C' drive will have faster and more efficient (less fragments) performance than your 'D' , 'E', 'F', etc. Why is this? Is this because the 'C' drive is the center drive? Closer to the internal motor or whatever? Asking because my outer partitions need defragmented after every install of large game or media files and also have slower transfer speeds.

Speaking of Defrag... Smartdefrag is the "smartest" defrag tool I have found as freeware and use it after ANY fresh install or transfer of significant size. Unless one enjoys the hours of inactivity while awaiting a defragment to complete, which I would not. Not only does it defragment but observes activity of frequently used files and puts them ahead of the rarely used or locked files. Most importantly, low resource utilization; enabling one to multitask without screwing up the fragment process

Another aspect of significant dispute amongst RAID 0 users is the loss of data incase of disc failure. Is it not common practice to back-up data/files/system on a weekly/bi-weekly basis? If these simple rules of thumb are followed then what is the problem utilizing the greater performance/storage utilization of a RAID 0 array? For the average user a simple flash drive of 64 gigs would suffice, but if not there is the option of having a free disk available. There are also the many thousands of options of online storage; for a small price.

For those using Vista 64-bit, the most comprehensive and safe third-party utility for Benchmarking read/writes I have found is CrystalDiskMark 2.2. Simple and free; best overall, it is quick and easy to get rid of when there is no more need for it.

Another FAQ answered is that you CAN use different harddrives to configure your RAID 0 array; whether it be speed or size. Just keep in mind that the number of drives is multiplied by your smallest drive. So, of course it wouldn't make much sense to use a 80 gig with a 500 gig that in turn would give you 160 gigs +/-. However, there is no great loss of using a 3 or 400gig with a 500gig.

Another FAQ that I am at odds with is the use of RAID hardware controller compared to software if dealing with more than 2 drives. I am using three drives as RAID 0 and use the Nvidia striping and get a consistent 80 MBs reads/writes using the 'C' drive which is partitioned at 300 gigs. The reads/writes gradually drop the further away I get from the 'C' partition. Again, is this normal?
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
I AM NO PRO! I do not have any sort of academic knowledge on this subject; I only have the passion and enjoyment of custom building for many years, so please if I am mistaken or have left out anything, let me know.

Hope this helps any noobies looking to array their disks.


Q9550 OC @ 3.6 Ghz 1.2 Vcore w/ Tuniq Tower
1650 fsb linked/synched @ 1.1
EVGA 7901 FTW
OCZ3P2000 @ 1650
4Gb(2X2)
9-12-12-31-2T @ 1.65
BFG 8800GTS OC in SLI
1 x 750 gb Seagate; 1x WD 500 gb; 1x 500 GB Seagate
1.5 TB RAID 0 Array
Vista Home 64-bit
CoolerMaster HAF 932 Full tower
Air cooled with High Powered AC/DC fans
30-35C idle/ 40-55C Full load

February 18, 2009 4:49:22 PM

As far as hard drive maintenance goes, you may want to mention something along the lines of disabling the VSC that is built into Vista; or evaluating if it is really necessary.
February 19, 2009 12:25:48 AM

Quote:
I have heard of many paths, utilities, etc in order to benchmark the read/write speed of a RAID array and have seen or heard no consensus from which it is to be conducted. I have personally noticed that of all the 3rd party benchmarking utilities and admin tools within Vista that summarize harddrive performance, the most accurate way I have found to benchmark your write speed on any given partition with a solid "number" other than sequential writes, is to monitor the copy speed after dragging and dropping files from one partition to another. This can be done by dropping the 'more details' link at the bottom of the 'transfer progress' box. There you will see the speed at which the new files are being copied to the new location.


Synthetic benchmarks are only ever going to give you a ruff idea in any hardware, whether it be disk drives, GPUs, CPUs, etc. We get closer to true results when the benchmarking utilities get closer to what happens in real world usage. However, they cannot make benchmarking software to suit everyone, as everyone will use their machines in different ways different purposes, applications installed etc. Yes sure if you are doing alot of copying partiton to partition, disk to disk a stop watch on copies as suggested will give you a great idea on what will happen in further applications of that process. Changing to a database environment or gaming will give a completely different result.

Quote:
Try doing this with a large complex file...(my largest game is 'The Witcher' like 15 gigs). Also, it is worth adding that the 'C' drive will have faster and more efficient (less fragments) performance than your 'D' , 'E', 'F', etc. Why is this? Is this because the 'C' drive is the center drive? Closer to the internal motor or whatever? Asking because my outer partitions need defragmented after every install of large game or media files and also have slower transfer speeds.


It is not the inner sections of the disk that are faster, but the outer sections that are the outer sections are written to first. Meaning that your first partition on a disk will always be written to the outside, and any subsequent partition is written closer and closer to the center (or inner sections). It should have no bearing on fragmentation, this will be based on the amount of transfers to that partition or disk.

Quote:
Speaking of Defrag... Smartdefrag is the "smartest" defrag tool I have found as freeware and use it after ANY fresh install or transfer of significant size. Unless one enjoys the hours of inactivity while awaiting a defragment to complete, which I would not. Not only does it defragment but observes activity of frequently used files and puts them ahead of the rarely used or locked files. Most importantly, low resource utilization; enabling one to multitask without screwing up the fragment process


I use Ultimate defrag myself but no where need to the extent you have explained. Defragmenting constantly can put undue wear on any hard disk, and as we know any mechanical device will wear out. I personally will generally only do this when fragmentation starts to get bad ~15-25% any good defragmentation program will give you a good indication of this.

Quote:
Another aspect of significant dispute amongst RAID 0 users is the loss of data incase of disc failure. Is it not common practice to back-up data/files/system on a weekly/bi-weekly basis? If these simple rules of thumb are followed then what is the problem utilizing the greater performance/storage utilization of a RAID 0 array? For the average user a simple flash drive of 64 gigs would suffice, but if not there is the option of having a free disk available. There are also the many thousands of options of online storage; for a small price.


This is a fair point, however, this is generally only brought up to make the user well aware that if they dont currently backup then it would be a good time to start if you are using a RAID-0 array. Alot of people actually refer to RAID-0 as AID-0 as it doesnt deserve the R, in the same vain RAID isnt call BRAID because it IS NOT a substitute for backing up. Another great tool to use for this is imaging software, that can take a complete image of a drive and copy it to a different location. Giving you the near instant ability to boot the same OS from a different source if the initial has failed.

Quote:
Another FAQ answered is that you CAN use different harddrives to configure your RAID 0 array; whether it be speed or size. Just keep in mind that the number of drives is multiplied by your smallest drive. So, of course it wouldn't make much sense to use a 80 gig with a 500 gig that in turn would give you 160 gigs +/-. However, there is no great loss of using a 3 or 400gig with a 500gig.


Heterogeneous (disks of different model or manufacturer) can be made into arrays, not only will the smallest disk designate the size of the array as described, the slowest drive will also have a bearing of the performance of the array (rather obvious).

Quote:
Another FAQ that I am at odds with is the use of RAID hardware controller compared to software if dealing with more than 2 drives. I am using three drives as RAID 0 and use the Nvidia striping and get a consistent 80 MBs reads/writes using the 'C' drive which is partitioned at 300 gigs. The reads/writes gradually drop the further away I get from the 'C' partition. Again, is this normal?


A dedicated hardware controller should be able to increase read/write speeds on most any configuration as they aren't as limited as their onboard counter parts, your array of three disks seems to be hitting a limit of the onboard controller. If your "C" partition was created first this is consistent to the outer edge of the disk being faster, and subsequent partitions being slower. Hardware controllers can help to negate this with onboard cache and processing techniques.

Quote:
Hope this helps any noobies looking to array their disks.


Hope i have helped you out with any inconsistencies and wholes in the information.

File Server:
Q6600 - Arctic Freezer 7 Pro
ASUS P5K-e Deluxe
4gb Corsair RAM 667mhz (4x1gb)
ASUS 7600GT
8x500gb WD AAKS
1x320gb Seagte
Adaptec 3805 RAID controller RAID-5 at 3.18TB
Server 2008 Standard 64bit - 3 VM's running off Hyper-V technology, XP SP3, Windows 7 Beta and Server 2003 Standard 32bit
Lian Li A70B
!