What raid setup to choose?

[guitarguy6]

Hi guys/gals, I've been running Raid 0 with my 2 X 500GB Seagate sata-2 hdd's. I recently picked up 2 more of the exact same model and am wondering what the fastest Raid setup would be. I have a Gigabyte EX58-UD5 which supports Raid 0, 1, 5, 10. Should I stick with Raid 0 or is there something faster.

Thanks!

 

[jay2472000]

I would say that adding the other drives in wont help performance a whole lot, and may actually hurt it in some types of reads. I read somewhere that in small data reads, it may actually take a raid 0 array longer as it has to position more heads ( 4 in this case assuming you add the new 2 into the array ). I cant for the life of me find the link, and take that with a grain of salt, but it seems to make sense.

If you really want to make a huge leap in performance start saving change till you have enough for a SSD. Thats what I did and I am almost there.

 

[drescherjm]

Hi guys/gals, I've been running Raid 0 with my 2 X 500GB Seagate sata-2 hdd's. I recently picked up 2 more of the exact same model and am wondering what the fastest Raid setup would be. I have a Gigabyte EX58-UD5 which supports Raid 0, 1, 5, 10. Should I stick with Raid 0 or is there something faster.

Thanks!

Raid0 is the fastest raid but is also the most risky.

 

[drescherjm]

I would say that adding the other drives in wont help performance a whole lot

Agreed. It may help benchmarks and STR but will have a limited effect in the real world where most reads and writes are small.

 

[pjkenned]

Fastest would be raid 0 (super low overhead no parity disk)... but do you really want to have 4x the chance of failure v. single drive? Yikes. As stated earlier benchmarks would be faster but 4x the chance of failure v. a single drive? Even with backups I wouldn't do it for anything other than a benchmark drive.

 

[terayon1515]

RAID0 running over Fake or SoftRAID is the fastest.

If you can handle losing some of the extra capacity these drives could afford - I would go Raid10. Keep the same solid speed you have now with added redundancy.

A RAID10 on an Intel Matrix powered board of mine consisting of four WD500 Blues had a failure today, the rig used to be a RAID0 .RAID10 kicks butt and takes numbers, RAID5 without a hardware engine seriously lacks any decent speed. (25-30mb/s writes, 48-60mb/s reads typically)

Worth the 50% loss in capacity. Especially with these cheap 500's...

 

[guitarguy6]

I think I'll try a Raid 0 with the 4 disks. Is there a significant advantage to short stroking them? The possible crash and data loss doesn't bother me since this pc is only used for gaming.

 

[jay2472000]

Best case scenario you could do with those 4 drives is raid 0 and shortstroke them. On paper........


Basically, think of it as building a car, do you cram 4 VW engines in ; or just drop in a hot bigblock and call it a day???


I hate to say it, but I dont think youre gonna be as impressed by the real world performance as the on paper performance.

 

[pjkenned]

Best case scenario you could do with those 4 drives is raid 0 and shortstroke them. On paper........


Basically, think of it as building a car, do you cram 4 VW engines in ; or just drop in a hot bigblock and call it a day???


I hate to say it, but I dont think youre gonna be as impressed by the real world performance as the on paper performance.

^This is a pretty good one. 4 points of failure vs 1. Plus even if you could go 200mph with the 4x VW engines and only 170mph with the bigblock, how often would it matter in a world where you get rapidly increasing tickets over 70mph and silver bracelets at 105mph?

Just to give you an idea, I had 8x 15k rpm 2.5" SAS drives in Raid 5 on a hardware controller with 512MB of battery backed write cache (read MUCH faster than you are going to see on a 4 drive raid 0 short stroked setup... but with redundancy). That was faster in most benchmarks than my SSD's. Real world, except for large sequendial reads/ writes, the SSD is faster. Then I moved my main system to 2x OCZ Vertex 120GB in raid 0 which was faster in benchmarks yet real-world usage there was no noticible difference. Same thing when I went to the Intel X25-V Raid 0 setup from a single (except sequential write speeds are now much higher which is noticible on Gigabit LAN).