:lol:
The operating system…and office…is written in small chunks across all 24 drives…if you pull one drive you will have lost that section of data…
Other variables come into play like stripe size/width…but my previous comment stands true with a generic example of striping.
yep, now I remember. No data redundancy or parity bits used here.
hey now…
what you are mis-understanding is 2 things.
1- that the performance begins to decrease as you start to decrease the number of drives. With an array of 24 drives, you are always reading/writing to 24 drives in parrallel.
Example, we have 48 bytes of data being written. In this case, you are going to simultaneously write 2bytes of the data to each of the 24 drives, spreading the data across them. So, it is very fast since since all the data gets written at the same time. Assume that each drive can write 1byte per second, it would take 2 seconds to write all 48bytes to disk.
With one drive, it wouldn’t all get written simultaneously. Assuming the same speed, it would take a full 48 seconds to write the data.
2 - Which leads to the second point. You are spreading your data across 24 drives. Lets say for example, you put a 240byte file into your filesystem. That file would be spread across the 24 drives, and you would have 10bytes of it on each drive. Now if you lose a drive, you just lost 10bytes of the file, and it is un-usable.
Remember this is an overview, and the examples are simply there to illustrate a point, they are not real world scenarios.
RAID 0 - More lanes on a highway allow more trucks (traveling at the same speed) to get to where they are going.
RAID 1 - Single file… but if one truck crashes, all of the other trucks know, and can pick up the pieces.
Can you please use cars/trucks to explain raid 5 and 10…thanks 
I would love to see a SSD benchmark against the Velociraptor. Not saying it’d be close one way or the other, just a comparison.
raid 5 has multiple trucks on multiple highways, but there are repair stations at every exit.
omg, I am awesome
RAID 5 - 2 lane highway. 2 trucks can go fast at the same time with a 3rd truck behind them. 3rd truck is empty in case either 1 of the first trucks lose their data.(In real life that 3rd truck driver has half of the map… which would allow him to get to the destination based on 1/2 ofthe data)
RAID 10 - 3 lane highway with 2 trucks in each lane. You have the benefit of striping (the #lanes) and also mirroring (2 trucks in each). Benefit here is that both trucks in every lane has a whole map. So they don’t have to piece together directions from 1/2 a map.
:mamoru:
Not even close. :spank:
Granted… 24 10k drives in RAID would be badass-fast.
For the quantity of drives, 2GB/s is VERY poor. That array has to be significantly slower than your average 8-12 SSD array since its about the same speed, and has to access double the drives to get the information.
They needed a better RAID card.
Theres several people who do over 1GB/s no problem, on 4 SSD’s. 24? Really?.. Ugh, they needed to know what theyre doing. They used QX9770’s with those horrible coolers. They would have been better off just using the stock Intel coolers.
Yea, I was thinking it of a ram kinda deal. I like learning…
Could do the same thing with 15k SAS drives…
Well, the seek times with SAS drives would be significantly higher. Those samsungs run .1 ms or so, 15K SAVVIO’s generally do 6 or so ms. Although it depends on the application, since sustained IOPS on SAS would be much better than the Samsung SSD’s due solely to the poor SATA interface on the SSD’s. The SATA controllers are pathetic for high IO operations.
SAS SSD’s are going to be nice…
NERDS with hardhats
still not faster than the human brain.
huh?
I didn’t know you could learn 2GB of books and retain it in a few sec :lol:
I didn’t know a computer could analyze a room, and perform collision detection on visible items providing rough estimates of how it would react in a timely manner.
I took “math in action” at a SUNY school and got a c-. Speak for yourself smarty pants.
I keed I keed
storing data is not learning data. read bacon’s post for a layman’s breakdown.
truth. no computer currently can process data faster than the human brain. One day it will. Just not today.