Hard drives and Trainz performance - is faster better?
- Thread starter Paulsw2
- Start date
oh! I was hoping to have it for Christmas! I'm prepared to wait though.I can just repeat: from my point of view there is no difference in fail of simply connected single drive and drive connected in RAID0, only mathematical probability of fail should be 2x higher, but it is more than well balanced by performance of such RAID, which is very necessary for any still-almost-the-same-graphic-engine-Trainz fan playing it under 1600x1200 and higher resolution going to load some well done larger map with a lot of up to date assets used.
johnwhelan
Well-known member
So if we expect a hard drive crash every 3 years on average running RAID0 brings this down to 18 months and even then do you replace the pair of drives or just one? Then you need to replace the drive with one that exactly matches it etc. If the MTBF is three years and you've already done 18 months on one drive you can then expect a crash in what roughly 12 months? This looks like diminishing returns to me.
Cheerio John
johnwhelan
Well-known member
Probably based on personal experience as much as anything. I've had at least three drives replaced under a three year warranty, I have a two year old lap top that currently needs the hard drive replacing and all these are run on UPSes so its not the hydro causing the problems. Google run their drives 24 by 7 I think in Raid 5 until they die, so a disk dying isn't the same level of problem as it is on a home PC, the average PC gets turned off and on so has warm up considerations as well. The newer drives maybe more reliable but only time will tell.
SDD drives have no moving parts so should be much more reliable. Tomshardware.com has an article on gaming PC which said that for most games they load faster on SDD but the actual game experience isn't much different. This would tally with the perfmon stats we got from earlier versions of Trainz. Note also that Raid0 improves one type of performance loading big files throughput, but what it doesn't improve is loading small files where you need to get the head in position first hence the comment above that the raptor gave better performance than two 7k disks in Raid 0.
Cheerio John
my current 160GB HD has been going for 4.75 years and only has 1.3GB left on it. I hope it can hold out!Paul
johnwhelan is just too skeptic, I say there is almost no difference between fail of single drive and drive in RAID0 (it is not necessary to replace failed HDD by exactly the same, but it is best for sure) and performance of RAID0 especially using Velociraptors is worth regardless of small or big files read or written. Such configuration improves significantly run of entire OS and all running applications as well. I have many experiences with HDD fails and I've never lost any data yet (even on failed RAID0).
Thanks John for that extensive answer. It makes sense. Warmup and cooldown is always a problem for hardware. I've also read about Google running their drives fairly warm for longer life.
I must be lucky. No drive failiure so far. (touch wood) The oldest is five years now and nothing else than a damaged block that's gone after chkdsk plus had to replace one faulty .dll from the Windows cd.
I must be lucky. No drive failiure so far. (touch wood) The oldest is five years now and nothing else than a damaged block that's gone after chkdsk plus had to replace one faulty .dll from the Windows cd.
Just to clarify a point here, raid0 is not double the speed of an individual drive, this is due to the fact that the seek times are not altered as both drives must go through it, so if the read/write and seek times are equal the best you can expect is a 25% increase in overall speed on average files, trainz uses a lot of small files for an object so the percentage of seek time is going to be a lot more than 50%, decreasing the effects of raid0.
So raiding two 7,500rpm drives. will give you a 25% speed increase at best.
Now compare the above to a 10,000rpm drive, which will give a real time and consistent increase of 33.3% over 7,500rpm drives, so putting in a faster drive will give you a bigger increase in speed.
Raid was started to use up all the smaller hdd's that were being replaced by bigger and faster drives but still had most of their service life to go, I think the meaning of the acronym says it all
Redundant Array of Inexpensive Disks
Cheers David
So raiding two 7,500rpm drives. will give you a 25% speed increase at best.
Now compare the above to a 10,000rpm drive, which will give a real time and consistent increase of 33.3% over 7,500rpm drives, so putting in a faster drive will give you a bigger increase in speed.
Raid was started to use up all the smaller hdd's that were being replaced by bigger and faster drives but still had most of their service life to go, I think the meaning of the acronym says it all
Redundant Array of Inexpensive Disks
Cheers David
:hehe: very good!
johnwhelan
Well-known member
Quote:
Originally Posted by pommie
....Redundant Array of Inexpensive Disks
Cheers David
About 35 years ago before PCs, in the universities there were three problems with disk drives on mini computers, these were the ones say with 1k of memory upwards. First the big ones were practically hand built and cost large chunks of money, the smaller ones weren't big enough and when they failed you lost the data.
So they put together a redundant array of the small inexpensive disks so that if one disk failed the computations could continue and the data would not be lost. You could get the same amount of storage with these arrays than you could with the bigger drives for less money and more importantly you weren't limited to a fixed drive size. ie having to buy more disk space than was required. That's why RAID1 has redundancy, RAID0 actually came later. The computer manufacturers then picked up the idea so RAID 5 etc followed.
Cheerio John
Originally Posted by pommie
....Redundant Array of Inexpensive Disks
Cheers David
About 35 years ago before PCs, in the universities there were three problems with disk drives on mini computers, these were the ones say with 1k of memory upwards. First the big ones were practically hand built and cost large chunks of money, the smaller ones weren't big enough and when they failed you lost the data.
So they put together a redundant array of the small inexpensive disks so that if one disk failed the computations could continue and the data would not be lost. You could get the same amount of storage with these arrays than you could with the bigger drives for less money and more importantly you weren't limited to a fixed drive size. ie having to buy more disk space than was required. That's why RAID1 has redundancy, RAID0 actually came later. The computer manufacturers then picked up the idea so RAID 5 etc followed.
Cheerio John