New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
Why would they both have to die at the same time?
Also unless you are thrashing the SSD's you will more than likely not write enough to them to kill the flash chips that fast I would have thought?
I'm not sure how the fact that SSDs could die at the same time changes anything? This is true for normal disks also. It is preferable to provision servers with mixed age disks so you reduce the likelihood of multiple drive failures.
I would host your MySQL databases on an SSD VPS or if the database is small enough, purchase a high ram VPS like those from VPSDime and serve the database from RAM.
@wych maybe not in 5 Months but some years and Raid 1 helps so keep everything in sync, so they are booth used in the same dimensions and Or not?
HDD's dont have chips they will not die after the amount of writen data, So they wont fail in the same time i guess.
We are using 6 ssd drives in raid 10 configuration on our byteshack plans so your db can be fast and safe in same time.
Use mixed drives and the problem is solved. However I do agree that it is more likely for 2 brand new SSDs in RAID1 to fail at the same time, than 2 brand new HDDs.
I wonder what are the odds of 6 SSD's failing at once
The question is what are the chances of two drives failing at once (assuming a RAID 10 array).
I was paraphrasing @drserver who implied raid 10 was safer. With SSD I guess concurrent failure is a very real and likely event, where the drives are equally read/written to.
Still the same applies. You only need two drives to fail at the same time in that array.
Mediatemple does that(stores mysql on SSD). Not sure why
Faster queries
if you rent a dedicated i doubt it would have same brand and same age
if you colo just use a different brand ssd, but of course it cant guarantee it wont failure in same time ( that's why backup is important)
Thats why you buy 2 SSDs from 2 series (or if larger at least X+1) of the same manufacturer (i.e. Corsair GT and GX) and replace one SSD each year.
Indeed. If you have control over the choice of drives, ensure they are different. This is especially true with RAID arrays as a rebuild can put additional stress on already weak components leading to failures.
You should have a spare on-hand anyways. Rotate your RAID disks.
It is not just about multiple drive failure. A RAID controller can fail and corrupt the data on the array int a way that you would not even imagine. A software bug can do the same. It is the responsibility of the host to monitor all the hardware to make sure the data is safe. It is unlikely that 2 drives will fail at the exact same time. What we do with our ssds is simple, if we are deploying a node with a raid 10 array of 4 drives we will overwrite one of the drives 150 times the second drive about 100 times the third drive about 50 times and the forth drive about 10 times just to be sure they will not fail at the same exact time due to regular ssd wear. It is also important to have backups in any half say serious application. Both the host (If advertised) and the end user should have backups of all the data they don't want lost. How often the backups should take place is up to the end user.
I will be quick.
Raid is not backup
Mysql replication is not backup
Always test your backups
If your data is as important as it sounds follow thouse rules each layer has more fult tolerance and more availability but with out all of them random acts of god will own you.
Get an Sad server from @drserver or someone the chances that two drives if a 6 drive araay both on the same side of the array failing r slim.
Get 2 servers from From @serverian one from VPSDime one from backupsy. Make sure the nodes are all in different locations if you can. If not make sure the VPSDime and Backupsy r in different locations. Setup backups from the VPSDime server to go to the Backupsy install mysql on Backupsy and have scripts auto restore backups there. Also do something like row count test between all nodes to make sure they are close.
If your data still means more to you download backups from backup server md5 them and burn them to tap if not CD or DVD and mail them to your lawyer.
The average enterprise SSD which you'll find in most decent servers have an MTBF of around 3 million hours which is 342 years. Putting a heck of a lot of load on these SSD's might put them down to 1 million hours but that's still 144 years even then only 500,000 hours of running is 70+ years.
Any server hardware these days are not going to last 70+ years so not really realistic.
Use SSD's and keep backups.
I need SSD VPS's in germany which no one can offer here but thanks @wojons.
I will dump all databases every hour to an external server, same with the files should be enought.
I have worked at compines that used 24 ssds in a database server really nice hardware and we have had ssd drives fail the way they get these MTBF rates is not really what mostst people think. They have not been testing the drive for 342 years so they get lets say 1000 drives going at once or something crazy like that and they wait til they get ethe first failure they count the total number of hours between the dries and there you go
http://nomtbf.com/2012/11/sample-size-and-mtbf/
I will dump all databases every hour to an external server, same with the files should be enought.
DOes it have to be germany of some where close?
Should be Germany.
https://www.vultr.com/locations/
They seem to have a frankfurt location not sure about stock
30GB SSD and 8$ dam cheap but out of stock.
Might be worth waiting and getting it then rushing into other options.
It's fairly unlikely they will die at the exact same time, but if it's risk you want to avoid, burn one in for a bit.