When to replace storage

[ggadrian]

I have a computer acting as a file server and running a VM (web app only accessible from the LAN) which has the following drives:

-OCZ Vertex 3 60GB as system drive
-WD Green 500GB for storage and the VM
-WD Scorpio Blue 700GB (or 750) as local backup drive

All the drives have around 25K hours of use as reported by CristalDiskInfo and the SSD as got little less than 9TB worth of writes.

Everything is backed up local and using a cloud service, so losing a drive would be really inconvenient, but virtually nothing would get lost.

Should I begin to be worried and replace the drives or am I OK for another year?

Thank you.

 

[PliotronX]

There are three things certain in life: death, taxes, and spontaneity of storage implosion. There is no way to tell if they will be good in a year, but I would say if they do not have any pending reallocation of sectors, performance issues, make noises, or otherwise behave strangely, I'd say there is a good chance they will be fine for over another year.

 

[Red Squirrel]

Typically I run them till failure and replace as they fail and let the raid rebuild. Data and uptime not affected. If I see lot of smart errors on lot of the drives I sometimes force them out and replace before they do fail. The key is to keep them spinning 24/7 as it greatly reduces the chance of failure. Minute they are turned off and back on there's a bigger chance of failure as the spin up and initialization puts a bit more stress on it than when it's just spinning away. Of course you can still get a failure on a spinning drive, it's just my experience that usually the failures are on cold boot. My storage server has redundant PSU and 4 hours of battery backup. I can extend that if I shut down other servers. At the 3 hour mark if power is not back yet I start shutting down most other servers. Been a while since I've had a power outage more than an hour though.

I was looking at the smart data on some of my drives, for some odd reason the ones in my oldest array are only showing as like 10k hours. Do those numbers roll over after a while? I know as a fact the drives are at least 3 years old and have been spinning 24/7. I'd even say they're more like 5 years old. Looking at some of the newer drives and they show as about 14k which sounds right.

As drives get bigger and cheaper sometimes it does make sense to retire raid arrays so bigger ones can be built using the same number of slots though.

 

[Sabrewings]

I have about 5 years on four 500GB WD Blues in a RAID 5 and all are going strong. I also have about four years on a 120GB OCZ Vertex 3, still going strong. I would say you should be good, but it's hard to say. Run them till they die. Just make your backups.

 

[COPOHawk]

I have about 5 years on four 500GB WD Blues in a RAID 5 and all are going strong. I also have about four years on a 120GB OCZ Vertex 3, still going strong. I would say you should be good, but it's hard to say. Run them till they die. Just make your backups.

I would go get your lottery ticket then...

There is no definitive answer...my experience shows that it is a best practice to look at swapping out mechanical drives around 4-5 year mark...depending on usage. Either way, take the time to have proper backups (Acronis TI) or equivalent.

I agree with PliotronX...death of your hard drives will happen...just depends on when.

 

[dave_the_nerd]

25k power on hours is just shy of three years.

At work, we have SANs with around 400 HDDs between them - failure rates really went up at around the four-year mark.

So I'd say you're probably okay for now, OP. On the other hand, a 1TB HD is <$50, so if your VM and stuff is important, it might be worth it to grab a replacement HDD and keep it on a shelf just in case. (If it means saving some time when one of the other drives finally does go.)

How many people access the VM at once? A single WD green isn't setting any performance records, that's for sure. I'd be tempted to replace that with an SSD if it's in the budget and can be justified based on heavy load.

 

[ggadrian]

The computer serves files for about 5 to 6 people and the VM has around 1 or 2 users, so the performance is not a problem.

Having spare parts IMO is not worth it, as I can have a replacement the next day if I buy at Amazon and anyway I have my job so I probably can't do anything about it until the next day.

The backup plan is very good, and I have tested it many times since every then and again the delete a file and I have to recover it.

In any case, as my idea is to avoid downtime, and it seems to be consensus that 4 to 5 years of use is OK, I'll replace the OS and data drive during 2016.

 

[Charlie98]

I'd probably DX that Vertex SSD, those were not known for reliable service. A 120GB Intel or Crucial would be my choice.

As cheap as drives are, maybe replace one of the HDDs this year, and continue on with the other (or just add a new drive and leave the other as a redundant backup drive.)

 

[ggadrian]

I'd probably DX that Vertex SSD, those were not known for reliable service. A 120GB Intel or Crucial would be my choice.

As cheap as drives are, maybe replace one of the HDDs this year, and continue on with the other (or just add a new drive and leave the other as a redundant backup drive.)

They are buying a new laptop in September, so I'll probably use the Vertex for the laptop and I'll replace it with a 250GB MX200 for the OS drive and I'll move the current VM and a new one they need to the SSD. Given how cheap hey are I'll try to get them to buy 2 and RAID them.

 

[mikeymikec]

IMO if it ain't broke don't fix it. For all you know, the replacement drives might last a fraction of the time these have lasted.

 

[Charlie98]

They are buying a new laptop in September, so I'll probably use the Vertex for the laptop and I'll replace it with a 250GB MX200 for the OS drive and I'll move the current VM and a new one they need to the SSD. Given how cheap hey are I'll try to get them to buy 2 and RAID them.

That's what I did with my 60GB Agility3... it's in my daughter's hardly used laptop (where it does pretty well, I might add.) Having said that, I expect it to die at any given moment, but in the laptop that's a simple reload of the OS.

 

[ggadrian]

That's what I did with my 60GB Agility3... it's in my daughter's hardly used laptop (where it does pretty well, I might add.) Having said that, I expect it to die at any given moment, but in the laptop that's a simple reload of the OS.

Sure, if it dies in a laptop I just load the OS again and it's good to go, as they have everything in the file server.