Raid 5, Yay or Nay..
- Thread starter Si.
- Start date
Don
because we destroy the old disks rather than send them back to HP, so they need ordering rather than a quick call to HP
very irritating.
Eh? Even if it's sensitive data, surely you just make use of HP Defective Media Retention as part of your care pack, to have a replacement on site next day at least, no approving/purchasing required, without having to send the failed disk back?
And if you aren't covered by care packs (e.g. supporting older servers etc), then surely you just have a couple of "cold spare" drives on a shelf ready to cover a failure. If you regularly have to run any RAID level in a degraded state, then you are defeating the point of having it.
hp defective media retention:
http://www8.hp.com/h20195/v2/GetPDF.aspx/4AA6-4603EEP.pdf
Deleted member 138126
D
Deleted member 138126
Huh, I didn't know this existed. You learn something every day!
raid 5 + a backup > raid 10 (for a home user) half of "I need to restore scenarios" will be non disk failure related
Don
Depends - for us it's a none issue. We don't have vendor cover on any of our servers (HP DL360 G5 and G7s, DL380 G6 and a couple of DL380p G8), what we do have is server redundancy (i.e. every server has a "warm" duplicate that could be swapped to with a few minutes of effort), and spares of older kit is cheap and plentiful (e.g. via I.T. recycling/brokerage companies or even ebay) enough that we have a couple of complete servers on a shelf as "cold" spares, in addition to common spares such as hard drives, PSUs, fans, RAM etc.
When I took over I.T. for our site, it was more important to me to standardise our servers (e.g. before we had a mix of Dell, HP, self built etc), and have "lots" of inexpensive servers rather than refreshing a couple of really high spec new servers (and having all your eggs in one basket type position) every 3 years or so.
didn't know HP offered that service and no, we don't carry spare disks unfortunately - daft if you ask me. But that's why i spec three disk raid 1 now as that means the servers are carrying an additional 'spare' by default. Plan was originally raid 1 + hotspare but ADM looked superior.
crazy thing is if we carried spare disk's i would have been happy going with 2 disk approach to raid 1 with a couple of spares in the cupboard, saving the cost of a good few disks. The bean counters didn't like that idea though, so it ended up costing them more.
crazy thing is if we carried spare disk's i would have been happy going with 2 disk approach to raid 1 with a couple of spares in the cupboard, saving the cost of a good few disks. The bean counters didn't like that idea though, so it ended up costing them more.
I meant if you have bean counters who ensure that a purchase takes weeks and also aren't allowed to purchase standby hardware, then it's a bad idea to let things go out of support. Your approach seems reasonable.
I just use RAID internally on my NAS for uptime convenience as it makes it quick to get up and running again if one drive in the array fails than if a main (single) drive dies and has to be restored from backup. For actual data backup redundancy I've got some stuff replicated to cloud storage and the entire array replicated in realtime to an external USB drive plugged into the back of the NAS - I also regularly snapshot the whole array via the USB copy port and store that drive elsewhere. Fortunately I only have around a TB of data that I like to keep backed up though would be a bit trickier with multiple TBs.
Soldato
I use SHR2 so Hybrid Raid 6 on my NAS as I have a number of older drives that could fail in any order and having protection against two drives seemed reasaonble; plus I have 12 bays available so its not such a loss to run 2 drive redundancy.
All data is either hyper-backup'ed to Amazon Drive or synced if pictures/photos & non-contentious.
All data is either hyper-backup'ed to Amazon Drive or synced if pictures/photos & non-contentious.
Soldato
I personally have no problem running 4 disk RAID5 at home for a reasonable balance of availability and capacity. But I do have a completely separate backup so if there was a problem with a rebuild then I can go to that (and if there is another issue where I lose files which is much more likely).
In a work environment I wouldn't be using it for anything important (i.e. anything whose SLA meant that recovering from backup is an issue if there was a problem) but I have come across things which I cannot comment on which made me cringe.
In a work environment I wouldn't be using it for anything important (i.e. anything whose SLA meant that recovering from backup is an issue if there was a problem) but I have come across things which I cannot comment on which made me cringe.
yeah lots of approaches are to people in I.T. but it's the bean counters who need to be reasoned with and many times they don't understand I.T at all - then something goes wrong, they get the ultimate blame, then they are OK for a few months then back to normal.
