Sas drives or SSD ?

[Armageus]

Our arrays actually already have spares built in, so the active disks run in a raid 6 config, and as soon as one has signs of failure then one of the spares kicks in, and replacement drive is sent to the customer.

Our arrays are smaller, typical 6-10 drive RAID10 direct attached arrays. We still use hot-spares, so like you say as soon as a drive fails it's automatically replaced and the array rebuilt, but we also have a huge quantity of cold spares.

As servers are replaced the drives are reused (assuming they are still the same type etc), and as drives are replaced with larger capacity, the smaller drives get reused.
For the price of 1 new HP hard drive, typically we can source 4-5 used drives from either ebay or specific IT recyclers/brokers. We add a small amount of new drives into the "pool" every now and then, further diversifying the sources and ages of drives. So far this has worked fine for our use case, with no issues in the last 6 years since we implemented this, and arguably is a true example of the "Inexpensive" part of the RAID acronym that seems to be missing in a lot of deployments.

We've just started buying SSDs for some of our servers, but again it's mostly been used SSDs from reputable resellers who are able to supply wear level etc and most offer there own independent warranty (normally of a year - due to how confident they are of the parts they resell)

 

[Liquidfox]

Yeah pretty useless right when two ssds hit their drive limit at the same time and drop off.

If you hit that, you've spec'd the solution wrong to begin with. Any half-decent enterprise lights-out management solution also reports the SSD lifespan remaining giving you plenty of warning to replace the drives in advance, or purchase an additional drive and configure it as a hot-spare.

Most SSDs will have a 5+ year lifespan, by which point you'd be typically replacing the tin anyways.