I have an odd use case I could do with some suggestions for. I use my own PC as a home server for my family. It’s a Lian Li PC-A71B with space for up to 15 3.5" drives and I am currently using a SAS3 HBA card (8 drive capacity) along with my motherboards 6 SATA (5 drives + DVD drive), I am running windows 10.
All these drives produce a lot of noise and waste a lot of power if left spun up all the time and they are fairly infrequently accessed (a couple of times a day).
They all spin down fine, but there are far too many things that wake the whole lot up without them actually needing to be accessed. The worst culprit is windows explorer, if I leave it open then it wakes all my drives up every 10 minutes or so. If I try to install something then getting the drive listing wakes everything up too.
I could do with some kind of caching software which caches all the folder structure/indexes as well as all the files that are under a certain size, but i’ve tried programs like primocache and they do not work this way as far as I can tell.
I want to avoid buying a new NAS since I already have this giant PC case that supports holding all these drives (a NAS of this size is expensive), the drives only need to be accessed during the day too so I want to avoid having something running all the time. I have considered converting the case into a kind of NAS, but I am not sure what kind of hardware would be a good option for providing 15 sata connections whilst not using much power.
It’s hard to say either way really, drives have an unknown lifespan with an average based on statistics rather than a concrete figure per drive.
My system uses 150W less power if I force spin all the drives down, given the electricity prices are going up a lot where I live it would cost about £45 a month to keep them spinning all the time.
The thermal stress from heating up and cooling down devices also does some degree of damage, which can be important for precision devices like hard drives.
Google’s study found: “for drives 3 years and older, higher power cycle counts can increase the absolute failure rate by over 2%.”
I am not disagreeing with this, but I don’t run a data centre and the statistics mentioned don’t really scale down, I don’t have 10,000 drives whereby I might lose an extra 10 or so drives. My drives might last 119 months instead of 120 months? With the current power costs I could easily afford to replace any failed drives from the power/money saved.