RAID6: Samsung or WD?

??

Having run this array for ten months on a 600VA UPS system, moving it ~3000 miles from CA to KY, and having it live through September thru now March in the power-blippy mess that is Louisville's power grid, I have only one response: this is NOT an issue.

Should I spend $150 raid card that can do RAID6 on commodity Western Digital BLUE drives that feature the dreaded TLER capability which normally trashes RAID arrays they live in, or should I spend $600 on a fancy caching Adaptec card + another $100 for the BBU unit, only to then potentially encounter the RAID issues with TLER firmware in the drives?

Since there's no hardware caching on this controller, and I spent the extra $90 on an uninterruptable power supply, howabout instead you stop assuming things that aren't true and instead simply ask if I had thought about it? I did think about it, and whatever problem you perceive there to be simply isn't.

 

I didn't know there was an OpenWRT for x86... Nice, I have something new to play with tonight

Also keep in mind that the "head end" of that box is being used as an HTPC for my wife to visit varying streaming websites or to peruse the myriad movies, TV shows, home videos, family pictures and music files we've stuffed into the WHS2011 guest machine. It gets used from pretty much every angle!

Also, the only Hyper-V usage of IOMMU (that I'm currently aware of) is the RemoteFX capabilities where it virtualizes your 3D video card. Unlike ESXi, Hyper-V allows you to have significantly more guests using the 3D resources. The downside is those resources are all artificially constrained to 256mb of VRAM each, so the remote VDI session is only of "limited" gaming use compared to what many of us would want.

 

I don't even understand why you're posting. You went from your boiler-room PC to why a UPS doesn't tell me why my system crashed.

What?

 

Are you trying to insinuate that, at some indeterminate point in the future, I might experience a disk failure? Oh my, you don't say?!?

</sarcasm>

Again, I'm not sure why you're having this conversation. If I somehow hard-lock the box and it somehow misses writing a stripe of data to the array, my world isn't going to end. It's not worth the time, money or effort to go find a gently used RAID controller, buy a new BBU (because how am I to trust the life expectancy of a used battery?), perform a full backup of all my data on the current array, open the box and rewire that entire setup (did you see the pictures?) to swap RAID controllers, rebuild the array on the new controller, and then restore all my data onto the new volume...

That's easily three days of work, all said and done.

You say I should perform all of this to avoid some trivial, remote possibility that a single 64kb stripe of unwritten data sends my entire 6TB RAID6 volume to dev/null? Even if it does, I make backups of that volume once a week to a 4GB USB drive. There's nothing I will lose in one week's time that will outweighs the cost of three days of my life.

I'll take my chances.