Posts by zootle 17 publicly visible posts • joined 10 Jun 2014
I agree that prevention is the best defence. However regular snapshots, replicated to another host (preferably off-site) are, to my mind at least, essential for mission-critical data. Add offline backups into the mix and you can control your recovery point.
An attacker would have to understand how the system is configured and penetrate at least two systems before any data can be corrupted.
Batteries are but one part of the EV equation.
Existing power grids will require extensive upgrades (especially for those fast charge times) requiring even more resources. More copper than we currently mine just for starters. If renewable generation is required (which it will be) there will also be massive demand for the rare Earths used in wind turbines.
Back in the 90s, I worked on a Nortel (remember them?) ATM switch that had become extremely popular as a replacement for large PABXs. A telephone exchange you could stick in a closet with no special power and cooling requirements! They sold by the shipload, replacing floor-sized exchanges with a small box on each floor.
Then along came IP telephony and the rest is, as they say, history.
You appear to have little or no experience with ZFS storage and more than a little difficulty with reading comprehension. What part of "I've built a number of systems with RAID controllers that can't do JBOD (the Dell PERC 710 in particular). The virtual devices just appear to the OS as drives." didn't you see? All ZFS features work on RAID volumes.
At no point did I claim ZFS has no management overhead, only that ZFS+hardware RAID adds an extra unnecessary layer. I see you cite old problems with a core duo 2 system that probably cost less than a RAID capable HBA and way less than one with a decent cache. How much memory do your hardware RAID systems use for deduplication?
If you want ZFS in production, use Solaris, Illumos or BSD. All of the commercial storage vendors use one of these.
Oracle fits hardware RAID cards so they can sell to users of operating systems that need them (Windows!).
All of my vital write data is safe and sound in the power protected log devices.... and yes, the power protection cam free with the Intel SSDs.
I'm not saying ZFS can't work with hardware RAID, it is quite happy to work with it. I've built a number of systems with RAID controllers that can't do JBOD (the Dell PERC 710 in particular). The virtual devices just appear to the OS as drives. What I am saying is it an inconvenience for the system builder (extra $, time wasted stetting up virtual volumes) and administrator (one more tool to use to manage disks). It can be more of an inconvenience when the RAID hides errors from ZFS which prevents the OS from recognising a failing drive. This has happened to me twice with Dell systems. The resource load on the system CPU is negligible. As for RAM, I'm sure you pay a lot less for system memory than your RAID vendor charges for cache memory.
I suggest JBOD because it is a cleaner, lower cost solution. It lets ZFS (and the host OS) directly manage the drives without an unnecessary intermediate layer getting in the way.
ZFS is tried and tested. It has been used for all of Sun, now Oracle's storage products for many years. The battery backed cache saves nothing. All my production systems have power protected log devices, so I can pull the system power until I get RSI before loosing data. I have tried and I got very bored.... The drive write cache is only a problem when the drive lies about honouring sync writes.
"because ZFS can't work with hardware RAID" is a myth. ZFS doesn't need it and works best with JBOD, but it works perfectly well with hardware RAID. If ZFS was reporting corruption, you had a hardware problem. Don't shoot the messenger!
All hardware RAID gives you is another potential source of firmware bugs in the drive path...
It never ceases to amaze me how much fuss is made in the tech press each time Linux adopts another piece Solaris technology. Those of us who manage Solaris based infrastructure have been using single service zones for nigh on a decade....
Hybrid storage has been the norm in Solaris and other platforms that support ZFS for nearly a decade now. All of the the ZFS storage I've built over the past six or seven years has the flash acceleration, snapshot and replication, write reduction and cache features people appear to be getting excited about now. I guess Sun/Oracle didn't shout loud enough or long enough.
https://blogs.oracle.com/ahl/entry/flash_hybrid_pools_and_future
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
