Right on, thanks Quentin -- this better-defined "fakeraid" is much worse than soft-raid in many cases, I think, because drive recovery after say mobo failure could be extremely difficult unless you have an identical board (or close-enough chipset) available on reserve... with soft-raid, provided you backup & clearly document your config, drivers, certain kernel settings, etc, you should be able to move the array to another host without much headache or surprise liability :) Plus, it has been said for years now that for commodity hardware & consumer-grade arrays, modern CPU's are *plenty* fast enough to maintain throughput using simple soft-raid.
I have not seen the most recent benchmarks or fully examined the latest HA (high-avail) compromises, but SCSI is still much better that SATA (bye, PATA) for high-volume multithreaded uses needed for many email, DB, web, and other servers. Still, the underlying platters are not very different: vertical/perpendicular recording pushed Moore's wall again, so we've got way better volume and speed from today's single drives than we had with low- and even some mid-range arrays of yesteryear... so even when just hobbying about, a question is HOW are you going to usefully test the throughput of your array? Is it a feat in itself, or a practical solution? :) I digress. At the LUG level, I see the ability to save data as being a great promotion of open source. Proprietary RAID solutions have a built-in "easter egg of danger" since the redundancy is an illusion if their secret technology fails for any given unknown reason -- one is not likely able to make any clever assessments beyond the advertised capabilities, when the inevitable situational surprise arises. In short, a responsible customer gets "stuck". If this isn't making any sense to you, I'd like to simply propose the breaking and recovery of a raid array at a public function like the eugene celebration. I would be happy to bring down a sledgehammer and provide the single drive which we'll rip out and destroy (to great effect but in a contained manner). After that hooplah, a simple showing of the recovery: rebuilding in progress, etc, and some well-planted statements from the audience about whether this reliable and well-supported, free technology could help our government's accountability or something :) thanks y'all, ciao, ben On Nov 16, 2007 1:42 PM, Garl Grigsby <[EMAIL PROTECTED]> wrote: > Quentin Hartman wrote: > > On Nov 15, 2007 9:29 AM, Ben Barrett <[EMAIL PROTECTED] > > > <mailto:[EMAIL PROTECTED]>> wrote: > > > > https://help.ubuntu.com/community/FakeRaidHowto ("Fake" raid is > > soft raid, nothing really fake about it) > > I don't know why software RAID does not get more respect these > > days. When processors were just babies, > > hardware raid was needed for reasonable performance, but now soft > > raid is really fast, unless you're comparing > > it with enterprise-grade NAS setups. IMO soft raid is fine for > > non-SCSI RAIDs :) > > > > > > Just a clarification on this point. Fakeraid is (every time I've seen > > it, anyway..) used to refer to raid cards that claim to have RAID, but > > actually offload all the raid functionality to the host machine via a > > driver. Most of these cards only have a firmware program to manage the > > drives, nothing else is done in hardware. Hence the name Fakeraid, it > > looks like hardware raid, but it's not, it's fake. This is also > > sometimes called firmware raid. A very small percentage (I only know > > of one) of these cards also have a hardware XOR engine for offloading > > the work required for Raid 5. Ooohhh... Fakeraid+ :) > > > > My rule of thumb is to use true hardware when it's available, and to > > require it in "important" servers. Failing that, I use software raid. > > You get 80-90% of the performance of hardware raid in most use cases, > > and management is consistent from machine to machine. The > > "middleground" Fakeraid gets you none of the performance advantages of > > true hardware raid since everything is offloaded to the host anyway, > > and introduces a whole bunch of vendor-specific complications, so I > > avoid it entirely. > I'll throw in a 'yup, me too'. Fakeraid is a tremendous headache waiting > to happen. Either go true hardware raid (if you've got the budget AND > you can afford to have a spare card on hand) or go software raid. > > garl > > _______________________________________________ > EUGLUG mailing list > [email protected] > http://www.euglug.org/mailman/listinfo/euglug > _______________________________________________ EUGLUG mailing list [email protected] http://www.euglug.org/mailman/listinfo/euglug
