Filesystem choice for fileserver?

[Shannon Hendrix]

On Apr 18, 2008, at 16:31 , Ron Wickersham wrote:

> i'm also planning to put a ZFS file server together (an E450 so i'll  
> have
> good disk bandwidth) and have read lots (but certainly not all) of the
> Sun blogs and papers on ZFS.   there were several recommendations to  
> use
> JBOD because this ensures that the ZFS does its job and that when  
> you use
> other software or hardware RAID you don't get the full ZFS advantage  
> for
> bit-rot and other features that ensure data integrity.     so after a
> failure, the other software or hardware RAID system makes sure that  
> the
> copies are consistent (in the case of full mirror, both copies will be
> identical even if both have errors) and ZFS can't fix the error if i  
> understand the point correctly.

This is partially true for current Solaris releases, and not true in  
the near future.

Current official releases of ZFS have redundant metadata blocks, so  
they can correct filesystem data corruption even on a single drive.

Future releases of ZFS, which I believe only exist in the Nevada/
Indiana releases right now, will let you have data redundancy as well.

Once this is done or you download one of the "beta" versions of  
Solaris, then ZFS will be able to correct errors even without RAID and  
multiple drives.

--
Shannon Hendrix
shannon@...
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Carl R. Friend]

    On Sat, 19 Apr 2008, Shannon Hendrix wrote:

> Has anyone actually done a CPU in hardware?

    I know that this was intended in a humourous vein, but give
the MUNIAC a look -- a modern 12-bit machine done with vacuum
tubes.

    Really -- there was life before microcode and firmware.  Oh, and
ECC, too, which can be applied to disks if one wants to expend the
energy.

+------------------------------------------------+---------------------+
| Carl Richard Friend (UNIX Sysadmin)            | West Boylston       |
| Minicomputer Collector / Enthusiast            | Massachusetts, USA  |
| mailto:crfriend@...                        +---------------------+
| http://users.rcn.com/crfriend/museum           | ICBM: 42:22N 71:47W |
+------------------------------------------------+---------------------+
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Lionel Peterson]

>From: Shannon Hendrix <shannon@...>
>Date: 2008/04/19 Sat PM 04:11:08 CDT
>To: The Rescue List <rescue@...>
>Subject: Re: [rescue] Filesystem choice for fileserver?

>On Apr 18, 2008, at 10:10 , der Mouse wrote:
>
>>>> A lot of hardware RAID cards have horribly slow processors on them.
>>> [..."hardware RAID"...]
>>
>> Does anyone actually do hardware RAID?  I've never seen any.  All the
>> supposedly-hardware RAID I've seen is actually software RAID running  
>> on
>> a dedicated processor (what would more honestly be called firmware
>> RAID).  Usually it doesn't even try to hide that - for example, I've
>> seen POST-time messages like "booting controller kernel".
>>
>> Have I missed something?  Is there anyone who actually does RAID in
>> hardware?
>
>Well, thinking that way:
>
>I've never seen a hardware CPU.  They all are programmed in microcode  
>or "fixed logic instructions".
>
>Has anyone actually done a CPU in hardware?
>
>:)

Following your logic, in order to sell a controller as "hardware RAID" you'd have to fully implement RAID functionality in hard-wired gates, with no "soft logic" inside? That threshold is a bit high - I think it would be fair to set the bar at "RAID implemented to not rely on host RAM and/or CPU"...

Lionel
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[der Mouse]

> Following your logic, in order to sell a controller as "hardware
> RAID" you'd have to fully implement RAID functionality in hard-wired
> gates, with no "soft logic" inside?

Mostly, yes.  I'm not sure whether I'd count an FPGA as hardware for
those purposes.

> That threshold is a bit high - I think it would be fair to set the
> bar at "RAID implemented to not rely on host RAM and/or CPU"...

I disagree.  Firmware RAID, coprocessor RAID, there are plenty of
perfectly good _accurate_ terms to describe this sort of "auxiliary CPU
dedicated to RAID" setup.  Hardware RAID should, it seems to me, be
reserved for RAID done in silicon instead of software, even if that
software is running on a dedicated CPU.

Your "threshold is a bit high" sounds as though there is some kind of
privilege being granted.  I'm trying to be descriptive here.  There's
nothing wrong with firmware RAID; the term is not a pejorative - it's
just that "hardware RAID" is a wrong term for it.  Just as, for
example, there's nothing wrong with using nails to fasten wood
together, but "gluing" is a wrong term for it.

/~\ The ASCII der Mouse
\ / Ribbon Campaign
 X  Against HTML       mouse@...
/ \ Email!     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Shannon Hendrix]

On Apr 19, 2008, at 18:35 , Carl R. Friend wrote:

>   On Sat, 19 Apr 2008, Shannon Hendrix wrote:
>
>> Has anyone actually done a CPU in hardware?
>
>   I know that this was intended in a humourous vein, but give
> the MUNIAC a look -- a modern 12-bit machine done with vacuum
> tubes.
>
>   Really -- there was life before microcode and firmware.  Oh, and
> ECC, too, which can be applied to disks if one wants to expend the
> energy.

Very true... although I remember in college we used to have lively  
debates about exactly where you draw the line.

Technically speaking an ALU made up of tubes is still instructions,  
they are just non-flexible... :)





--
Shannon Hendrix
shannon@...
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Shannon Hendrix]

On Apr 20, 2008, at 00:42 , Lionel Peterson wrote:

> Following your logic, in order to sell a controller as "hardware  
> RAID" you'd have to fully implement RAID functionality in hard-wired  
> gates, with no "soft logic" inside? That threshold is a bit high - I  
> think it would be fair to set the bar at "RAID implemented to not  
> rely on host RAM and/or CPU"...

Following my logic would have led you to a joke.

Hardware RAID is a coprocessor.  Same basic definition applies.  The  
work is not being done by the host.

Otherwise, how many sub-atomic state changes are involved in  
incrementing a register?

        :-)



--
"Where some they sell their dreams for small desires."
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Lionel Peterson]

>From: Shannon Hendrix <shannon@...>
>Date: 2008/04/24 Thu PM 10:45:19 CDT
>To: The Rescue List <rescue@...>
>Subject: Re: [rescue] Filesystem choice for fileserver?

>On Apr 20, 2008, at 00:42 , Lionel Peterson wrote:
>
>> Following your logic, in order to sell a controller as "hardware  
>> RAID" you'd have to fully implement RAID functionality in hard-wired  
>> gates, with no "soft logic" inside? That threshold is a bit high - I  
>> think it would be fair to set the bar at "RAID implemented to not  
>> rely on host RAM and/or CPU"...

I should also have included "software on the host system" (no need for special drivers, etc.), but I thought that "host RAM" sort of covered that... IMHO, an add-in card that completely handles all RAID controller work, and simply presents itself as a disk drive controller, with each "container" presented to the OS as if it were a single, physical device would qualify as a hardware-implementation.

>Following my logic would have led you to a joke.

I guess I didn't follow it... ;^)

>Hardware RAID is a coprocessor.  Same basic definition applies.  The  
>work is not being done by the host.

Agreed.

Lionel
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue