[kwlug-disc] Synology, and proprietary disk formats ...

unsolicited unsolicited at swiz.ca
Tue Aug 5 22:33:28 EDT 2014


Yes, but the ECC versions of the memory itself has significant cost 
differentials.

I do take your point though, on a card ... for the amount of memory 
present, no good reason not to have ECC on such a card. But, same cost 
differential applies. A quick poke of 'ecc sata controller' at canada 
computers comes in at $630. Non-ECC RATHER less - but, to your point, 
rather pointless? vis a vis software raid, but I expect the same non/ecc 
argument can be similarly made even then.

Not what I meant on btrfs checksumming ...

A wonky bit of memory would presumably cause a checksum mismatch at 
verification, and so that hunk would never get committed to disk?

On 14-08-05 10:06 PM, L.D. Paniak wrote:
> It brings the same kind of grief that it would for ZFS.  These storage
> systems depend on the CPU/RAM subsystem for error-free calculation of
> checksums for data and, more importantly, metadata.  Flipped bits in
> data will ruin a file.  Flipped bits in metadata can ruin the whole
> storage system (pool, volume).
>
> In the modern world, there is no good reason for server/workstation
> hardware to not have ECC memory.  CPUs, motherboards and RAM that
> support ECC operation can be had for minimal premium over "regular"
> hardware.  Even the low-cost options like a $60 Pentium:
> http://ark.intel.com/products/74749/Intel-Pentium-Processor-G2030-3M-Cache-3_00-GHz?wapkw=g2030
> or low-power Atom Avoton board:
> http://www.supermicro.com/products/motherboard/Atom/X10/A1SAM-2550F.cfm
> are ECC-capable.
>
>
> On 08/05/2014 09:37 PM, unsolicited wrote:
>> Does btrfs' checksumming ability bring anything to a non-ECC party?
>>
>> As to the posters comment in the link of 'Shame on you for suggesting
>> software raid.' ... guess even I.T. keeps building better idiots.
>>
>> On 14-08-05 08:52 PM, L.D. Paniak wrote:
>>>
>>> On 08/05/2014 07:29 PM, Khalid Baheyeldin wrote:
>>>
>>>> But what caught my eye is this comment about the disk formats being
>>>> proprietary and unreadable on a Linux system
>>>>
>>>> http://soylentnews.org/comments.pl?sid=3232&cid=77644
>>>>
>>>> Hostage but for another reason ...
>>>>
>>>> Posting this for those who use Synology in this group ...
>>>>
>>>
>>> Thanks for the important PSA!  Be sure to expand the comments for bonus
>>> mention of the importance of ECC memory in software-RAID systems that
>>> checksum data written to disk (eg ZFS).
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> kwlug-disc mailing list
>>> kwlug-disc at kwlug.org
>>> http://kwlug.org/mailman/listinfo/kwlug-disc_kwlug.org
>>>
>>
>>
>> _______________________________________________
>> kwlug-disc mailing list
>> kwlug-disc at kwlug.org
>> http://kwlug.org/mailman/listinfo/kwlug-disc_kwlug.org
>
>
>
>
> _______________________________________________
> kwlug-disc mailing list
> kwlug-disc at kwlug.org
> http://kwlug.org/mailman/listinfo/kwlug-disc_kwlug.org
>





More information about the kwlug-disc mailing list