Thread: Why you should turn on Checksums with SSDs

Why you should turn on Checksums with SSDs

From
Josh Berkus
Date:
Explained here:
https://www.usenix.org/system/files/conference/fast13/fast13-final80.pdf

13 out of 15 tested SSD's had various kinds of corruption on a power-out.

(thanks, Neil!)

--
Josh Berkus
PostgreSQL Experts Inc.
http://pgexperts.com


Re: Why you should turn on Checksums with SSDs

From
"Tomas Vondra"
Date:
On 30 Červenec 2014, 5:12, Josh Berkus wrote:
> Explained here:
> https://www.usenix.org/system/files/conference/fast13/fast13-final80.pdf
>
> 13 out of 15 tested SSD's had various kinds of corruption on a power-out.
>
> (thanks, Neil!)

Well, only four of the devices supposedly had a power-loss protection
(battery, capacitor, ...) so I guess it's not really that surprising the
remaining 11 devices failed in a test like this. Although it really
shouldn't damage the device, as apparently happened during the tests.

Too bad they haven't mentioned which SSDs they've been testing
specifically. While I understand the reason for that (HP Labs can't just
point at products from other companies), it significantly limits the
usefulness of the study. Too many companies are producing crappy
consumer-level devices, advertising them as "enterprise". I could name a
few ...

Maybe it could be deciphered using the information in the paper
(power-loss protection, year of release, ...).

I'd expect to see Intel 320/710 to see there, but that seems not to be the
case, because those devices were released in 2011 and all the four devices
with power-loss protection have year=2012. Or maybe it's the year when
that particular device was manufactured?


regards
Tomas



Re: Why you should turn on Checksums with SSDs

From
Merlin Moncure
Date:
On Wed, Jul 30, 2014 at 4:01 AM, Tomas Vondra <tv@fuzzy.cz> wrote:
> On 30 Červenec 2014, 5:12, Josh Berkus wrote:
>> Explained here:
>> https://www.usenix.org/system/files/conference/fast13/fast13-final80.pdf
>>
>> 13 out of 15 tested SSD's had various kinds of corruption on a power-out.
>>
>> (thanks, Neil!)
>
> Well, only four of the devices supposedly had a power-loss protection
> (battery, capacitor, ...) so I guess it's not really that surprising the
> remaining 11 devices failed in a test like this. Although it really
> shouldn't damage the device, as apparently happened during the tests.
>
> Too bad they haven't mentioned which SSDs they've been testing
> specifically. While I understand the reason for that (HP Labs can't just
> point at products from other companies), it significantly limits the
> usefulness of the study. Too many companies are producing crappy
> consumer-level devices, advertising them as "enterprise". I could name a
> few ...
>
> Maybe it could be deciphered using the information in the paper
> (power-loss protection, year of release, ...).
>
> I'd expect to see Intel 320/710 to see there, but that seems not to be the
> case, because those devices were released in 2011 and all the four devices
> with power-loss protection have year=2012. Or maybe it's the year when
> that particular device was manufactured?

Take a look here:
http://hardware.slashdot.org/story/13/12/27/208249/power-loss-protected-ssds-tested-only-intel-s3500-passes

"Only the end-of-lifed Intel 320 and its newer replacement, the S3500,
survived unscathed. The conclusion: if you care about data even when
power could be unreliable, only buy Intel SSDs.""

merlin