Cloud Backup

Cloud Backup

E.S. Rosenberg esr+linux-il at g.jct.ac.il
Sun May 19 17:37:38 IDT 2013


I think Shachar is missing one point about S3 and similar amazon services.

You are assuming that amazon created infrastructure specifically for S3.
S3 and other products is amazon renting off it's over-capacity, as
such it *pays* for amazon to have a very reliable and stable
infrastructure because it's for their total business and amazon.com
being down probably costs them in the order of millions per second.

That is why for amazon it pays to make big investments in
infrastructure and renting space there is probably mainly offsetting
the cost of building very robust infrastructure...
Now they probably invest in extra infrastructure specifically for S3
and similar products...

Other then that no one in their right mind making even 5 nines SLA
claims will accepts responsibility for downtime you suffer as a result
of problems on your side, "I" the company commit that my systems will
be available and reachable from the Internet, if your ISP has a
problem you can't blame me (that's why "I" built datacenters in
multiple locations all over the world and you can fail-over instantly
and transparently to any of them).

I don't work for amazon, but this is what it looks like from my end.
Regards,
Eliyahu - אליהו

2013/5/18 Ghiora Drori <ghioradrori at gmail.com>:
> Shachar Shemesh enjoys being rude and wrong.
> I suggest he install new fuses.
>
>
>
> On Fri, May 17, 2013 at 10:05 AM, Shachar Shemesh <shachar at shemesh.biz>
> wrote:
>>
>> On 17/05/13 11:43, Oleg Goldshmidt wrote:
>>
>> Shachar Shemesh <shachar at shemesh.biz> writes:
>>
>> On 17/05/13 10:13, Ghiora Drori wrote:
>>
>>     https://aws.amazon.com/glacier/#highlights
>>     Quote: "Amazon Glacier is designed to provide average annual
>>     durability of 99.999999999% "
>>
>>     If this is not good enough for you too bad.
>>
>>
>> When you see someone, anyone, saying such a thing, run. As fast and as
>> far as you can.
>>
>> This level of assurance is called "nine nines"(henceforth 9*9). It
>> amounts to one thousandth of a second of downtime a year.
>>
>> I think you are misreading the claim, Shachar. It is not about
>> availablity, it is about "durability". I read it as a measure of the
>> probability that your data will not be lost before a year passes.
>>
>> You are right that this is not about availability. The previous response
>> was my fuse jumping because of the pure ludiciority of people claiming 9*9
>> availability. After reading the actual text, however, it is not clear what
>> it is about.
>>
>> It is possible that this means that they will lose (on average) ten bits
>> per Terabyte per year. If that is the case, honestly, this does not sound
>> very good. Assuming they have several exabytes of customers data, this means
>> that they have several actual cases of customer data lose all the time. Not
>> a particularly good track record.
>>
>> Or, and this is the more likely scenario, they are talking out of their
>> asses, and put the number in because it sounds impressive.
>>
>> Omer Zak wrote:
>>
>> IMO, the quote does not promise a nine nines assurance.
>> It only says that Amazon Glacier WAS DESIGNED to provide this kind of
>> assurance.
>>
>> See my previous comment for why this is equally ludicrous.
>>
>> Shachar
>>
>> Disclaimer: I have never used the service and th above is my "common
>> sense and reading comprehension" take on what is written in the above
>> website.
>>
>>
>
>
>
> --
> Mark Twain - "If you don't read the newspaper, you're uninformed. If you
> read the newspaper, you're mis-informed."
>
> _______________________________________________
> Linux-il mailing list
> Linux-il at cs.huji.ac.il
> http://mailman.cs.huji.ac.il/mailman/listinfo/linux-il
>



More information about the Linux-il mailing list