ALL NPPS Members

[Torre Wenaus <wenaus AT gmail.com>]

Regarding cost, the ATLAS experience several years ago I refer to that it takes expensive hardware to support S3 well, was interesting. Wen was in the middle of that, and Doug who has seen the document. If you mean using Amazon itself, yes that's even more expensive (and did not work well either back then). Our Rubin folks are writing to Google's S3 equivalent at a high enough rate to hit choke points getting through the SLAC network. But the issue here isn't using actual commercial clouds, but in-house S3 instances.

  Torre

On Fri, Apr 28, 2023 at 3:29 PM Maxim Potekhin via Phys-npps-members-l <phys-npps-members-l AT lists.bnl.gov> wrote:

Hi Brett,

Just my 0.25 cents from a discussion with SDCC, about two years back -
the claim was that there was available expertise (one person) to set up a
S3 access point + back end, and at the same time there was not enough
available effort to build a well scaled XRootD cluster. Take it "for
what it's worth",
but I think I got main points right.

Maxim


On 4/28/2023 3:22 PM, Brett Viren via Phys-npps-members-l wrote:
> Torre Wenaus via Phys-npps-members-l <phys-npps-members-l AT lists.bnl.gov>
> writes:
>
>> FYI this googledoc is for the ePIC S&C conveners so they will see its content. COmmunications best kept among
>> us can be put on this email thread
> I'm rather ignorant of S3 but is cost an issue?  XrootD lets us use
> disks and bandwidth we already own.
>
> -Brett.
>
> _______________________________________________
> Phys-npps-members-l mailing list
> Phys-npps-members-l AT lists.bnl.gov
> https://lists.bnl.gov/mailman/listinfo/phys-npps-members-l

_______________________________________________
Phys-npps-members-l mailing list
Phys-npps-members-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/phys-npps-members-l

--

-- Torre Wenaus, BNL NPPS Group, ATLAS Experiment

-- BNL 510A 1-222 | 631-681-7892