Skip to Content.
Sympa Menu

perfsonar-user - Re: [perfsonar-user] Amazon / Wasabi S3 storage performance utility for perfSONAR

Subject: perfSONAR User Q&A and Other Discussion

List archive

Re: [perfsonar-user] Amazon / Wasabi S3 storage performance utility for perfSONAR


Chronological Thread 
  • From: James Deaton <>
  • To: Alex Moura <>, Edward Colone <>
  • Cc:
  • Subject: Re: [perfsonar-user] Amazon / Wasabi S3 storage performance utility for perfSONAR
  • Date: Thu, 19 Dec 2019 10:34:47 -0600

Ed,

My apologies for not catching up on some of these emails until today but I wanted to state that I found a broad set of interest in this concept. However there were some concerns about how such benchmarking may impact costs given that S3 and some of the other S3 compatible providers have charges for egress and storage that may need to be taken into account.

The benchmark you linked tests both the performance of the object creation and deletion of storage providers as well as the throughput in the tests. I’ve done a handful of tests against both Wasabi and AWS but I’ve not had an opportunity to judge impact on potential provider costs at scale. I still think this type of test provides a very close to production environment test case that would serve the community well. We have not reliably (and affordably) had success with test points in the commercial cloud but perhaps this is a path to develop baselines for the act of moving data in and out of the commercial cloud.

-- James
On Dec 17, 2019, 2:27 PM -0600, Edward Colone <>, wrote:
Alex, I believe that is correct.  It is a way for someone to benchmark performance from their deployed perfSONAR testpoints to S3 cloud storage.  This would include anyone providing S3 as a service: Amazon itself, Wasabi, or any other provider.

Anyone please correct me if I am wrong.

Thanks,

-Ed


On Tue, Dec 17, 2019 at 3:11 PM Alex Moura <> wrote:
Hi, Ed

Just to confirm my understanding about this tool: it would allow a pS node to perform tests to AWS Cloud provider only and the results would show the network performance between that pS node and the chosen AWS region, is this correct? I mean this component would allow one to verify the network performance to the AWS cloud storage provider only, and would not perform tests between different pS nodes.

Regards
 

----- Em 17 de Dez de 2019, em 16:45, Edward Colone <> escreveu:
TIME SENSITIVE EMAIL
I would like a response right now if anyone is interested in the Wasabi S3 storage benchmarking tool to be incorporated in to perfSONAR.  We may have a very short term resource availability, and if this is of community interest it would be good to know.


Please respond in the next 48 hours, if possible.

Thanks,

-Ed

--
To unsubscribe from this list: https://lists.internet2.edu/sympa/signoff/perfsonar-user
--
To unsubscribe from this list: https://lists.internet2.edu/sympa/signoff/perfsonar-user



Archive powered by MHonArc 2.6.19.

Top of Page