Skip to main content

Hi,

i have a question regarding the implementation of a cloud library with Scality Ring.

We can create two type of mount path S3 Compatible Storage or Scality Ring.

Which is required ? (i have some cloud libraries already created in S3 compatible storage instead of scality ring type).

There is a difference between them ?

Kind regards,

 

Christophe

Hi @christophe,

In most cases you want to pick the vendor specific library type rather than choosing the generic S3 connector. We typically co-develop or partner with storage vendors to use their optimized protocols that work best for their system - so you may find some benefits using the proprietary implementation (less overhead, more performance etc.).

I’m not sure about the benefits of this particular implementation but I’ll ask around.


Hi @christophe,

In most cases you want to pick the vendor specific library type rather than choosing the generic S3 connector. We typically co-develop or partner with storage vendors to use their optimized protocols that work best for their system - so you may find some benefits using the proprietary implementation (less overhead, more performance etc.).

I’m not sure about the benefits of this particular implementation but I’ll ask around.

Hi Damian,

our cloud libraries was implement with this type S3 compatible storage. i ask this because we currently have some issues with the backend with job integrity and performance.

Kind regards,


Hi @christophe,

In most cases you want to pick the vendor specific library type rather than choosing the generic S3 connector. We typically co-develop or partner with storage vendors to use their optimized protocols that work best for their system - so you may find some benefits using the proprietary implementation (less overhead, more performance etc.).

I’m not sure about the benefits of this particular implementation but I’ll ask around.

Hi Damian,

our cloud libraries was implement with this type S3 compatible storage. i ask this because we currently have some issues with the backend with job integrity and performance.

Kind regards,

Got it. You should never see integrity issues regardless of the library type! Performance could be for a few reasons. Usually we see slow restore or copy performance due to the increased latency of object storage - but usually mitigated by a larger dedupe object size if needed. Some systems prefer many many streams for best performance.


Reply