[pve-devel] [RFC qemu/storage/qemu-server/container/manager 00/23] backup provider API

[Jonathan Nicklin]

> On Jul 27, 2024, at 11:20 AM, Dietmar Maurer <dietmar at proxmox.com> wrote:
> 
>> Would adding support for offloading incremental difference detection
>> to the underlying storage be feasible with the API updates? The QEMU
>> bitmap strategy works for all storage devices but is far from
>> optimal.
> 
> Sorry, but why do you think this is far from optimal?
> 

The biggest issue we see reported related to QEMU bitmaps is
persistence. The lack of durability results in unpredictable backup
behavior at scale. If a host, rack, or data center loses power, you're
in for a full backup cycle. Even if several VMs are powered off for
some reason, it can be a nuisance. Several storage solutions can
generate the incremental difference bitmaps from durable sources,
eliminating the issue.

That said, using bitmaps or alternative sources for the incremental
differences is slightly orthogonal to the end goal. The real
improvement we're hoping for is the ability to eliminate backup
traffic on the client.

Today, I believe the client is reading the data and pushing it to
PBS. In the case of CEPH, wouldn't this involve sourcing data from
multiple nodes and then sending it to PBS? Wouldn't it be more
efficient for PBS to read it directly from storage? In the case of
centralized storage, we'd like to eliminate the client load
completely, having PBS ingest increment differences directly from
storage without passing through the client.