We regularly test DAOS over DRAM (to emulate SCM). We found out that fallocate(2) isn’t always reliable over tmpfs and it is better to set “PMEMOBJ_CONF=prefault.at_create=1;” in the server’s yaml file if you want to
have stable performance.
Regarding scalability, we have tested DAOS at reasonable scale with both OPA and verbs, but we are still investigating several network-related issues (e.g. FI_UNIVERSE_SIZE limitation) that happen when scaling the total
number of nodes (i.e. client + server nodes).
Cheers,
Johann
From:
<daos@daos.groups.io> on behalf of Wu Huijun <huijunw91@...> Reply-To: "daos@daos.groups.io" <daos@daos.groups.io> Date: Saturday 6 June 2020 at 14:11 To: "daos@daos.groups.io" <daos@daos.groups.io> Subject: [daos] Scalability of DAOS on RAM simulation
Hi all,
Has anyone tried the max scalability of DAOS while using DRAM to simulate NVDIMM?
We continuously saw one or more server instances exited and several clients reported crt timeout when the number of clients increases.
Does it mean that DAOS only scales well under 3dxpoint but not the case for DRAM? Thanks!
Cheers,
Hector
---------------------------------------------------------------------
Intel Corporation SAS (French simplified joint stock company)
Registered headquarters: "Les Montalets"- 2, rue de Paris,
92196 Meudon Cedex, France
Registration Number: 302 456 199 R.C.S. NANTERRE
Capital: 4,572,000 Euros
This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.
