Loading...

Xporter

XML

Word

Printable

ARTs:

Services
Benefit hypothesis:

Hide

Provide a clear path towards a potential ECP in PI9.

Show
Provide a clear path towards a potential ECP in PI9.
Acceptance criteria:
Hide

At least one of the following needs to be provided:

Lightweight Software Document on the results of the investigation on RDMA using RoCE v2 (NZAPP implementation), and use of conventional sockets with io_uring backend, with recommendation on whether the baseline (SPEAD2 over normal sockets) needs to be changed, including potential impact on science performance (or demonstration that there are none, or improve the baseline).

Stretch: Draft ECP with the recommendation above, and order-of-magnitude costing.
Show
At least one of the following needs to be provided: Lightweight Software Document on the results of the investigation on RDMA using RoCE v2 (NZAPP implementation), and use of conventional sockets with io_uring backend, with recommendation on whether the baseline (SPEAD2 over normal sockets) needs to be changed, including potential impact on science performance (or demonstration that there are none, or improve the baseline). Stretch: Draft ECP with the recommendation above, and order-of-magnitude costing.
Feature Points:
2
WSJF:
7.5
Epic Link:
Platform & Networks
Agile Teams:

Team_PLANET
Story Point Burn-up:
Overdue:

In order to evaluate if an ECP for the CSP SDP visibility protocol is required, we will produce a comparison of the various available options.

Story Point Burn-up: (0%)

Feature Estimate: 2.0

Produce a comparison of the various options for SDP ingest