[Lustre-discuss] QOS between clients

Wallior, Julien Julien.Wallior at sig.com
Mon Sep 21 18:02:13 PDT 2009


Hi everyone,
In our lustre setup, we have a few hundred native linux clients and a few samba gateways sharing the filesystem over CIFS to about a hundred windows clients.
When the linux clients are idle, we can drive up to 400-500MB/s per samba gateway. When the linux clients are crushing the filesystem, we only get about 100MB/s per gateway. This makes sense, because in this case our storage backend is the bottleneck.
We are looking for a way to give a bigger share of the available bandwidth to the samba gateways. We played with max_rpcs_in_flight but that helped only so much.
Here is an idea we had:. mount the lustre filesystem multiple times on the samba gateways and share multiple SMB path out. We would then use some kind of load balancing between all the UNC paths. We tried it on a small lustre and it is possible to do it. We are a little worried that some race condition shows up when we do it on a larger scale.
How good or bad is that idea? Anybody else found a solution for this particular problem (other than add more gateways)?




Thanks,
Julien

IMPORTANT: The information contained in this email and/or its attachments is confidential. If you are not the intended recipient, please notify the sender immediately by reply and immediately delete this message and all its attachments. Any review, use, reproduction, disclosure or dissemination of this message or any attachment by an unintended recipient is strictly prohibited. Neither this message nor any attachment is intended as or should be construed as an offer, solicitation or recommendation to buy or sell any security or other financial instrument. Neither the sender, his or her employer nor any of their respective affiliates makes any warranties as to the completeness or accuracy of any of the information contained herein or that this message or any of its attachments is free of viruses.



More information about the lustre-discuss mailing list