From: Andi Kleen <andi@firstfloor.org>
To: Andrew Grover <andy.grover@gmail.com>
Cc: Andi Kleen <andi@firstfloor.org>,
Andy Grover <andy.grover@oracle.com>,
netdev@vger.kernel.org, rds-devel@oss.oracle.com,
general@lists.openfabrics.org
Subject: Re: [ofa-general] [PATCH 0/26] Reliable Datagram Sockets (RDS), take 2
Date: Sat, 28 Feb 2009 23:36:53 +0100 [thread overview]
Message-ID: <20090228223653.GD26292@one.firstfloor.org> (raw)
In-Reply-To: <c0a09e5c0902281244u356acbecn98f37ac3abd5ebc0@mail.gmail.com>
> The previous solution for IPC that Oracle was using was based on UDP,
> which I think could be considered very close to using raw sockets --
> each process is responsible for its own acks, retransmits, everything.
> Doing this on a highly loaded machine resulted in a cascade where
> performance got worse and worse.
Could you describe that cascade in more detail?
The problem was that the retransmits didn't have high enough priority?
> Additionally, our interconnect is primarily Infiniband. It natively
> implements a reliable datagram connection type so RDS leverages that.
So perhaps it would make more sense to have a thin direct interface
to that IB service? Or perhaps it already exists? (I admit I don't know
the IB interfaces very well)
-andi
--
ak@linux.intel.com -- Speaking for myself only.
next prev parent reply other threads:[~2009-02-28 22:18 UTC|newest]
Thread overview: 48+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-02-25 1:30 [ofa-general] [PATCH 0/26] Reliable Datagram Sockets (RDS), take 2 Andy Grover
2009-02-25 1:30 ` [ofa-general] [PATCH 01/26] RDS: Socket interface Andy Grover
2009-02-25 1:30 ` [PATCH 02/26] RDS: Main header file Andy Grover
2009-02-25 1:30 ` [PATCH 03/26] RDS: Congestion-handling code Andy Grover
2009-02-25 1:30 ` [PATCH 04/26] RDS: Transport code Andy Grover
2009-02-25 1:30 ` [PATCH 05/26] RDS: Info and stats Andy Grover
2009-02-25 1:30 ` [PATCH 06/26] RDS: Connection handling Andy Grover
2009-02-25 1:30 ` [PATCH 07/26] RDS: loopback Andy Grover
2009-02-25 1:30 ` [PATCH 08/26] RDS: sysctls Andy Grover
2009-02-25 1:30 ` [PATCH 09/26] RDS: Message parsing Andy Grover
2009-02-25 1:30 ` [PATCH 10/26] RDS: send.c Andy Grover
2009-02-25 1:30 ` [PATCH 11/26] RDS: recv.c Andy Grover
2009-02-25 1:30 ` [PATCH 12/26] RDS: RDMA support Andy Grover
2009-02-25 1:30 ` [PATCH 13/26] RDS/IB: Infiniband transport Andy Grover
2009-02-25 1:30 ` [PATCH 14/26] RDS/IB: Ring-handling code Andy Grover
2009-02-25 1:30 ` [PATCH 15/26] RDS/IB: Implement RDMA ops using FMRs Andy Grover
2009-02-25 1:30 ` [PATCH 16/26] RDS/IB: Implement IB-specific datagram send Andy Grover
2009-02-25 1:30 ` [PATCH 17/26] RDS/IB: Receive datagrams via IB Andy Grover
2009-02-25 1:30 ` [PATCH 18/26] RDS/IB: Stats and sysctls Andy Grover
2009-02-25 1:30 ` [PATCH 19/26] RDS: Add iWARP support Andy Grover
2009-02-25 1:30 ` [PATCH 20/26] RDS: Common RDMA transport code Andy Grover
2009-02-25 1:30 ` [PATCH 21/26] RDS: Documentation Andy Grover
2009-02-25 1:30 ` [PATCH 22/26] RDS: Kconfig and Makefile Andy Grover
2009-02-25 1:30 ` [PATCH 23/26] RDS: Add AF and PF #defines for RDS sockets Andy Grover
2009-02-25 7:26 ` [ofa-general] " David Miller
2009-02-25 11:36 ` Rémi Denis-Courmont
2009-02-25 11:58 ` David Miller
2009-02-25 18:45 ` Andrew Grover
2009-02-25 1:30 ` [PATCH 24/26] RDS: Add MAINTAINERS entry Andy Grover
2009-02-25 1:30 ` [PATCH 25/26] RDS: Add userspace header Andy Grover
2009-02-25 1:30 ` [PATCH 26/26] RDS: Add RDS to AF key strings Andy Grover
2009-02-25 7:28 ` [PATCH 0/26] Reliable Datagram Sockets (RDS), take 2 David Miller
2009-02-25 8:04 ` [ofa-general] " Or Gerlitz
2009-02-25 8:06 ` David Miller
2009-02-25 9:46 ` Or Gerlitz
2009-02-25 18:16 ` Roland Dreier
2009-02-25 18:43 ` Andrew Grover
2009-02-25 21:45 ` [ofa-general] " David Miller
2009-02-27 8:01 ` David Miller
2009-02-27 18:21 ` Andrew Grover
2009-02-27 17:08 ` [ofa-general] " Andi Kleen
2009-02-28 1:53 ` ***SPAM*** " Andrew Grover
2009-02-28 5:56 ` Andi Kleen
2009-02-28 20:44 ` ***SPAM*** " Andrew Grover
2009-02-28 22:36 ` Andi Kleen [this message]
2009-03-01 0:58 ` Andrew Grover
2009-03-01 1:50 ` Andi Kleen
2009-03-03 22:19 ` ***SPAM*** " Andrew Grover
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20090228223653.GD26292@one.firstfloor.org \
--to=andi@firstfloor.org \
--cc=andy.grover@gmail.com \
--cc=andy.grover@oracle.com \
--cc=general@lists.openfabrics.org \
--cc=netdev@vger.kernel.org \
--cc=rds-devel@oss.oracle.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).