From bfab6b1508b8dd357f30928a26cf223d1e642878 Mon Sep 17 00:00:00 2001 From: Benjamin Kaduk Date: Sun, 16 Apr 2017 00:38:53 +0000 Subject: [PATCH] Add 2017Q1 pNFS entry from rmacklem Some additions to the introductory material by me, as well. --- .../news/status/report-2017-01-2017-03.xml | 49 +++++++++++++++++++ 1 file changed, 49 insertions(+) diff --git a/en_US.ISO8859-1/htdocs/news/status/report-2017-01-2017-03.xml b/en_US.ISO8859-1/htdocs/news/status/report-2017-01-2017-03.xml index d9ee0d8cde..3bdebc3e42 100644 --- a/en_US.ISO8859-1/htdocs/news/status/report-2017-01-2017-03.xml +++ b/en_US.ISO8859-1/htdocs/news/status/report-2017-01-2017-03.xml @@ -244,4 +244,53 @@ least 3 more bits out, one at each level. + + + pNFS server Plan B + + + + + Rick + Macklem + + rmacklem@FreeBSD.org + + + + +

Parallel NFS (pNFS) is an extension to the NFSv4 protocol that + allows for file accesses within a single logical mount to be + performed against multiple NFS file servers, "in parallel". + Thus, a pNFS server separats the MetaData operations from the + Data operations (Read/Write/Setattr of size N), letting + existing NFSv3 servers be used for the bulk data storage + while still taking advantage of NFSv4 protocol enhancements, + among other things.

+ +

My first attempt at a pNFS server using GlusterFS was a dud. + It worked, but performance was so poor that it was not + usable. This attempt that I call Plan B, only uses &os;, + with one &os; server handling the MetaData operations and K + &os; servers configured to serve Data. An NFSv4.1 client + that supports the pNFS File Layout will be able to + Read/Write to the Data servers directly, spreading out the + RPC load and allowing growth beyond that of what a single + &os; NFS server could achieve.

+ +

There is no support for the Flex Files Layout or mirroring + at this time. I hope to use the Flex Files Layout to add + mirroring support over the next year or so. Striping is + also not supported, but I have no plans for implementing it + at the moment.

+ +

Plan B is working quite well now and should be available + for testing by the end of April. I will announce how to do this on + the freebsd-fs@FreeBSD.org mailing list when it is available.

+ + + + Testing by others will be needed, once it is available. + +