Add a section to nfs(5) that discusses the trade-offs between the UDP and
TCP transports.
Signed-off-by: Chuck Lever <[email protected]>
---
utils/mount/nfs.man | 56 +++++++++++++++++++++++++++++++++++++++++++++++++++
1 files changed, 56 insertions(+), 0 deletions(-)
diff --git a/utils/mount/nfs.man b/utils/mount/nfs.man
index 10e5cd2..0debdb0 100644
--- a/utils/mount/nfs.man
+++ b/utils/mount/nfs.man
@@ -683,6 +683,62 @@ This example can be used to mount /usr over NFS.
.TA 2.5i +0.7i +0.7i +.7i
server:/export/share /usr nfs ro,nolock,nocto,actimeo=3600 0 0
.FI
+.SH "TRANSPORT METHODS"
+NFS clients send requests to NFS servers via
+Remote Procedure Calls, or
+.IR RPCs .
+The RPC client discovers remote service endpoints automatically,
+handles per-request authentication,
+adjusts request parameters for different byte endianness on client and server,
+and retransmits requests that may have been lost by the network or server.
+RPC requests and replies flow over a network transport.
+.P
+The
+.BR mount (8)
+command, the NFS client, and the NFS server
+can usually automatically negotiate proper transport
+and data transfer size settings for a mount point.
+There are some cases, however, where it pays to specify
+these settings explicitly using mount options.
+.P
+TCP is the default transport for all modern NFS implementations.
+It performs well in almost every conceivable network environment
+and provides excellent guarantees against data corruption
+due to network unreliability.
+TCP is often a requirement for mounting a server through a network firewall.
+.P
+The UDP transport has many limitations
+that prevent smooth operation and good performance
+in some common deployment environments.
+However, UDP can be quite effective in specialized settings
+where the network's MTU is large relative to NFS's data transfer size.
+This includes the use of jumbo Ethernet frames
+or high bandwidth local area networks.
+Thus, with UDP, trimming the
+.B rsize
+and
+.B wsize
+settings
+so that each NFS read or write request fits in just a few network frames,
+or even in a single frame, is advised.
+This reduces the probability that the loss of a single MTU-sized network frame
+results in the loss of an entire NFS request.
+TCP itself manages the reliability of network transmissions,
+thus
+.B rsize
+and
+.B wsize
+can safely be allowed to default to the largest settings
+supported by both client and server.
+.P
+Reasonable timeout and retransmit settings for NFS over TCP
+are in the one to several minute range
+because the TCP transport already manages network-related packet loss.
+For UDP, even an insignificant packet loss rate results in the loss of
+whole NFS requests,
+thus retransmit timeouts are usually in the subsecond range.
+The Linux RPC client employs an RTT estimator that
+dynamically manages the timeout settings for requests sent via UDP.
.SH FILES
.I /etc/fstab
.SH "SEE ALSO"
-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs