Macklem, Rick, "Lessons Learned Tuning the 4.3BSD Reno Implementation of the NFS Protocol," Winter USENIX Conference Proceedings, USENIX Association, Berkeley, CA, January 1991. Describes performance work in tuning the 4.3BSD Reno NFS implementation. Describes performance improvement (reduced CPU loading) through elimination of data copies. Home/Search Document Not in Database Summary Related Articles Check |
This paper is cited in the following contexts:
Linux NFS Client Write Performance - Lever, Honeyman (2001) (6 citations) (Correct)
....investigate NFS behavior on high performance networks, but do not address implementation specific issues in existing clients [10] The closest previous work we found describes performance improvement (reduced CPU loading) through elimination of data copies in the 4. 3BSD Reno NFS implementation [8]. 2.3. Inter run variance on Linux Our experience with performance measurement on Linux has taught us to expect large variations in results between individual benchmark runs on the same O S version and software and hardware configurations. Other benchmarks performed by the authors in the past ....
Macklem, R. "Lessons Learned Tuning the 4.3BSD Reno Implementation of the NFS Protocol." USENIX Technical Conference Proceedings, January 1991.
NFS Dynamics Over Flow-Controlled Wide Area Networks - Chang, al. (1997) (2 citations) (Correct)
....of three factors: high drop rate, long propagation delay, and low available bandwidth. The latter is most significant for NFS operations such as read and write that involve large amounts of data. However, large transfers are substantially less common than small ones in most NFS workloads [4]. In such cases, NFS performance is affected more by packet drop rate and propagation delay. Our research explores four issues relating to NFS and WANs. ATM WANs may use flow control to reduce losses; NFS is affected by both the flow control and the remaining losses. NFS can be used over TCP ....
.... less CPU overhead than TCP [9] and because NFS is almost always used on local networks where TCP s sophisticated error recovery and congestion control are not required [5] However, implementations of NFS on TCP have recently demonstrated performance better than UDP over congested networks [4]. The reason is that TCP has better congestion control and faster error recovery than UDP based NFS. Whenever a packet is lost, NFS UDP waits a timeout period on the order of one second and retransmits the entire request, which might involve multiple packets. TCP, in contrast, can often respond to ....
[Article contains additional citation context not shown here]
Rick Macklem, "Lessons Learned Tuning the 4.3 BSD Reno Implementation of the NFS Protocol", Winter USENIX Conference, Dallas, TX, 1991.
Use of Link-by-Link Flow Control in Maximizing ATM Network.. - Kung, al. (1993) (21 citations) (Correct)
....flow control. Consider the NFS [17] file system protocol, which is based on a remote procedure call (RPC) mechanism. NFS implementations usually adjust time out periods based on round trip times, but do not explicitly adjust the amount of load offered in response to congestion (though see [15]) The underlying RPC mechanism limit the load to some extent, since clients must often wait for the reply to one RPC before sending the next. However, NFS requests and responses range in size from a few hundred bytes to 8 kilobyte disk blocks, so this limit is not very stringent or precise. ....
R. Macklem, "Lessons Learned Tuning the 4.3BSD Reno Implementation of the NFS Protocol," USENIX Winter Conference Proceedings, January 1991.
Unknown - (Correct)
No context found.
Macklem, Rick, "Lessons Learned Tuning the 4.3BSD Reno Implementation of the NFS Protocol," Winter USENIX Conference Proceedings, USENIX Association, Berkeley, CA, January 1991. Describes performance work in tuning the 4.3BSD Reno NFS implementation. Describes performance improvement (reduced CPU loading) through elimination of data copies.
Nfs Version 4 Protocol - Shepler, Callaghan, Robinson.. (2000) (10 citations) (Correct)
No context found.
Macklem, Rick, "Lessons Learned Tuning the 4.3BSD Reno Implementation of the NFS Protocol," Winter USENIX Conference Proceedings, USENIX Association, Berkeley, CA, January 1991. Describes performance work in tuning the 4.3BSD Reno NFS implementation. Describes performance improvement (reduced CPU loading) through elimination of data copies.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC