Modularity and efficiency in protocol implementation

Clark, D. D.

doi:10.17487/rfc0817

Cited by 58 publications

(29 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another scheme is to mer ge the copying and checksum operations to reduce transfers between memory and the CPU [8,11,26]. On some machines this strate gy is sufficient to result in the ef fective elimination of checksum o verhead [16].…”

Section: Related Workmentioning

confidence: 99%

Profiling and reducing processing overheads in TCP/IP

Kay

Pasquale

1996

IEEE/ACM Trans. Networking

View full text Add to dashboard Cite

This paper presents detailed measurements of processing overheads for the Ultrix 4.2a implementation of TCP/IP network software running on a DECstation 5000/200. The performance results were used to unco ver throughput and latency bottlenecks. We present a scheme for impro ving throughput when sending large messages by a voiding most checksum computations in a relatively safe manner. We also show that for the implementation we studied, reducing latency (when sending small messages) is a more dif ficult problem because processing overheads are spread over many operations; gaining a significant savings would require the optimization of many different mechanisms. This is especially important because, when processing a realistic w orkload, we ha ve found that non-data-touching operations consume more time in aggre gate than data-touching operations. IntroductionWe analyze TCP/IP [30] and UDP/IP [29] processing overheads given a real w orkload on a DECstation 5000/200 running Ultrix 4.2a, and we use this information to guide our development of new optimizations. The cost of various processing overheads depend on message size; consequently, our optimizations take into account the message size distributions derived from the network traffic in our environment (which is not atypical of many academic and office environments).In our analysis of the TCP/IP and UDP/IP LAN and WAN traffic we were able to collect, we fi nd that message sizes are f ar from uniformly distributed; rather, most messages are either very small or very large. Small messages are usually used to carry control information, whereas lar ge messages typically carry bulk data. Different kinds of optimizations can improve processing speed for each type of traffic; in this paper we discuss both.Typical processing time breakdowns for short (i.e. 64-128 byte) control messages fundamentally differ from those of long multiple kilobyte data messages. The processing time of large messages is dominated by data-touching operations such as cop ying and computing checksums [4, 8-11, 15, 24, 36] because these operations must be applied to each byte. However, small messages have few bytes of data, and thus their processing time is dominated by non-data-touching operations.To optimize processing of large messages, we describe a checksum redundancy a voidance algorithm that eliminates most checksum processing without sacrificing reliability. Since checksum processing alone consumes nearly half the total processing time (of large messages), this optimization improves throughput considerably.On both the LAN and WAN we studied, both of which are typical Unix-networking environments, small messages far outnumber large messages. In fact, even though processing a large message requires more time, the large proportion of small messages causes the cumulative non-data-touching processing time to exceed the cumulative data-touching processing time. We show that it w ould be difficult to significantly reduce the a verage processing time of non-datatouching overheads, at lea...

show abstract

Section: Related Workmentioning

confidence: 99%

Profiling and reducing processing overheads in TCP/IP

Kay

Pasquale

1996

IEEE/ACM Trans. Networking

View full text Add to dashboard Cite

show abstract

“…The amount of information transferred is the same for both dissemination algorithms. The disadvantage of splitting the packets is that with each extra packet transferred there are additional requirements such as buffering and routing in the network [15].…”

Section: System Models and Simulationsmentioning

confidence: 99%

On packet loss performance under varying network conditions with path diversity

Haywood

Peng

2008

Proceedings of the 2008 International Conference on Advanced Infocomm Technology - ICAIT '08

View full text Add to dashboard Cite

In this paper we investigate the properties of packet losses when packets are transferred using an overlay network. Three packet loss models are established for different packet loss scenarios: random, burst and congestion. Based on these models, it has been shown through simulation that in random and burst packet losses environments the packet loss rate increases as the number of diverse paths increases, whilst the opposite is true in the congestion case. A combined model is presented and then developed to address the situation where the last link of the system experiences different loss patterns to the rest of the network, such as a wireless access link.

show abstract

“…However, more data has to be transferred in the smaller packet case as each packet requires a header, and this could result in more packet losses because the likelihood of random errors is related to the number of bits transferred. Also, with each extra packet transferred there are additional requirements such as buffering and routing in the network [15]. The content of the packet is simply an identifying number with padding to make the payload up to 1000 bytes long which should not be fragmented by the network as it is smaller than the Ethernet maximum transmission unit (MTU).…”

Section: Planetlab Setupmentioning

confidence: 99%

Effect of path diversity on the loss performance of UDP packets over the internet

Haywood

Peng

2008

2008 5th International Conference on Broadband Communications, Networks and Systems

View full text Add to dashboard Cite

Abstract-In this paper we investigate the effect of path diversity on the packet loss performance through simulations and experiments using three overlay networks created on top of the PlanetLab testbed. The overlay network consists of a number of relay nodes that are positioned around the Internet. In each of the overlay networks UDP packets are streamed from a client to a destination for a period of 50 days, across up to six paths with a static packet dispatching algorithm. From the results obtained, we can see that increasing the number of paths could increase as well as decrease the packet loss probability, depending on the network environment and the structure of the multiple paths created. We have also investigated the correlation in the underlying network and demonstrated the performance of the metric defined.

show abstract

Modularity and efficiency in protocol implementation

Cited by 58 publications

References 1 publication

Profiling and reducing processing overheads in TCP/IP

Profiling and reducing processing overheads in TCP/IP

On packet loss performance under varying network conditions with path diversity

Effect of path diversity on the loss performance of UDP packets over the internet

Contact Info

Product

Resources

About