Analysis of TCP Performance in Data Center Networks by Santosh Kulkarni, Prathima Agrawal

By Santosh Kulkarni, Prathima Agrawal

This ebook addresses the necessity to increase TCP’s functionality inside of info facilities by way of supplying ideas which are either functional and backward appropriate with common TCP models. The authors method this problem first by means of deriving an analytical version for TCP’s functionality below normal info heart workload site visitors. They then talk about a few strategies which are designed to enhance TCP functionality by means of both proactively detecting community congestion via probabilistic retransmission or through keeping off timeout penalty via dynamic resizing of TCP segments. Experimental effects express that every of recommendations mentioned outperforms common TCP within a knowledge middle.

Show description

Read Online or Download Analysis of TCP Performance in Data Center Networks (SpringerBriefs in Electrical and Computer Engineering) PDF

Best networking books

Vehicular Networks: From Theory to Practice (Chapman & Hall/CRC Computer & Information Science Series)

In spite in their value and power societal impression, there's presently no complete resource of data approximately vehicular advert hoc networks (VANETs). Cohesively integrating the state-of-the-art during this rising box, Vehicular Networks: From idea to perform elucidates many matters interested in vehicular networking, together with site visitors engineering, human elements reviews, and novel desktop technological know-how research.

Divided into six extensive sections, the booklet starts off with an outline of site visitors engineering matters, reminiscent of site visitors tracking and site visitors movement modeling. It then introduces governmental and commercial efforts within the usa and Europe to set criteria and practice box exams at the feasibility of vehicular networks. After highlighting leading edge purposes enabled by way of vehicular networks, the ebook discusses numerous networking-related matters, together with routing and localization. the next part makes a speciality of simulation, that's at the moment the first process for comparing vehicular networking platforms. the ultimate half explores the level and effect of driving force distraction with in-vehicle displays.

Encompassing either introductory and complex techniques, this consultant covers many of the parts that impression the layout of purposes for vehicular networks. It information key examine demanding situations, deals suggestions on constructing destiny criteria, and provides useful info on latest experimental studies.

Additional info for Analysis of TCP Performance in Data Center Networks (SpringerBriefs in Electrical and Computer Engineering)

Example text

Online]. txt 16. J. Schlyter, W. Griffin, Using DNS to securely publish secure shell (SSH) key fingerprints, RFC 4255 (Proposed standard), Internet engineering task force, Jan 2006. [Online]. txt 17. F. Cusack, M. Forssen, Generic message exchange authentication for the secure shell protocol (SSH), RFC 4256 (Proposed standard), Internet engineering task force, Jan 2006. [Online]. txt 18. J. Kristoff, The transmission control protocol. [Online]. html 19. R. Braden, Requirements for internet hosts-communication layers, RFC 1122 (Standard), Internet engineering task force, updated by RFCs 1349, 4379, 5884, 6093, 6298, 6633.

When the third duplicate ACK is received, following the reset of ssthresh, TCP sets its cwnd to (ssthr esh + 3 × M SS) ensuring that the cwnd is artificially inflated by the number of segments that are outstanding in the network. • For each additional duplicate ACK that the sender receives, TCP increments its cwnd by one MSS. • When finally the sender receives an ACK that acknowledges previously unacknowledged data, TCP sets cwnd to ssthresh. This sequence is also known as “deflating” of the congestion window (cwnd).

U k are ACKed, another k packets, s1 , . . , sk are sent in the next round, which we will refer to as the “loss” round. This round of packets may have another loss, say packet sm+1 . Again, our assumptions on packet loss correlation mandates that packets sm+2 , . . , sk are also lost in the loss round. The m packets successfully sent in the loss round are responded to by ACKs for packet u k , which are counted as duplicate ACKs. If the number of such ACKs is higher than three, then a TD indication occurs, otherwise an IBTT occurs.

Download PDF sample

Rated 4.52 of 5 – based on 14 votes

About the Author