ACM Transactions on Graphics (TOG)
Congestion avoidance and control
SIGCOMM '88 Symposium proceedings on Communications architectures and protocols
Network performance effects of HTTP/1.1, CSS1, and PNG
SIGCOMM '97 Proceedings of the ACM SIGCOMM '97 conference on Applications, technologies, architectures, and protocols for computer communication
Performance interactions between P-HTTP and TCP implementations
ACM SIGCOMM Computer Communication Review
Proceedings of the ACM SIGCOMM '98 conference on Applications, technologies, architectures, and protocols for computer communication
Emergent (mis)behavior vs. complex software systems
Proceedings of the 1st ACM SIGOPS/EuroSys European Conference on Computer Systems 2006
The MooDS protocol: a J2ME object-oriented communication protocol
Mobility '07 Proceedings of the 4th international conference on mobile technology, applications, and systems and the 1st international symposium on Computer human interaction in mobile technology
Hop, a Fast Server for the Diffuse Web
COORDINATION '09 Proceedings of the 11th International Conference on Coordination Models and Languages
An efficient implementation of GPU virtualization in high performance clusters
Euro-Par'09 Proceedings of the 2009 international conference on Parallel processing
A transport-layer approach for improving thin-client performance in a WAN environment
International Journal of Internet Protocol Technology
| Hi-index | 0.00 |
Modern TCP implementations include a mechanism, known as the Nagle algorithm, which prevents the unnecessary transmission of a large number of small packets. This algorithm has proved useful in protecting the Internet against excessive packet loads. However, many applications suffer performance problems as a result of the traditional implementation of the Nagle algorithm. An interaction between the Nagle algorithm and TCP's delayed acknowledgement policy can create an especially severe problems, through a temporary “deadlock.” These flaws in the Nagle algorithm have prompted many application implementors to disable it, even in cases where this is neither necessary nor wise. We categorize the applications that should and should not disable the Nagle algorithm, and we show that for some applications that often disable the Nagle algorithm, equivalent performance can be obtained through an improved implementation of the algorithm. We describe five possible modifications, including one novel proposal, and analyze their performance on benchmark tests. We also describe a receiver-side modification that can help in some circumstances.