[xcp] New XCP expr results

Yongguang Zhang ygz at hrl.com
Sat Aug 7 02:15:05 PDT 2004


Dear Colleagues,

We have done some experimental study on XCP in a real test bed.
The manuscript explaining the results is placed (temporarily) at
http://www.yongguangzhang.net/research/xcp.html.
The abstract is also included below.  We solicit all your comments
on this work.  Thanks!

Aaron: do you still have regular XCP meetings down in ISI?
Maybe I can come down sometime and talk about this.

Regards,

Yongguang


====
An Implementation and Experimental Study of the eXplicit Control Protocol (XCP)

Yongguang Zhang and Tom Henderson

The eXplicit Control Protocol (XCP) has been proposed as a
multi-level network feedback mechanism for congestion control of Internet
transport protocols. Theoretical and simulation results have suggested
that the protocol is stable and efcient over high bandwidth-delay product
paths, while being more scalable to deploy than mechanisms that do require
per-ow state in routers. However, there is little operational experience with
the approach. Since the deployment of XCP would require changes to both
the end hosts and routers, it is important to study the implications of this
new architecture before advocating such wide scale changes to internets.
This paper presents the results of an experimental study of XCP. We
first implemented XCP in the Linux kernel and solved various systems issues.
After validating previously reported simulation results, we studied
the sensitivity of XCP's performance to various environmental factors and
identied two sources that can signicantly reduce XCP's ability to control
congestions and achieve fairness. Our contributions are two fold. First,
through implementation we have revealed the challenges in platforms that
lack large native data types or floating point arithmetic, and the need to
keep fractions in XCP protocol header. Second, through experiments and
analysis we have identied several possibilities that XCP can enter into 
incorrect
feedback control loops and adversely affect the performance. These
are deployment challenges intrinsic to XCP design. More research is needed
before we can advocate a wide scale adoption.






More information about the xcp mailing list