viro - VIRO: Scalable, Robust and Naming-proof Plug &...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: VIRO: Scalable, Robust and Naming-proof Plug & Play Routing Sourabh Jain, Saurabh Jain, Yingying Chen, Zhi-Li Zhang Department of Computer Science, University of Minnesota - Twin Cities { sourj, saurabh, yingying, zhzhang } Abstract —In recent years, researchers have argued for the clean-slate design of the Internet to meet the growing require- ments. One of the most promising proposal for this clean-slate design was to design the networks using the “Flat-Id” based routing architecture. These designs can solve the problems related to mobility and scalability of the Internet, but suffers from the poor efficiency in terms of routing stretch and network delay. In this work, we propose a routing scheme which work on the top of “Flat-ID” based network architecture and promises much better support for the mobility and scalability. Furthermore, unlike other similar proposals it does not suffer from the poor routing efficiency. It is achieved by the insertion of a “Virtual-id layer” which embeds the network topology, and uses it to reduce the routing stretch. We evaluated our routing protocol on various topologies including backbone topologies, line, star topology, and also on recently presented network architectures for data centers such as fat-tree and Dcell. We found that routing stretch using VIRO for all these topologies is very close to 1. Index Terms —Routing, Mobility, Networks I. INTRODUCTION Current Internet has proven to be quite successful during the last decade. This can be partly attributed to its ’hourglass’ design in which the use of IP as the glue allows various protocols to run on top of the IP over many different net- works. However, using IP addressing scheme can create a lot problems. First, attaching ’identifier’ to location makes multi-homing and mobility services difficult. Second, we have to assign IP addresses very carefully in order to keep its global uniqueness and match the physical network topology. Moreover, the large number of IP prefixes can overwhelm the capacity of routers. In particular, the number of routing entries in the current ’core routers’ is around 120k+ and is still increasing. Most commonly used intra-domain routing protocols such as OSPF, IS-IS are all based on IP addressing, making it quite restrictive in terms of routing efficiency. Routing effi- ciency is defined to be the path length computed using the routing protocol over the best possible path length, which is often used to measure the performance of a routing protocol. Moreover, these IP-based routing protocols can hardly meet the increasing need of address prefixes. That’s simply because each IP prefix requires exactly one routing entry, resulting in an extremely large routing table hard to maintain. Additionally, current routing protocols could have scalability and manage- ability issues. For example, OSPF uses flooding in exchanging link state information to ensure consistent views of the network topology among all routers. As such, local failure can havetopology among all routers....
View Full Document

This note was uploaded on 10/09/2009 for the course CS 5211 taught by Professor Xyz during the Fall '09 term at Minnesota.

Page1 / 11

viro - VIRO: Scalable, Robust and Naming-proof Plug &...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online