An inter-domain load balancing mechanism and performance evaluation | IEEE Conference Publication | IEEE Xplore

An inter-domain load balancing mechanism and performance evaluation


Abstract:

Load balancing technique may improve the performance and scalability of the Internet greatly. Many researchers focus on intra-domain load balancing which distributes traf...Show More

Abstract:

Load balancing technique may improve the performance and scalability of the Internet greatly. Many researchers focus on intra-domain load balancing which distributes traffic over multiple paths or server farms in a single domain. However, there is little research on interdomain (such as inter-AS or inter-PoP) load balancing due to the loose relationship and complex policies among domains. It is important to study inter-domain load balancing under the trend toward a denser mesh of inter-domain connectivity within the Internet. In this paper, we propose a distributed and dynamic load balancing mechanism using a hashing-based algorithm to distribute traffic between domains which preserves per-flow packet ordering and supports unequal weighted distribution when dynamic adaptation is used. We also evaluate the performance of the mechanism.
Date of Conference: 27-30 May 2005
Date Added to IEEE Xplore: 15 August 2005
Print ISBN:0-7803-9015-6
Conference Location: Hong Kong, China

I. INTRODUCTION

With the dramatic increase in Internet traffic and users. the original simple backbone may become bottleneck. In order to provide scalability and high availability, redundant denser mesh connectivity and multiple parallel trunks between major Points of Presence (POPs) or Internet Service Providers(JSPs) are often engineered to ensure multiple paths. Load balancing is a key technique to distribute the-traffic all over these multiple paths, which is in best utilizing the network resources and improving the network performance effectively. There are lots of load balancing mechanisms from data link layer to application layer, Inverse multiplexing is designed for use over point-to-point links and load balancing is based on round robin distribution of packets or bytes[1] [2] [3]. The paths for Internet load balancing make use of the natural redundancy in the network topology. Equal-cost or unequal-cost multi-paths within an autonomous system (AS) are discovered dynamically by routing protocols such as OSPF[4], IS-IS[5], [GRP[6], The Border Gateway Protocol (BGP)[7] is currently the only inter-domain routing protocol employed on the Internet. As required of any inter-domain protocol, BGP allows policy-based metrics to override distance-based metrics and enables each AS to independently define its routing policies with little or no global coordination. It chooses the best route to the destination network and becomes non-trivial to figure out how to manually direct specific portions of internal traffic(prefixes)in a distributed fashion across multiple T external gateways. So, Standard BGP does not load balancing traffic. It is difficult for BGP to implementing the additional load balancing mechanisms based on real-time traffic and dynamic resource status by complicated configuration. [8]describes an extensibJe mechanism that will allow a BGP speaker to advertise equal-cost multi-path routes for a destination to its peers by introducing a new BGP attribute without changing the semantics of the UPDATE message. [9]The Link Bandwidth feature is used to advertise the bandwidth of an AS exit link as an extended community to BGP. The BGP Link Bandwidth feature supports both internal BGP (iBGP), external BGP (eBGP) multi-path load balancing, and multi-path load balancing in Multiprotocol Label Switching (MPLS) Virtual Private Networks (VPNs). The link bandwidth extended community indicates the preference of an AS exit link in terms of bandwidth. But BGP can originate the link bandwidth community only for eBGP peers that are directly connected. This paper discusses the load balancing among great-granularity and far-ranging networks, such as ASes or POPs.

Contact IEEE to Subscribe

References

References is not available for this document.