Distributed packet rewriting and its application to scalable server architectures

Bestavros, Azer; Crovella, Mark; Líu, Jùn; Martin, Denis

doi:10.1109/icnp.1998.723750

Cited by 53 publications

(23 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The IMX scenario is different from web and call server scenarios, since IMX nodes will always stay in the path for processing or policing tasks. A problem closer to the IMX scenario is presented in [8], where web servers forward new connections to other servers by rewriting addresses directly in the network stack. However, the resource allocation mechanism statically forwards a fix rate of requests, which is inappropriate for IMX where relocation is more costly.…”

Section: B Differences To Server Clustersmentioning

confidence: 99%

“…Even in the ideal case, there are free resources at IMX2 when IMX1 blocks, which is an intended property that avoids overloading remote nodes by relocating calls to them up to their limit. From Figure 5, we can derive the following relations: Full utilization in IMX1 (7), utilization in IMX2 (8), and (9) from comparing occupation of IMX1 and IMX2.…”

Section: B Considerations For the Ideal And Static Casementioning

confidence: 99%

See 1 more Smart Citation

Load Sharing in a Distributed IMS Architecture

Kögel

Wahl

Scharf

et al. 2009

VTC Spring 2009 - IEEE 69th Vehicular Technology Conference

View full text Add to dashboard Cite

Abstract-The IP Multimedia Subsystem (IMS) serves as universal platform for fast and standardized creation of mobile services. Typical deployment models for the IMS favor centralized session control and application servers. Furthermore, they rely on sophisticated border elements for, e.g., policy enforcement. One step towards a simpler architecture is the integration of call control, application server, and media functions into border elements. This leads to a distributed IMS architecture with equal processing nodes, which drastically reduces system complexity and scales on functional module instantiation basis.However, the distributed nature of the architecture reduces the statistical gain. In this work, we study a suitable load sharing concept that counters or even over-compensates this problem. In principle, this mechanism realizes one large virtual central server with small control and message overhead. Our simulative evaluation shows the feasibility and performance of our approach.

show abstract

Section: B Differences To Server Clustersmentioning

confidence: 99%

Section: B Considerations For the Ideal And Static Casementioning

confidence: 99%

Load Sharing in a Distributed IMS Architecture

Kögel

Wahl

Scharf

et al. 2009

VTC Spring 2009 - IEEE 69th Vehicular Technology Conference

View full text Add to dashboard Cite

show abstract

“…From Table 1, we can immediately notice that hashing does an extremely good job in scattering flows. In fact, a load balancing system relying mainly on flow distribution, e.g., web servers [5] , works very well with static hashing. But hashing is not so effective in distributing packets as shown in Table 2 and is even worse for distributing bytes as shown in Table 3 .…”

Section: 2mentioning

confidence: 99%

<title>Internet traffic load balancing using dynamic hashing with flow volume</title>

Kim

Chao³

et al. 2002

SPIE Proceedings

View full text Add to dashboard Cite

Sending IP packets over multiple parallel links is in extensive use in today's Internet and its use is growing due to its scalability, reliability and cost-effectiveness. To maximize the efficiency of parallel links, load balancing is necessary among the links, but it may cause the problem of packet reordering. Since packet reordering impairs TCP performance, it is important to reduce the amount of reordering. Hashing offers a simple solution to keep the packet order by sending a flow over a unique link, but static hashing does not guarantee an even distribution of the traffic amount among the links, which could lead to packet loss under heavy load. Dynamic hashing offers some degree of load balancing but suffers from load fluctuations and excessive packet reordering. To overcome these shortcomings, we have enhanced the dynamic hashing algorithm to utilize the flow volume information in order to reassign only the appropriate flows. This new method, called dynamic hashing with flow volume (DHFV), eliminates unnecessary flow reassignments of small flows and achieves load balancing very quickly without load fluctuation by accurately predicting the amount of transferred load between the links. In this paper we provide the general framework of DHFV and address the challenges in implementing DHFV. We then introduce two algorithms of DHFV with different flow selection strategies and show their performances through simulation.

show abstract

“…Distributed packet rewriting is similar to packet rewriting (Bestavros, Crovella, Liu, and Martin, 1998). When a packet reaches a server, the server uses a hash function to calculate the packet's destination by the use of the client's IP address and port number to determine to which server the packet should be redirected.…”

Section: Request Redirection Mechanismsmentioning

confidence: 99%

Load Balancing on the Internet

Jiang

Zhou

2004

The Internet Encyclopedia

View full text Add to dashboard Cite

This chapter gives a survey of state‐of‐the‐art load‐balancing strategies on the Internet. Internet services must be scalable, available, and responsive to handle large traffic volumes and support up to millions of concurrent clients. This scalability requirement goes beyond the capability of any single machine servers. Cluster server architecture is gaining momentum in support of scalable Internet services. The servers in a cluster can be either distributed locally at the same site or globally distributed in different geographical locations. Load balancing is a key integration technology at both layers for performance. This chapter presents a characterization of Web services and gives a classification of load balancing strategies according to their deployment locations (server, network, and client sides), information taken into account (state‐blind, client‐state aware, and server‐state aware), and implementation levels (network and application layers). Representative load balancing strategies in each category are reviewed.

show abstract

Distributed packet rewriting and its application to scalable server architectures

Abstract: Abstract

Cited by 53 publications

References 11 publications

Load Sharing in a Distributed IMS Architecture

Load Sharing in a Distributed IMS Architecture

<title>Internet traffic load balancing using dynamic hashing with flow volume</title>

Load Balancing on the Internet

Contact Info

Product

Resources

About