Distributed data center load balancing method based on fault tolerance of side

The invention discloses a distributed data center load balancing method based on the fault tolerance of a side. A path congestion table with a self-aging mechanism is maintained at a source host side. The distributed data center load balancing method comprises the steps of path state evaluation base...

Full description

Saved in:
Bibliographic Details
Main Authors LU SANGLU, WEN KAIYUAN, QIAN ZHUZHONG
Format Patent
LanguageEnglish
Published 08.07.2015
Subjects
Online AccessGet full text

Cover

Loading…
More Information
Summary:The invention discloses a distributed data center load balancing method based on the fault tolerance of a side. A path congestion table with a self-aging mechanism is maintained at a source host side. The distributed data center load balancing method comprises the steps of path state evaluation based on the side and self-adapting load balancing, wherein according to the path evaluation based on the side, the congestion degree of a current path is evaluated by using information at least comprising ACK arrival time and the confirmed ACK data size in a TCP data stream, information of a fault path is obtained in time through fast retransmission and RTO overtime events in a TCP, and the obtained path congestion degree and the fault information are inserted into the congestion table of a host; according to the self-adapting load balancing, a congestion window in the TCP stream is taken as a basic load unit, and a path state in the congestion table of the host is used for carrying out a bidirectional fine grit load balancing mechanism. According to the distributed data center load balancing method, special hardware is not needed for support, deployment can be carried out quickly and effectively in a large scale data center, low expenditures and easy deployment are achieved, and the communication performance in the data center is effectively improved.
Bibliography:Application Number: CN201510198276