| Applying Internet into embedded systems, monitoring embedded equipments through Internet, these are the developing trends of combination of Internet and embedded systems. Embedded web server (EWS), inexpensive, convenient, expandable, has become the main method of connecting embedded equipments to Internet. With the gradually wide use of EWS, how to improve its processing capacity has become more and more important, and this will directly affect the management efficiency of remote equipments.The paper first introduced the load balancing techniques of traditional web server. Discussed the factors and strategies of implementing load balancing, and presented several basic load balancing algorithms and implementation techniques. Then the paper analyzed the critical issues of implementing EWS. Presented the software and hardware architectures of current EWS according to its application envirenment. Concluded that the limitation of computing and memory resources of EWS has . become the bottleneck of its developments.Under the discussion above, to improve the performance of EWS, the paper analyzed the distributing rules of requests arriving according to the queuing theory and Markov process and optimized the parameters of M/M/1 queuing system model according to the EWS working environment, such as length of queue, time of customer waiting, time of system busy and idle. Based on these, the paper brought forward the load balancing technique of EWS cluster, which adopted several embedded web servers to serve one application as a whole. We divided requests visiting EWS into two kinds, requests monitoring embedded equipments and requests browsing states documents. According to the characteristics of EWS, we brought forward the least connection load-balancing algorithm based on the priority of requests, which distinguishes the priorities of requests (the first kind is prior to the second) and transmits the requests with higher priority first to the EWS with least connections. |
PDF Full Text Request