Fully pre-splicing (FPS) is an extension of TCP splicing which is content-blind that prevents the switches from using application layer information for forwarding decisions. FPS extends TCP splicing to support content...
详细信息
ISBN:
(纸本)0769526160
Fully pre-splicing (FPS) is an extension of TCP splicing which is content-blind that prevents the switches from using application layer information for forwarding decisions. FPS extends TCP splicing to support content-aware load balancing algorithms and pre-splices the client's connections to webservers. In addition, FPS extracts the application-information in kernel-space for eliminating the cost of moving data twice through user/kernel protection boundaries and the latency of scheduling the processes. To achieve our design, we extended the TCPSP project of LVS using Linux Netfilter which defines specific hooks to provide a verdict for the packet. On the performance results, FPS improves the TCPSP throughput dramatically.
We study approximate algorithms for placing a set of documents into M distributed web servers in this paper. We define the load of a server to be the summation of loads induced by all documents stored. The size of a s...
详细信息
We study approximate algorithms for placing a set of documents into M distributed web servers in this paper. We define the load of a server to be the summation of loads induced by all documents stored. The size of a server is defined in a similar manner. We propose five algorithms. Algorithm 1 balances the loads and sizes of the servers by limiting the loads to k(l) and the sizes to k(s) times their optimal values, where 1/k(l)-1 + 1/k(s)-1 <= 1. This result improves the bounds on load and size of servers in [ 10]. Algorithm 2 further reduces the load bound on each server by using partial document replication, and Algorithm 3 by sorting. Algorithm 4 employs both partial replication and sorting. Last, without using sorting and replication, we give Algorithm 5 for the dynamic placement at the cost of a factor O(log M) in the time-complexity.
＜正＞While content-aware distribution policies getting more popular in cluster-based web systems,they make the dispatching node a *** achieve scalable server performance,we present a completely distributed architectu...
详细信息
＜正＞While content-aware distribution policies getting more popular in cluster-based web systems,they make the dispatching node a *** achieve scalable server performance,we present a completely distributed architecture named *** CODA,we employ both layer-4 and layer-7 web switching *** the front-end layer-4 switch distributes the client requests among the server nodes using some simple dispatching algorithm,the server nodes may forward the incoming request to another server node according to some content-aware *** eliminating the centralized dispatcher,CODA affords better scalability and fault ***,we have implemented this architecture on the basis of Linux IP stack and evaluated its performance with trace-based *** performance results show it can achieve favorable throughput compared to existing architectures.
While content-aware distribution policies getting more popular in cluster-based web systems,they make the dispatching node a *** achieve scalable server performance,we present a completely distributed architecture in ...
详细信息
While content-aware distribution policies getting more popular in cluster-based web systems,they make the dispatching node a *** achieve scalable server performance,we present a completely distributed architecture in which all hosts participate in request *** cope with the architecture,we propose a distributed dispatching policy named *** DWARD,all the server nodes may participate in request dispatching on the basis of local access *** addition, the DWARD policy inherits several desirable features of the centralized algorithms,including load balance,high locality and less forwarding ***,a testbed is implemented on the basis of Linux kernel to benchmark various dispatching *** performance results show that DWARD can achieve favorable throughput compared with state-of-the-art dispatching policies.
暂无评论