A Portable Client/Server Communication Middleware over SANs: Design and Performance Evaluation with Virtual Interface and InfiniBand
نویسندگان
چکیده
In this paper, we address the problem of using high performance SAN technologies to provide efficient communication support in client/server type of environments. Our objective is to build a portable substrate which not only bridges the functionality gap between applications and SAN networks and preserves the performance of the underlying communication layer, but also provides the required flexibility for analyzing the impact of various modern communication techniques. We propose an architecture which exposes a simple queue based interface to applications. We take advantage of the features of modern SANs (such as OS bypass, zero copy, and RDMA) and use several techniques to improve the performance of our communication protocol. We have implemented our design for both InfiniBand and Virtual Interface Architecture (VIA). Our performance evaluation shows that the protocol processing overhead is low and the underlying communication performance can be preserved. We also discuss the performance impact of various communication techniques.
منابع مشابه
High Performance Sockets and RPC over Virtual Interface (VI) Architecture
Standard user-level networking architecture such as Virtual Interface (VI) Architecture enables distributed applications to perform low overhead communication over System Area Networks (SANs). This paper describes how high-level communication paradigms like stream sockets and remote procedure call (RPC) can be e ciently built over user-level networking architectures. To evaluate performance ben...
متن کاملDesign and Performance Evaluation of LiMIC (Linux Kernel Module for MPI Intra-node Communication) on InfiniBand Cluster
High performance intra-node communication support for MPI applications is critical for achieving the best performance out of clusters of SMP workstations. Although the performance of system area networks has improved in the recent years, intra-node communication still remains orders of magnitude faster than the network. Present day MPI stacks cannot make use of operating system kernel support f...
متن کاملA Comparison of Mobile Agent and Client-Server Paradigms for Information Retrieval Tasks in Virtual Enterprises
In next-generation enterprises it will become increasingly important to retrieve information efficiently and rapidly from widely dispersed sites in a virtual enterprise, and the number of users who wish to do using wireless and portable devices will increase significantly. This paper considers the use of mobile agent technology rather than traditional clientserver computing for information retr...
متن کاملBenchmarking communication middleware for cloud computing virtualizers
Virtualization technologies typically introduce additional overhead that is specially challenging for specific domains such as real-time systems. One of the sources of overhead are the additional software layers that provide parallel execution environments which reduce the effective performance given by the infrastructure. This work identifies the factors to be analysed by a benchmark for perfo...
متن کاملA Portable InfiniBand Module for MPICH2/Nemesis: Design and Evaluation
With the emergence of multi-core-based processors, it is becoming significantly important to optimize both intra-node and inter-node communication in an MPI stack. MPICH2 group has recently introduced a new Nemesis-based MPI stack which provides highly optimized design for intra-node communication. It also provides modular design for different inter-node networks. Currently, the MPICH2/Nemesis ...
متن کامل