Accurate Load Monitoring for Cluster-based Web Data-Centers over RDMA-enabled Networks

Monitoring a pool of resources in a cluster-based web datacenter environment can be critical for successful deployment of applications such as web servers, database servers, etc. In particular, the monitored information assists system-level services like load balancing in enabling the data-center environment to efficiently adapt to the changing system load and traffic pattern. This information is not only critical in terms of accuracy and content, but it must also be gathered without impacting performance or affecting other applications. In this paper, we propose two accurate load monitoring schemes, namely, user-level load monitoring (ULM) and kernel-level load monitoring (KLM) in a web datacenter environment and evaluate its benefits with respect to overall system load balancing. In our approach, we use the Remote Direct Memory Access (RDMA) operation (in user space or kernel space) provided by RDMA-enabled interconnects like InfiniBand. We further leverage the information provided by certain kernel data structures in designing these schemes without requiring any modifications to the existing data-center applications. Our experimental results show that the KLM and ULM schemes achieve an improvement of 22% and 12% in a single data-center and an improvement of 25% and 11% per web-site in shared data-centers, respectively. More importantly, our schemes take advantage of RDMA operations in accessing portions of kernel memory that is not exposed to user space for accurate load monitoring. Further, our design is resilient and well-conditioned to the load on the servers as compared to two-sided communication protocols such

[1]  Edward D. Lazowska,et al.  Adaptive load sharing in homogeneous distributed systems , 1986, IEEE Transactions on Software Engineering.

[2]  Songnian Zhou.  An Experimental Assessment of Resource Queue Lengths as Load Indices , 1986 .

[3]  S. Zhou,et al.  A Trace-Driven Simulation Study of Dynamic Load Balancing , 1987, IEEE Trans. Software Eng..

[4]  Ravishankar K. Iyer,et al.  Prediction-Based Dynamic Load-Sharing Heuristics , 1993, IEEE Trans. Parallel Distributed Syst..

[5]  Erich M. Nahum,et al.  Locality-aware request distribution in cluster-based network servers , 1998, ASPLOS VIII.

[6]  Guerney D. H. Hunt,et al.  Network Dispatcher: A Connection Router for Scalable Internet Services , 1998, Comput. Networks.

[7]  Michael Mitzenmacher,et al.  On the Analysis of Randomized Load Balancing Schemes , 1997, SPAA '97.

[8]  Nina Bhatti,et al.  Web server support for tiered services , 1999, IEEE Netw..

[9]  Willy Zwaenepoel,et al.  Scalable Content-aware Request Distribution in Cluster-based Network Servers , 2000, USENIX Annual Technical Conference, General Track.

[10]  Ludmila Cherkasova,et al.  Optimizing a "content-aware" load balancing strategy for shared Web hosting service , 2000, Proceedings 8th International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (Cat. No.PR00728).

[11]  Sang Hyuk Son,et al.  A feedback control approach for guaranteeing relative delays in Web servers , 2001, Proceedings Seventh IEEE Real-Time Technology and Applications Symposium.

[12]  Ricardo Bianchini,et al.  Efficiency vs. portability in cluster-based network servers , 2001, PPoPP '01.

[13]  David E. Culler,et al.  SEDA: an architecture for well-conditioned, scalable internet services , 2001, SOSP.

[14]  Raphael Rom,et al.  Application-aware admission control and scheduling in Web servers , 2002, Proceedings.Twenty-First Annual Joint Conference of the IEEE Computer and Communications Societies.

[15]  David K. Y. Yau,et al.  Admission control and dynamic adaptation for a proportional-delay diffserv-enabled web server , 2002, SIGMETRICS '02.

[16]  Weisong Shi,et al.  Modeling object characteristics of dynamic Web content , 2002, Global Telecommunications Conference, 2002. GLOBECOM '02. IEEE.

[17]  Tao Yang,et al.  Cluster load balancing for fine-grain network services , 2002, Proceedings 16th International Parallel and Distributed Processing Symposium.

[18]  Weisong Shi,et al.  Modeling object characteristics of dynamic Web content , 2003, J. Parallel Distributed Comput..

[19]  Prashant J. Shenoy,et al.  Dynamic resource allocation for shared data centers using online measurements , 2003, IWQoS'03.

[20]  Hyun-Wook Jin,et al.  Exploiting Remote Memory Operations to Design Efficient Reconfiguration for Shared Data-Centers over InfiniBand , 2004 .

[21]  D. Panda,et al.  Supporting Strong Coherency for Active Caches in Multi-Tier Data-Centers over InfiniBand , 2004 .

[22]  Hyun-Wook Jin,et al.  On the provision of prioritization and soft qos in dynamically reconfigurable shared data-centers over infiniband , 2005, IEEE International Symposium on Performance Analysis of Systems and Software, 2005. ISPASS 2005..

[23]  An observation-based approach towards self-managing web servers , 2006, Comput. Commun..