Enhanced server fault-tolerance for improved user experience

Marwah, Manish; Mishra, Shivakant; Fetzer, Christof

doi:10.1109/dsn.2008.4630085

Cited by 9 publications

(4 citation statements)

References 10 publications

(18 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thus, we intercept messages between the master and the workers to know the input location and sample them for the evaluation runs. For retaining the D datasets for the server applications, we leverage the replay strategy that is commonly used in large-scale servers for faulttolerant computing [Marwah et al 2008;Aghdaie and Tamir 2002]-because of the high failure rate in servers, software infrastructures always embed a replay capability to retain the datasets and rerun the program if the previous execution failed.…”

Section: Implementing Iodcmentioning

confidence: 99%

Practical Iterative Optimization for the Data Center

Fang¹,

Xu²,

Chen

et al. 2015

ACM Trans. Archit. Code Optim.

View full text Add to dashboard Cite

Iterative optimization is a simple but powerful approach that searches the best possible combination of compiler optimizations for a given workload. However, iterative optimization is plagued by several practical issues that prevent it from being widely used in practice: a large number of runs are required to find the best combination, the optimum combination is dataset dependent, and the exploration process incurs significant overhead that needs to be compensated for by performance benefits. Therefore, although iterative optimization has been shown to have a significant performance potential, it seldom is used in production compilers.In this article, we propose iterative optimization for the data center (IODC): we show that the data center offers a context in which all of the preceding hurdles can be overcome. The basic idea is to spawn different combinations across workers and recollect performance statistics at the master, which then evolves to the optimum combination of compiler optimizations. IODC carefully manages costs and benefits, and it is transparent to the end user. To bring IODC to practice, we evaluate it in the presence of co-runners to better reflect real-life data center operation with multiple applications co-running per server. We enhance IODC Extension of Conference Paper: This article is an extension of a previously published conference paper at ASPLOS 2012 [Chen et al. 2012]. The article contains the following contributions over the prior paper:(1) We add experiments and analysis to understand whether iterative optimization and IODC are effective under the co-running scenario in Section 6. (2) We propose and implement two practical techniques in Section 7 to enhance IODC with the capability of compatible scheduling and dynamic threshold adjustment, so as to guarantee improved robustness, and handle the extreme cases in the presence of performance noise due to co-running applications.L. S. Fang et al.with the capability to find compatible co-runners along with a mechanism to dynamically adjust the level of aggressiveness to improve its robustness in the presence of co-running applications. We evaluate IODC using both MapReduce and compute-intensive throughput server applications. To reflect the large number of users interacting with the system, we gather a very large collection of datasets (up to hundreds of millions of unique datasets per program), for a total storage of 16.4TB and 850 days of CPU time. We report an average performance improvement of 1.48× and up to 2.08× for five MapReduce applications, and 1.12× and up to 1.39× for nine server applications. Furthermore, our experiments demonstrate that IODC is effective in the presence of co-runners, improving performance by greater than 13% compared to the worst possible co-runner schedule.

show abstract

Section: Implementing Iodcmentioning

confidence: 99%

Practical Iterative Optimization for the Data Center

Fang¹,

Xu²,

Chen

et al. 2015

ACM Trans. Archit. Code Optim.

View full text Add to dashboard Cite

show abstract

“…IODC for server applications. For retaining the D data sets in IODC for server applications, we leverage the replay strategy which is commonly used in large-scale servers for fault-tolerant computing [2,23]. Because the failure rate in servers is high, software infrastructures on servers embed a replay capability to rerun a data set if the previous run did not complete.…”

Section: Implementing Iodcmentioning

confidence: 99%

Iterative optimization for the data center

et al. 2012

View full text Add to dashboard Cite

Iterative optimization is a simple but powerful approach that searches for the best possible combination of compiler optimizations for a given workload. However, each program, if not each data set, potentially favors a different combination. As a result, iterative optimization is plagued by several practical issues that prevent it from being widely used in practice: a large number of runs are required for finding the best combination; the process can be data set dependent; and the exploration process incurs significant overhead that needs to be compensated for by performance benefits.Therefore, while iterative optimization has been shown to have significant performance potential, it is seldomly used in production compilers. In this paper, we propose Iterative Optimization for the Data Center (IODC): we show that servers and data centers offer a context in which all of the above hurdles can be overcome. The basic idea is to spawn different combinations across workers and recollect performance statistics at the master, which then evolves to the optimum combination of compiler optimizations. IODC carefully manages costs and benefits, and is transparent to the end user. We evaluate IODC using both MapReduce and throughput compute-intensive server applications. In order to reflect the large number of users interacting with the system, we gather a very large collection of data sets (at least 1000 and up to several million unique data sets per program), for a total storage of 10.7TB, and 568 days of CPU time. We report an average performance improvement of 1.48×, and up to 2.08×, for the MapReduce applications, and 1.14×, and up to 1.39×, for the throughput compute-intensive server applications.

show abstract

“…Other methods concentrate on providing fault-tolerance as transparent as possible to the service [1], [2], but they are generally developed for specific applications like Web servers, whose client communication protocol is known. In order to provide transparency to the clients, they require service's collaboration, so they impose special design requirements on the service.…”

Section: Introductionmentioning

confidence: 99%

Transparent Fault-Tolerance Based on Asynchronous Virtual Machine Replication

Coleșa

Stan

Ignat

2010

2010 12th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing

View full text Add to dashboard Cite

Fault-tolerance of services has received great attention for years, yet it is not clear how this can be done efficiently for different types of services. Current research focuses on finding strategies to provide fault-tolerance using commodity hardware and as independent of the service and transparent to clients as possible.We propose a replication system to provide fault-tolerance for any type of service by running the service instance into a virtual machine and replicate the entire virtual machine using an asynchronous replication strategy.The testing prototype was implemented using Xen hypervisor and an asynchronous replication strategy that gives better replication time than the synchronous one.

show abstract

Enhanced server fault-tolerance for improved user experience

Abstract: Interactive applications such as email, calendar, and

Cited by 9 publications

References 10 publications

Practical Iterative Optimization for the Data Center

Practical Iterative Optimization for the Data Center

Iterative optimization for the data center

Transparent Fault-Tolerance Based on Asynchronous Virtual Machine Replication

Contact Info

Product

Resources

About