In exascale systems, increasing error rate -- particularly silent data corruption -- is a major concern. The Global ViewResilience (GVR) system builds a new model of application resilience on versioned global arrays. These arrays can be exploited for flexible, application-specific error checking and recovery. We explore a fundamental challenge to the GVR model -- the cost of versioning. We propose a novel log-structured implementation that appends new data to an update log, simultaneously tracking modified regions and versioning incrementally. We compare performance of log-structured arrays to traditional flat arrays using micro-benchmarks and three full applications, and show that versioning can be more than 10x faster, and reduce memory cost significantly. Further, in future systems with NVRAM, a log-structured approach is more tolerant onramp limitations such as write bandwidth and wear-out.

We present the Global View Resilience (GVR) system, a library that enables applications to add resilience in a portable, application-controlled fashion using versioned distributed arrays. We briefly describe GVR's interfaces for distributed arrays, versioning, and cross-layer error recovery. We illustrate how GVR can be used for rollback recovery and a wide range additional error recovery techniques including forward recovery for latent errors or silent data corruptions. Application results demonstrate that GVR's interfaces and implementation are portable, flexible (support a variety of recovery models), efficient and create a gentle-slope path to tolerate growing error rates in future systems.

Profiling is an effective dynamic analysis approach to investigate complex applications. ParaTrac is a user-level profiler using file system and process tracing techniques for data-intensive workflow applications. In two respects ParaTrac helps users refine the orchestration of workflows. First, the profiles of I/O characteristics enable users to quickly identify bottlenecks of underlying I/O subsystems. Second, ParaTrac can exploit fine-grained data-processes interactions in workflow execution to help users understand, characterize, and manage realistic data-intensive workflows. Experiments on thoroughly profiling Montage workflow demonstrate that ParaTrac is scalable to tracing events of thousands of processes and effective in guiding fine-grained workflow scheduling or workflow management systems improvements.

This paper describes the rational behind designing workflow systems based on the Unix make by showing a number of idioms useful for workflows comprising many tasks. It also demonstrates a specific design and implementation of such a workflow system called GXP make. GXP make supports all the features of GNU make and extends its platforms from single node systems to clusters, clouds, supercomputers, and distributed systems. Interestingly, it is achieved by a very small code base that does not modify GNU make implementation at all. While being not ideal for performance, it achieved a useful performance and scalability of dispatching one million tasks in approximately 16,000 seconds (60 tasks per second, including dependence analysis) on an 8 core Intel Nehalem node. For real applications, recognition and classification of protein-protein interactions from biomedical texts on a supercomputer with more than 8,000 cores are described.

The realistic characteristics of data-intensive workflows are critical to optimal workflow orchestration and profiling is an effective approach to investigate the behaviors of such complex applications. ParaTrac is a fine-grained profiler for data-intensive workflows by using user-level file system and process tracing techniques. First, ParaTrac enables users to quickly understand the I/O characteristics of from entire application to specific processes or files by examining low-level I/O profiles. Second, ParaTrac automatically exploits fine-grained data-processes interactions in workflow to help users intuitively and quantitatively investigate realistic execution of data-intensive workflows. Experiments on thoroughly profiling Montage workflow demonstrate both the scalability and effectiveness of ParaTrac. The overhead of tracing thousands of processes is around 16%. We use low-level I/O profiles and informative workflow DAGs to illustrate the vantage of fine-grained profiling by helping users comprehensively understand the application behaviors and refine the scheduling for complex workflows. Our study also suggests that current workflow management systems may use fine-grained profiles to provide more flexible control for optimal workflow execution.

Developing and deploying distributed file system has been important for the Grid computing. By GMount, non-privileged users can instantaneously and effortlessly build a distributed file system on arbitrary machines that are reachable via SSH. It is scalable to hundreds of nodes in the wide-area Grid environments and adapts to NAT/Firewall. Unlike conventional distributed file systems, GMount can directly harness local file systems of each node without importing/exporting application data and utilize the network topology to make the metadata operations locality-aware. In this paper, we present the design and implementation of GMount by using two popular modules: SSH and FUSE. We demonstrate its viability and locality-aware metadata operation performance in a large scale Grid with over 320 nodes spreading across 12 clusters that are connected by heterogeneous wide-area links.