Rights to individual papers remain with the author or the author's employer. Permission is granted for noncommercial reproduction of the work for educational or research purposes. This copyright notice must be included in the reproduced paper. USENIX acknowledges all trademarks herein.

This paper describes the design, implementation, and evaluation of TBBT, the first comprehensive NFS trace replay tool. Given an NFS trace, TBBT automatically detects and repairs missing operations in the trace, derives a file system image required to successfully replay the trace, ages the file system image appropriately, initializes the file server under test with that image, and finally drives the file server with a workload that is derived from replaying the trace according to user-specified parameters. TBBT can scale a trace temporally or spatially to meet the need of a simulation run without violating dependencies among file system operations in the trace.

Abstract not found

For typical workloads and file naming conventions, the size, lifespan, read/write ratio, and access pattern of nearly all files in a file system are accurately predicted by the name given to the file when it is created. We discuss some name-related properties observed in three contemporary NFS workloads, and present a method for automatically creating name-based models to predict interesting file properties of new files, and analyze the accuracy of these models for our workloads. Finally, we show how these predictions can be used as hints to optimize the strategies used by the file system to manage new files when they are created.

Rights to individual papers remain with the author or the author's employer. Permission is granted for noncommercial reproduction of the work for educational or research purposes. This copyright notice must be included in the reproduced paper. USENIX acknowledges all trademarks herein.

A common approach to benchmarking a server is to measure its behavior under load from a workload generator. Often a set of such experiments is required—perhaps with different server configurations or workload parameters— to obtain a statistically sound result for a given benchmarking objective. This paper explores a framework and policies to conduct such benchmarking activities automatically and efficiently. The workbench automation framework is designed to be independent of the underlying benchmark harness, including the server implementation, configuration tools, and workload generator. Rather, we take those mechanisms as given and focus on automation policies within the framework. As a motivating example we focus on rating the peak load of an NFS file server for a given set of workload parameters, a common and costly activity in the storage server industry. Experimental results show how an automated workbench controller can plan and coordinate the benchmark runs to obtain a result with a target threshold of confidence and accuracy at lower cost than scripted approaches that are commonly practiced. In more complex benchmarking scenarios, the controller can consider various factors including accuracy vs. cost tradeoffs, availability of hardware resources, deadlines, and the results of previous experiments.

Benchmarks are crucial to understanding software systems and assessing their performance. In file-system research, synthetic benchmarks are accepted and widely used as substitutes for more realistic and complex workloads. However, synthetic benchmarks are largely based on the benchmark writer’s interpretation of the real workload, and how it exercises the system API. This is insufficient since even a simple operation through the API may end up exercising the file system in very different ways due to effects of features such as caching and prefetching. In this paper, we describe our first steps in creating “realistic synthetic ” benchmarks by building a tool, CodeMRI. CodeMRI leverages file-system domain knowledge and a small amount of system profiling in order to better understand how the benchmark is stressing the system and to deconstruct its workload. 1

Access control in network file systems relies on primitive mechanisms like Access Control Lists and permission bits, which are not enough when operating in a hostile network environment. Network middleboxes, e.g., firewalls, completely ignore file system semantics when defining policies. Therefore, implementing simple context-aware access policies requires modifications to file servers and/or clients, which is impractical. We present FileWall, a network middlebox that allows administrators to define context-aware access policies for file systems using both the network context and the file system context. FileWall interposes on the client-server network path and implements administrator defined policies through message transformation without modifying either

File access traces have been used to drive simulations of storage management algorithms such as file caching, for workload characterization and modeling, and to identify interesting access patterns for performance optimization. Surprisingly they are rarely used to test the correctness and evaluate the performance of an actual file system or server. The main reason is that up until now there did not exist a flexible and easy-to-use player for file access traces. This paper describes the design, implementation, and evaluation of an NFS trace play-back tool called TBBT (Trace-Based file system Benchmarking Tool) that can automatically derive the file system hierarchy from an NFS trace, initialize the file system image with controllable aging effects, and speed up or slow down the trace play-back speed using temporal or spatial scaling without violating dependencies among trace entries. Experiments using a large NFS trace set show that TBBT can indeed produce different throughput and latency measurements than synthetic benchmarks such as SPECsfs. Moreover, TBBT's trace player is actually more efficient than SPECsfs's workload generator despite the fact that the former requires more CPU computation and disk I/O accesses.

Offloading to hardware components that support the primary task of a system enables sea-paration of concerns and allows both the primary and offloaded components of a system to be easy to understand, manage, and evolve independent of other components. In this dissertation, we explore the software mechanisms required to effectively offload functionality to idle processing elements. We present the design, implementation, and evalua-tion of three system architectures – TCPServers, Orion, and FileWall, which offload functional-ity for improving performance (TCPServers), improving availability (Orion), and for extending functionality (FileWall). We explore software mechanisms to offload functionality to a subset of processors in an Symmetric Multiprocessor (SMP) system, a programmable network inter-face, and an interposing network middlebox to realize the three system architectures. TCPServers is a system architecture that offloads network processing to a subset of proces-sors in an SMP system. Network processing imposes direct and indirect overheads on server systems. It directly affects system performance since it executes at a higher priority than ap-plication tasks and prevents other components of the system from executing simultaneously on the processors. It indirectly affects performance by causing cache pollution and Trans-