Dynamic memory allocation has been a fundamental part of most computer systems since roughly 1960, and memory allocation is widely considered to be either a solved problem or an insoluble one. In this survey, we describe a variety of memory allocator designs and point out issues relevant to their design and evaluation. We then chronologically survey most of the literature on allocators between 1961 and 1995. (Scores of papers are discussed, in varying detail, and over 150 references are given.) We argue that allocator designs have been unduly restricted by an emphasis on mechanism, rather than policy, while the latter is more important; higher-level strategic issues are still more important, but have not been given much attention. Most theoretical analyses and empirical allocator evaluations to date have relied on very strong assumptions of randomness and independence, but real program behavior exhibits important regularities that must be exploited if allocators are to perform well in practice.

Consistent checkpointing provides transparent fault tolerance for long-running distributed applications. In this paper we describe performance measurements of an implementation of consistent checkpointing. Our measurements show that consistent checkpointing performs remarkably well. We executed eight compute-intensive distributed applications on a network of 16 diskless Sun-3/60 workstations, comparing the performance without checkpointing to the performance with consistent checkpoints taken at 2-minute intervals. For six of the eight applications, the running time increased by less than 1% as a result of the checkpointing. The highest overhead measured for any of the applications was 5.8%. Incremental checkpointing and copy-on-write checkpointing were the most effective techniques in lowering the running time overhead. These techniques reduce the amount of data written to stable storage and allow the checkpoint to proceed concurrently with the execution of the processes. The overhead ...

Delta compression, which consists of compactly encoding one le version as the result of changes to another, can improve eciency in the use of network and disk resources. Delta compression techniques are readily available and can result in compression factors of ve to ten on typical data. Managing delta-compressed storage, however, is a dicult task. I will present a system that attempts to isolate the complexity of delta-compressed storage management by separating the task of version labeling from performance issues. I will show how the system integrates delta-compressed transport with delta-compressed storage. Existing tools for managing delta-compressed storage suer from weak le system support. Lack of transaction support is responsible for inecient application behavior. The only atomic operation in the traditional le system forces unnecessary disk activity due to copying costs. I will demonstrate that transaction support can improve application performance and extensibility wit...

Extended version An Archival Repository reliably stores digital objects for long periods of time (decades or centuries). The archival nature of the system requires new techniques for storing, indexing, and replicating digital objects. In this paper we discuss the specialized indexing needs of a write-once archive. We also present a reliability algorithm for effectively replicating sets of related objects. We describe an administrative user interface and a data import utility for archival repositories. Finally, we discuss and evaluate a prototype repository we have built, the Stanford Archival Vault, SAV.

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. For more information about the USENIX Association:

this report we discuss the strategic directions and challenges in the management and use of storage systems---those components of computer systems responsible for the storage and retrieval of data. Typical large-scale storage systems include the following components of secondary and tertiary storage:

It may be important to provide long-term preservation of digital data even when that data is stored in an unreliable system, such as a filesystem, a legacy database, or even the World Wide Web. In this paper we focus on the problem of archiving the contents of a web site without disrupting users who maintain the site. We propose an archival storage system, the InfoMonitor, in which a reliable archive is integrated with an unmodified existing store. Implementing such a system presents various challenges related to the mismatch of features between the components, such as differences in naming and data manipulation operations. We examine each of these issues as well as solutions for the conflicts that arise. We also discuss our experience using the InfoMonitor to archive the Stanford Database Group's web site. 1 Introduction Important and irreplaceable data is constantly being lost. For example, some data from the 1960 U.S. Census has disappeared, and files detailing land usage ...

Power consumption and speed are the largest costs for a virtual memory system in handheld computers. This paper describes a method of trading off computation and useable physical memory to reduce disk I/O. The design uses a compression cache, keeping some virtual memory pages in compressed form rather than sending them to the backing store. Efficiency is managed by a log-structured circular buffer, supporting dynamic memory partitioning, diskless operation, and disk spin-down.

The Self-certifying File System (SFS) currently exports Unix filesystems. Consequently, file owners on SFS servers who want to give other users access to their files can do so only through the coarse-grained Unix access control mechanisms, which are based on locally de ned user identifiers and group identifiers. Therefore, despite the fact that SFS was designed to be a global filesystem, it is impossible for a file owner to grant access permissions to a remote SFS user who does not maintain a Unix account on the local machine. Moreover, creating new user identifiers or group identifiers for the purpose of access control requires the involvement of the local machine's administrator, a limitation that runs counter to the egalitarian spirit of SFS.

SFS for Windows is a Microsoft Windows 2000 client for the Self-Certifying File System. It is a direct port of the original using Cygwin. It uses a commercial NFS client to connect Windows to SFS. It provides support for symbolic links in Cygwin by converting these into Windows Shortcuts. It automounts file systems in Windows using a mount daemon and the portmapper. The client also includes a new system to pass file descriptors between Cygwin processes. All this was accomplished with minimal code changes to the original UNIX code.