The Jaquith Archive Server

Abstract

Advances in robotic devices and storage media now make it possible to design near-line automated storage systems. These systems aim to provide responsive performance to users of tertiary storage devices. The Jaquith system is a prototype archive server that lets network users archive their own files using automated storage. It provides semi-interactive file access to its clients by combining a high-density robotic tape system with disk-based indexing. Jaquith presents an FTP interface whereby whole files are moved between the client and its storage archive. Each client’s archive is separately governed to provide independent namespaces, added security, and parallel operation. A wildcard query mechanism lets users manipulate arbitrary subsets of their files. Two important aspects of the query system are abstracts, text tags that can be associated with files, and versions, date-stamps that are applied to archived files. Jaquith throughput is about 135 KB/second when archiving small (10 KB) user files to disk buffers. The use of synchronous disk writes by the server to ensure durability of each user file degrades throughput to 40 KB/second. The performance when writing disk buffers to Exabyte or Metrum tape is severely limited by the time to write a hardware filemark. Consequently, it is important to write several megabytes of data between filemarks for good performance. Report Documentation Page Form ApprovedOMB No. 0704-0188 Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information,