Since the early days of UNIX, the administrator has searched for ways to make their life easier. Scripts provide the basic building blocks for automation and coupled with a job scheduler just about any task may be automated. CRON (and AT) are the standard tools distributed by the major UNIX vendors to manage scheduled jobs. While they perform the basic functions very well, today’s administrator (and user) is required to deliver more than ever before.

This paper is targeted towards UNIX administrators and managers and maybe even the highly adept (technical) end user. The concepts, methods and tools presented herein are designed for the UNIX savvy technician. While the concepts may carry over to other environments, such as Windows NT/2000, those will only understand the majority of the material familiar with UNIX.

CRON is a wonderful tool for scheduling everyday tasks on a UNIX server. It’s fairly straight forward, and once you get used to its features, syntax and methods you find many places you can take advantage of it. The problems it presents aren’t because of a lack of functionality, but rather age. At the time of its design, the problems that had to be solved were quite simple in nature and the design shows this. Think about how far the computer industry has evolved over the past 20 years, while programs such as CRON as just about the same as they were when they were designed (20 years ago). While there are several versions of the CRON program that deliver various features and functionality, they all have one foot in the same basic design principles. I believe, as well as others that have contributed to this paper, that it is time for a complete redesign of the UNIX job scheduling system. There are several commercial packages available today for the “enterprise” to choose from for job scheduling, but your choices are limited. We aim to change this. The goal of this paper is to deliver a design for just such a solution. The primary areas of concern are: flexibility, usability, standardization and scalability. We believe that if the solution is presented in an open-source type forum these goals may be achieved. In the end a draft proposal for a job scheduling standard (RFC) along with a basic implementation will be delivered.

To top