Univ. of Hawaii - ITS Technical Architecture - Brick

Enterprise Job Scheduling

Primary Architect: Jan Kawachi

Description:

This brick describes the ITS approach to Enterprise job scheduling (also known as “batch scheduling”).

“Enterprise job scheduling” encompasses any scheduling system used to run production services for UH systems that are run by ITS. This could include application programs, system programs, database programs or any other type of program.

Experimental

Strategic (3-5 Years)

ONE Automation (UC4) for Enterprise Apps and Systems
SLURM for HPC

Tactical (1-2 Years)

ONE Automation, used for KFS (UC4 9.00A SP4)
SLURM for HPC

Containment

CRON for Enterprise job scheduling, used a lot in Banner
CRON/anaCRON, for SysOps, such as for log rotations
Quartz, used in MIS custom engineered apps (not KFS though)
Windows Task Scheduler, used by AT Contract Services for SysOps

Retirement

Emerging Trend:

CRON has been traditionally used by ITS technical staff to schedule jobs. Over time, we would like to eliminate the use of CRON as a tool to support enterprise systems. There are several reasons for this including:

Batch jobs are an important part of the operation of a system. As such, it’s critical that we monitor the jobs and respond when they fail. Just like we monitor the health of servers from our operations center, we should monitor the health of batch jobs. If they fail, a response protocol should be implemented. (Essentially, batch job failures should eventually be added to our incident management protocols.)
Batch jobs are often dependent upon each other. When jobs are scheduled in distributed tools like CRON then there is no way to set up formal dependencies. Ideally, jobs would be in central system where dependencies can be defined and managed.

While we are trying to stop using CRON for production systems, it would still be appropriate as a productivity tool for individuals, but not as an ongoing operational tool.)
Cloud Software as a Service products (e.g. Kuali Curriculum Management) may have their own unique scheduling infrastructure, probably embedded Quartz (it’s a library). This will complicate our scheduling in the future.

Change History

Approved August 2016

Definitions

Experimental	Someone in ITS is currently investigating or experimenting with this technology.
Strategic	ITS will be investing in this technology for 3-5 years.
Tactical	ITS will be investing in this technology for 1-2 years.
Containment	ITS will continue to use this technology for existing systems, but will no longer invest in this technology and/or grow its use.
Retirement	ITS has a firm plan (and timeline) to retire this technology.