Automation - The Practice of Cloud System Administration - page 246

12.1.4 Automation for System Administration
In “A Model for Types and Levels of Human Interaction with Automation,” Parasuraman,
Sheridan, and Wickens ( 2000 ) propose an approach to automation. Much like the discus-
sionin Section14.2.4 ,theyobservethattherearefourstagesofinformationprocessing:in-
formation is gathered, that information is analyzed, a decision is made, and action is taken.
These authors also observe that there are gradations between fully manual and fully auto-
mated. The various degrees include automation levels where the system makes a sugges-
tionforapersontoapprove,thesystemmakesasuggestionandexecutesitifapersondoes
not veto it within a certain amount of time, the system executes its decision and informs
people after the fact, and the computer decides everything and acts autonomously, not re-
quiring or asking for any human input.
Parasuraman et al.'s conclusion is that the information gathering and analysis stages are
appropriate for a high level of automation. The decision stage should be automated at a
level appropriate for the function's risk. Low-risk functions should be highly automated
andhigh-riskfunctionsshouldbelessautomated.Whenitcomestotakingaction,however,
the appropriate level of automation is somewhere in the middle for high-risk functions.
Allspaw's ( 2012c ) two-part blog post “A Mature Role for Automation” gives a detailed
explanation of Parasuraman et al.'s work and relates it to system administration. He con-
cludesthatanevenbettermodelforautomationisthatofapartnership—applyingthecom-
plementarity principle. When automation and operations staff work together, they enhance
each other's value the same way that members of a team do.
Akeyelementofteamworkistrustandsafety.AllspawcitesLeeandSee's( 2004 )meth-
ods for making automation trustworthy:
• Design for appropriate trust, not greater trust.
• Show the past performance of the automation.
• Show the process and algorithms of the automation by revealing intermediate res-
ults in a way that is comprehensible to the operators.
• Simplify the algorithms and operation of the automation to make it more under-
standable.
• Show the purpose of the automation, design basis, and range of applications in a
way that relates to the users' goals.
• Train operators regarding its expected reliability, the mechanisms governing its be-
havior, and its intended use.
• Carefully evaluate any anthropomorphizing of the automation, such as using
speech to create a synthetic conversational partner, to ensure appropriate trust.