A case study in software adaptation
WOSS '02 Proceedings of the first workshop on Self-healing systems
High speed and robust event correlation
IEEE Communications Magazine
Emergent (mis)behavior vs. complex software systems
Proceedings of the 1st ACM SIGOPS/EuroSys European Conference on Computer Systems 2006
Towards an understanding of decision complexity in IT configuration
Proceedings of the 2007 symposium on Computer human interaction for the management of information technology
Discrete control for safe execution of IT automation workflows
Proceedings of the 2nd ACM SIGOPS/EuroSys European Conference on Computer Systems 2007
Refined failure remediation for IT change management systems
IM'09 Proceedings of the 11th IFIP/IEEE international conference on Symposium on Integrated Network Management
Toward quantifying system manageability
HotDep'08 Proceedings of the Fourth conference on Hot topics in system dependability
Hi-index | 0.00 |
The high cost of IT operations has led to an intense focus on the automation of processes for IT service delivery. We take the heretical position that automation does not necessarily reduce the cost of operations since: (1) additional effort is required to deploy and maintain the automation infrastructure; (2) using the automation infrastructure requires the development of structured inputs that have up-front costs for design, implementation, and testing that are not required for a manual process; and (3) detecting and recovering from errors in an automated process is considerably more complicated than for a manual process. Our studies of several data centers suggest that the up-front costs mentioned in (2) are of particular concern since many processes have a limited lifetime (e.g., 25% of the packages constructed for software distribution were installed on fewer than 15 servers). We describe a process-based methodology for analyzing the benefits and costs of automation, and hence for determining if automation will indeed reduce the cost of IT operations. Our analysis provides a quantitative framework that captures several traditional rules of thumb: that automating a process is beneficial if the process has a sufficiently long lifetime, if it is relatively easy to automate (i.e., can readily be generalized from a manual process), and if there is a large cost reduction (or leverage) provided by each automated execution of the process compared to a manual invocation.