Service Research has been exploring short- and long-term approaches to automation of server management. Among the multiple goals of these efforts are reduction of labor cost, faster resolution of problems, standardization of procedures, and minimization of human error. Automation can take the form of tools designed for human system administrators to use to improve their efficiency and effectiveness, in situations that require human intelligence and decision making. For situations that are fully characterized and for which reliable automation remediation exists, automation can be executed without human involvement. Our short term approach exploits the capabilities of IBM Endpoint Manager (IEM) to address the most commonly observed types of work in IT infrastructure management. Our longer term approach includes new frameworks and OS level enhancements to govern server state autonomically.
Service Delivery Automation (SDA)
The SDA project created an automation tool designed to reduce the labor hours spent by System Administrators working on three common types of problem tickets (disk full, OS alert, and application alert). We identified the repetitive activities to get diagnostic information and to resolve these types of problems. Using IBM Endpoint Manager, the strategic tool for GTS SO server management, we built a custom set of fixlets that collect diagnostic data useful for these problems and another set that implements commonly used remedial actions. These fixlets and the diagnostic information are accessible from a custom dashboard that is dedicated to resolving these types of problem tickets. The disk full fixlets search for files that are candidates for deletion, truncation, or archiving followed by compression based on a set of rules that is customizable for each account or server group. SDA supports a spectrum of Windows and Unix platforms and versions.
Runbook Automation (RBkA)
Runbook Automation is a suite of automation tools that expedite maintenance, configuration, installation, inventory, and activation/deactivation activities commonly performed on managed servers by system administrators. RBkA is built on IBM Endpoint Manager and includes a novel command line interface that can broadcast OS or shell commands to multiple target servers, and display the consolidated responses in the same user interface. RBkA saves system management labor by eliminating the need to log in to each server and issue the same commands. Moreover, OS or shell commands are validated against a white list to prevent execution of potentially harmful commands (on large numbers of servers) by unauthorized users. The system is easily extensible by adding new shell scripts without modification.
This project aims to develop frameworks that enable full automation without human involvement whenever possible. In the simplest case, we have implemented proactive automation such as disk cleanup scripts that remove or compress specific files or directories on a periodic schedule, thus reducing the frequency of disk full problem tickets. A more challenging goal is to develop a change management framework that captures system state and account server management policies, and enables system administrators to make only those changes that comply with the policies. This system will bring the benefit reduced human error as well as reduced labor cost through automated change execution.
Service Delivery Analytics for Productivity Improvement
Productivity improvement in Service Delivery Operations is a major enterprise level target. Solid and measurable productivity improvement results can be achieved only with a deep understanding of the current processes and procedures, their limitations and business implications. In this project we identify and quantify opportunities for process optimization and activity automation that lead to productivity improvements. Towards this end, we develop tools for analysis of service delivery content and perform targeted analysis of activity domains. Our methodology is based on integration across a multitude of content domains, on collection of new data based on a refined understanding of the scope, and on estimation of long-term expected impact by correlation of the designed process changes with the long-term workload patterns.
Security and Compliance are two important pillars of any IT Delivery organization. Many times these two are treated as one because they so inter-related, but they are not the same . While security is about getting the latest fix to the latest vulnerability, or making a system physically secure, compliance is about making sure these security operations are applied in a timely and orderly fashion. This project seeks to transform the IT compliance processes using formal business process modeling (BPM) techniques using Service Oriented Architecture (SOA) approach. This brings about the traditional benefits of BPM, like process efficiency, standardized documentation and process control, but also enables something completely new - the continuous compliance. It enables continuous monitoring and management of the compliance posture of a large group of systems in an audit friendly manner without increasing the cost of doing so. It also enables real-time (auto-)fixing of compliance deviations thus significantly reducing the risk exposure window. An implementation of this approach is done using IBM BPM v8 integrated with IBM Tivoli End Point Manager.
Universal Script Wrapper - An Innovative Solution to Manage Endpoints in Large and Heterogeneous Environments
Sai Zeng, Shang Guo, Frederick Wu, Constantin Adam, Long Wang, Cashchakanithara Venugopal, Rajeev Puri, Ramesh Palakodeti
IFIP/IEEE International Symposium on Integrated Network Management (IM 2013), Ghent, Belgium, May 27-31, 2013
Effort Instrumentation and Management in Service Delivery Environments
Melissa Buco, Daniela Rosu, Dikran Meliksetian, Frederick Wu, Nikos Anerousis
Connecting the Dots in IT Service Delivery
Daniela Rosu, Winnie Cheng, Ea-Ee Jan, Naga Ayachitula