5.4Service Management

Service management is the discipline of not just building a system but continuing to operate it stably. Based on the best practices ITIL lays out, the core of what is asked both in practice and on the exam is understanding how to manage the promised level of service (SLA/SLM) and how to distinguish the processes for responding to failures (incident/problem/change management).

5.4.1ITIL and SLA/SLM

• ITIL (Information Technology Infrastructure Library) is a collection of best practices for IT service management. It is a framework widely referenced as an industry standard, independent of any specific product or organization.
• SLA (Service Level Agreement) is a document agreed between a service provider and its users that sets numerical targets for the level of service provided (availability rate, response time, and so on). SLM (Service Level Management) is the overall activity of continuously monitoring and reporting on the levels agreed in the SLA, and managing toward achieving them. The relationship is that the SLA is "the content of the promise," while SLM is "the management process for keeping the promise."

5.4.2Incident, problem, change, and configuration management

• Incident management is the process aimed at restoring service as quickly as possible when a service interruption or degradation occurs (centered on immediate response and workarounds; it does not aim to pursue the root cause). Problem management is the process of identifying the root cause behind incidents and implementing permanent measures to prevent recurrence.
• Change management is the process of planfully evaluating and approving changes to a system (configuration changes, patch application, feature additions, and so on) and carrying them out through a controlled process, preventing new failures caused by the change itself. Configuration management (in the service management context) is the process of accurately grasping and recording configuration information (CIs: Configuration Items) for IT assets, equipment, software, and so on, using it as a basis for decisions in change management and incident response.
• The most important distinction is the difference between incident management and problem management. Incident management is "keep the service running now, restore it," while problem management is "never let the same failure happen again"--the two differ in their time horizon of purpose (immediate vs. permanent).

5.4.3Service desk, capacity/availability management, and facilities

• The service desk is the single point of contact that receives inquiries and failure reports from users. Anything that cannot be resolved at first contact is escalated to the appropriate specialist team. Capacity management is the process of planning so that processing capacity and resources do not fall short, anticipating future demand. Availability management is the process of planning redundant configurations and failure-response arrangements so a service stays usable whenever it is needed.
• Facility management is the area of managing the physical equipment of data centers and similar sites. A UPS (Uninterruptible Power Supply) is a device that continues supplying power for a limited time during an outage (its main role is not sustained long-term generation but bridging the gap until a backup generator takes over or a safe shutdown completes). Seismic isolation (menshin) is a structure that absorbs shaking between the building and the ground so earthquake motion is not transmitted directly to the building (note that seismic resistance (taishin) is a different approach--strengthening the building itself to withstand shaking--and the way motion is transmitted differs between the two).

Take a failure where order processing on an e-commerce site suddenly slowed down as an example of how the roles divide across processes. A user's inquiry is first received at the service desk, the single point of contact. The service desk does an initial triage of the symptom and escalates it to the responsible team. That team now enters the incident management process, and rather than digging deeply into "why is it slow" first, restoring the service to a normal state is the top priority. For example, temporarily pausing a heavy batch job or restarting a cache might restore normal response times, at which point the incident is closed via a workaround. But if the same symptom keeps recurring, the process moves into problem management, where the root cause--say, "a particular query is doing a full scan instead of using an index"--is investigated in depth over time, and a permanent fix (adding an index, revising the query) is planned. When this permanent fix is applied to production, it follows the change management process: the scope of impact and timing are evaluated and approved in advance, and the change is carried out through a controlled procedure. To ensure this change does not cause an unintended side effect, the system configuration information recorded under configuration management (which servers depend on which middleware, and so on) is referenced as a basis for the decision. In parallel, if the failure was caused by a surge in access, capacity management plans a resource increase anticipating future demand growth, and availability management considers strengthening the redundant configuration. Underlying all of this, only once there is a physical foundation--the data center's servers can shut down safely on a UPS's limited-duration power supply during an outage, and a seismic isolation structure keeps them less directly affected by earthquake shaking--do the various service management processes have a base on which to function.

5.4.4Section summary

• SLA = the document of agreed levels; SLM = the activity of continuously monitoring/managing toward them. ITIL = the best-practice collection for IT service management
• Incident management aims for swift restoration (does not pursue root cause); problem management identifies root cause and applies permanent fixes to prevent recurrence. Change management plans, evaluates, approves, and controls changes
• UPS bridges a short gap during an outage (a backup generator substitutes long-term). Seismic isolation absorbs shaking between building and ground (seismic resistance withstands via the building itself)