I85PREP
⌘ K
CDFOM study materials
CDFOM study resources

CDFOM: Data Centre Operations Management: Service Levels, KPIs and Process Control

CDFOM · Work through your exam, one topic at a time.

All guide sections

Data Centre Operations Management: Service Levels, KPIs and Process Control

Data Centre Operations Management

This domain tests how you run the data centre day to day: the processes that turn an SLA into repeatable work, the KPIs that prove service, and the governance of incidents, changes and handovers. The exam expects you to distinguish contractual service levels from internal targets and to calculate performance from agreed definitions.

Core rules and concepts

  • A service catalogue defines what operations delivers; an SLA is the customer-facing commitment, an OLA is an internal support agreement, and an underpinning contract (UC) governs a supplier.
  • KPIs must be defined with formula, data source, owner, frequency and threshold. Use leading indicators (PM completion, change success, training currency) alongside lagging indicators (incidents, availability, MTTR).
  • Incident management restores service; problem management removes root cause; change management controls risk. A change advisory board (CAB) reviews normal changes, while standard changes follow pre-approved procedures.
  • Shift handover must be structured: open incidents, changes in progress, capacity and security alerts, pending maintenance and vendor activity. A daily operations meeting reviews the same dashboard.
  • Documentation includes SOPs for steady-state tasks, MOPs for a specific change and EOPs for emergencies. Each needs version control, approval and review dates.

Worked example

A 30-day month has 43,200 minutes. The SLA commits to 99.95% monthly availability for unplanned downtime, excluding agreed planned maintenance. Allowed unplanned downtime is 0.05% × 43,200 = 21.6 minutes. The month records 35 minutes of unplanned outage. Availability = (43,200 − 35) ÷ 43,200 = 99.919%. The SLA is missed by 13.4 minutes of downtime. If the contract excluded a 10-minute outage caused by a customer-authorized test, chargeable downtime is 25 minutes and availability becomes 99.942%, still below target. Reasoning: apply the agreed measurement boundary before declaring a breach.

Traps that cost marks

  • Calculating availability without stating inclusions and exclusions, or mixing planned and unplanned downtime.
  • Treating an OLA as an SLA, or a supplier contract as an OLA.
  • Averaging KPIs across sites without weighting by exposure or volume.
  • Calling an incident closed when service is restored but the root cause remains unknown; that belongs to problem management.