ITIL Availability, Capacity, Performance and Service Continuity: RTO, RPO and Downtime Explained

A service that is not there when needed, too slow to use or unrecoverable after a disaster has no value, however good its features. Three ITIL practices protect the warranty side of a service (see ITIL Key Concepts: Value, Utility and Warranty): availability management, capacity and performance management, and service continuity management.

Availability Management

Availability management ensures that services deliver agreed levels of availability to meet the needs of customers and users. Availability is usually expressed as a percentage of agreed service time:

Availability % = (Agreed service time − Downtime) ÷ Agreed service time × 100

AvailabilityDowntime per yearDowntime per month (about)
99%3.65 days (87.6 hours)7.3 hours
99.5%43.8 hours3.65 hours
99.9%8.76 hours43.8 minutes
99.95%4.38 hours21.9 minutes
99.99%52.6 minutes4.4 minutes

Each extra “nine” costs much more: redundancy, automated recovery, 24x7 staffing and tested processes. Agree the level with the business based on what an outage really costs (see Service Level Management and SLAs).

Reliability measures

  • MTBF (mean time between failures): average operating time between failures. Higher is better.
  • MTTR (mean time to repair or restore): average time to recover from a failure. Lower is better.
  • Single point of failure: a component whose failure stops the service. Availability work aims to find and remove them.

Capacity and Performance Management

This practice ensures that services achieve agreed and expected performance, satisfying current and future demand in a cost-effective manner. It looks at capacity (how much the service can handle) and performance (how quickly it responds under load). Activities include:

  • Monitoring usage of CPU, memory, storage, bandwidth, licences and people.
  • Analysing trends and forecasting demand, including seasonal peaks.
  • Modelling and load testing before launch or major change.
  • Planning to add, remove or resize capacity, including cloud auto-scaling.
  • Balancing cost against risk: too little capacity causes slowness, too much wastes money.

For example, an online retailer expects traffic to triple during a festival sale. Capacity management reviews last year’s peak, runs a load test at three times normal traffic, finds the database is the bottleneck and arranges extra capacity for the sale week, then scales back to avoid paying for it all year.

Service Continuity Management

Service continuity management ensures that availability and performance are maintained at a sufficient level in the event of a disaster. Disasters include fire, flood, power loss, cyberattack, major supplier failure and loss of a key site. The practice relies on:

  • Business impact analysis (BIA): identifies critical services, how long the business can survive without them and the effect of loss.
  • Risk assessment: identifies threats and how likely and damaging they are.
  • Continuity strategy and plans: how to keep running or recover, including roles, contacts and steps.
  • Testing and exercises: an untested plan is only a hope.

RTO and RPO

TermMeaningExample
Recovery time objective (RTO)The maximum acceptable time to restore a service after a disruptionThe order system must be back within 4 hours
Recovery point objective (RPO)The maximum acceptable amount of data loss, measured in timeNo more than 15 minutes of orders may be lost

A tight RPO needs frequent backups or replication; a tight RTO needs standby systems and rehearsed recovery. Both cost money, so set them per service according to business impact.

Disaster Recovery vs Business Continuity

Disaster recovery focuses on restoring IT systems and data. Business continuity is wider: how the business keeps operating, including people, premises and manual workarounds. IT service continuity is the IT contribution to the business continuity plan.

Worked Example

A bank’s BIA shows that online payments are critical: every hour of outage costs revenue and customer trust. It sets RTO to 1 hour and RPO to 5 minutes, which requires replicating data to a second site and rehearsing failover. Twice a year the team runs a test: they switch to the standby site during a planned window and measure recovery time. The first test takes 2 hours because a firewall rule was missing; the team fixes it and the next test takes 40 minutes. The test results are reported to the service level review.

Common Mistakes

  • Promising high availability without funding redundancy.
  • Forecasting capacity only after users complain about slowness.
  • Writing a continuity plan and never testing it.
  • Backing up data without testing that it can be restored.
  • Ignoring supplier dependence in continuity planning.
Key takeaways
  • Availability is a percentage of agreed service time; each extra nine costs much more.
  • Capacity and performance management forecast demand and balance cost against risk.
  • Service continuity relies on business impact analysis, tested plans and clear RTO and RPO targets.
  • An untested backup or continuity plan is an assumption, not a safeguard.

Related Lessons