Monitoring
What is IT Monitoring?
IT Monitoring describes the continuous observation, measurement, and analysis of IT systems, networks, and applications. The goal is to detect anomalies, performance degradations, and failures early – ideally before they impact operations or users. Monitoring is a core component of any professional IT operations strategy.
Types of monitoring
Infrastructure monitoring
Infrastructure monitoring tracks the health and performance of physical and virtual hardware components:
- Server metrics: CPU utilisation, memory usage, disk I/O, temperature
- Network metrics: Bandwidth utilisation, packet loss, latency, interface errors
- Storage metrics: Disk usage, IOPS, throughput
- Hypervisor metrics: VM density, overcommitment ratios
Application performance monitoring (APM)
APM tools track the performance and availability of applications from the end-user perspective:
- Response times (page load time, API latency)
- Error rates and exception tracking
- Transaction tracing (distributed tracing in microservices)
- Dependency mapping
Log monitoring
Application and system logs contain valuable information about errors, security events, and unusual behaviour. Log management platforms (ELK Stack, Splunk, Graylog) aggregate logs from all sources, enable searching and correlation, and trigger alerts on defined patterns.
Synthetic monitoring
Synthetic monitors simulate user interactions (e.g. login, form submission, checkout process) at regular intervals from multiple geographic locations. This proactively detects availability and performance issues before real users are affected.
Security monitoring (SIEM)
Security Information and Event Management (SIEM) correlates security-relevant events from across the infrastructure to detect attack patterns, anomalies, and policy violations.
Alerting and escalation
Monitoring only delivers value if anomalies trigger timely, actionable alerts. Effective alerting:
- Uses thresholds based on historical baselines rather than arbitrary fixed values
- Implements alert fatigue countermeasures (deduplication, correlation, suppression)
- Defines clear escalation paths (email → SMS → phone → on-call)
- Includes runbooks so on-call engineers know how to respond
Key monitoring metrics
The Four Golden Signals (Google SRE)
- Latency: Response time of service requests
- Traffic: Volume of requests per second
- Errors: Rate of failed requests
- Saturation: Resource utilisation (how "full" the system is)
RED method (for services)
- Rate: Requests per second
- Errors: Error rate
- Duration: Response time distribution
USE method (for resources)
- Utilisation: Percentage of time the resource is busy
- Saturation: Amount of work queued
- Errors: Error count
Monitoring in managed service contexts
For Managed Service Providers, monitoring is central to service delivery. 24/7 monitoring allows proactive incident detection and resolution, often before the customer is affected. Monitoring data forms the basis for SLA reporting and capacity planning.
