InfraWatch — System Case Study
From Customer Complaints to Controlled Response
Why an owner-readable monitoring system had to exist
The Context
Across multiple client engagements, the same failure pattern repeated: downtime was discovered by a customer, a partner or a founder refreshing a page — never by the systems themselves.
Monitoring tools existed in the market, but they reported in a vocabulary that operations owners could not act on, and they produced alert volumes that teams learned to ignore.
The Requirement
Why This System Was Needed
Before the System
The Operational Headaches
Discovery by Complaint
Outages surfaced through support tickets, meaning the commercial damage was already done before anyone in the business knew.
Alert Fatigue
Untuned thresholds produced constant noise, so genuine incidents were dismissed alongside routine flapping.
Undefined Escalation
No agreed chain of responsibility meant incidents stalled in group chats while the clock ran.
No Incident History
Post-mortems relied on recollection; repeat failures were never recognised as patterns.
Unprovable Uptime
Client SLA reviews turned into negotiations over whose recollection of downtime was accurate.
Maintenance Distortion
Planned work counted against availability, making every reliability report structurally wrong.
The Turning Point
The decision was to stop buying monitoring for engineers and build monitoring for the business — a system where detection, escalation, communication and evidence lived in one operating record.
The Build
What We Engineered
Monitor Registry
A single inventory of every business-critical endpoint, each with an owner, interval and threshold.
Check Engine
Scheduled availability, latency and status checks with degradation detection ahead of hard failure.
Incident Lifecycle
Automatic incident creation, acknowledgement, timeline capture and resolution recording.
Escalation Policies
Tiered, time-based routing so unacknowledged incidents rise automatically.
War Room
A live consolidated view for active incidents, replacing scattered chat threads.
Maintenance Windows
Declared planned work excluded from alerting and availability maths.
SLA Reporting
Period-based uptime, response trends and SLA attainment for internal and client review.
The Shift
Before and After
| Dimension | Before | After |
|---|---|---|
| Detection | Customer or founder reports the outage | Automated check detects and opens an incident |
| Escalation | Ad-hoc messages to whoever responds | Tiered policy escalates on a timer |
| Incident record | Chat history and recollection | Structured timeline per incident |
| Uptime reporting | Estimated, contested | Calculated from recorded checks |
| Planned maintenance | Counted as downtime | Excluded via declared windows |
| Owner visibility | Requires an engineer to interpret | Readable business status view |
The Result
Outcomes
Next Step
See InfraWatch in action
We can walk you through the system and how it maps to your operation.