Real-time observability and alerting

Deployment of Zabbix, Nagios, and Grafana with instant Telegram bot alerts for critical routers and servers. Before: no unified visibility, reactive failure detection dependent on user reports.

Key metrics

Monitored hosts: 0 → 200+
Checks per minute: 5k
MTTR: 90 min → 30 min (67% less)
Alerts: Telegram < 60 s
Coverage: 100% of critical infra

Key facts

Scope: core routers and servers in production (200+ hosts).
Alerts: notifications in under 60 s via Telegram.
Goal: operational visibility, lower MTTR, and fast response.

Stack

Zabbix
Nagios
Grafana
Telegram Bot

Sanitized observability diagram

Routers/servers
Zabbix/Nagios
Grafana
Telegram alerts

Logical view without hosts, IPs, internal thresholds, or sensitive service names.

Context

The ISP's operation relied on manual failure detection. Alerts arrived through customer calls, not monitoring. There was no unified dashboard and no response-time metrics. The on-call team needed real-time visibility and actionable alerts.

Actions

Deployed Zabbix to monitor core infrastructure and services (SNMP, ICMP, custom checks).
Built Grafana dashboards for real-time tracking of traffic and router/server health.
Integrated alerts with a Telegram bot: priority-based notifications, on-call group, and automatic escalation.

Anonymized global Zabbix view with problems, hosts, and monitoring metrics — Anonymized Zabbix: global view of problems, hosts, and operational status.

Anonymized Nagios summary with host groups, services, and check status — Anonymized Nagios: status summary for critical hosts and services.

Anonymized Grafana dashboard with traffic, CPU, memory, and storage metrics — Anonymized Grafana: operational metrics for traffic, resources, and availability.

Results

MTTR reduced from 90 to 30 minutes: incident response 3x faster.
Alerts in under 60 seconds for critical events (link down, high CPU/memory usage, full disk).
Unified operational dashboard: the NOC team moved from reactive to proactive monitoring.