Monitoring and uptime
Some links on this page are affiliate links: we may earn a commission at no extra cost to you. Our approach
Monitoring makes visible what happens behind the company's back: a payment failure at midnight, an overloaded server, a form that no longer records anything. It fixes nothing on its own, but it reduces the blind spot and the reaction time. ## Alerts you can trust A useful alert is rare, precise and actionable. Too many alerts create fatigue that leads to ignoring precisely the one that matters. Fewer, well-chosen thresholds are better. ## Monitor journeys, not just pages Checking that the home page responds is not enough: you also need to monitor the purchase funnel, the contact form, login and transactional email sending. ## From signal to improvement Every incident should lead to an adjustment: a threshold, a procedure, a fix or a redundancy. Monitoring has value when it feeds decisions.When this guide is for you
- You find out about outages from unhappy customers rather than from an alert
- Your forms or payment fail and nobody notices for hours
- You run several sites or applications and do not know which one is struggling
- You are preparing a launch or a busy period (Ramadan, Black Friday)
- You want to measure an availability target (for example 99.9%) and meet it
Step by step
- Defining what matters — Choose the useful indicators: uptime, response time, error rate, conversions, database, email. Too many metrics dilutes attention.
- Setting up the probes — Configure uptime tests from several regions, journey tests (login, payment) and collection of server and application logs and metrics.
- Sending alerts to the right people — Define thresholds and escalations, avoid redundant alerts and false positives. An alert should trigger an action.
- Shared dashboards — Create a readable dashboard for the technical team and a summary one for management (uptime, incidents, recovery time).
- Review and improvement — Analyse incidents after the fact, adjust thresholds and document procedures. Monitoring evolves with the product.
What it costs (in MAD)
| Service | Indicative price | Good to know |
|---|---|---|
| Setting up monitoring | 3,000 – 15,000 MAD (≈ 270 – 1,400 €) | Probes, alerts and dashboards |
| Hosted monitoring service | 0 – 1,500 MAD / month (≈ 0 – 140 €) | UptimeRobot or Better Stack depending on the number of probes |
| Application monitoring (APM) | 1,000 – 8,000 MAD / month (≈ 90 – 730 €) | Traces, errors and transaction performance |
| On-call duty and alert handling | 2,000 – 10,000 MAD / month (≈ 180 – 920 €) | Incident response according to a defined SLA |
| Uptime and performance audit | 4,000 – 15,000 MAD (≈ 370 – 1,400 €) | Failure points, dependencies and recommendations |
Indicative ranges for the Moroccan market in 2026, excluding tax. Prices vary with scope and provider.
Mistakes to avoid
- Creating dozens of alerts that all end up ignored
- Monitoring only the home page while payment or the form is down
- Alerting one person without a procedure or a backup owner
- Measuring without ever analysing incidents and their causes
- Assuming the host's uptime is enough to guarantee your service's uptime
Frequently asked questions
What uptime should you aim for?
99.9% (about 43 minutes of downtime per month) is a reasonable target for an SME. 99.99% requires far more redundancy and budget. What matters is measuring what you promise.
Is monitoring useful for a simple brochure site?
Yes, at least to know when the site goes down and be alerted before your clients. A free probe every five minutes is enough at this level.
What is the difference between uptime and APM?
Uptime checks that the service responds. APM (application performance monitoring) analyses errors and response times at transaction level: it identifies where the slowdown happens.
Should external dependencies be monitored?
Yes: payment gateway, email service, delivery, DNS host. A dependency incident can bring your service down without your servers being at fault.
How long does it take to set up monitoring?
A basic setup with two or three probes and an email alert takes a few hours. Full monitoring with dashboards and procedures is built over a few weeks.
Updated on
Related guides
Version française: /services/devops-securite/monitoring/ · AR · NL