Monitoring
We watch the platform continuously to detect problems, ideally
before they affect you.
What we monitor
- Availability: that your application responds correctly.
- Performance: response times and resource usage.
- Component health: the status of each piece of the service.
- Capacity: compute, memory and storage usage, to anticipate needs.
- Security signals: anomalous behavior.
How we do it
The platform has a monitoring and observability system that collects
operational metrics and logs, and automatic alerts that notify our team
when an indicator falls outside expected ranges.
How we determine your service's status
We continuously probe your service from outside (approximately every 30
seconds) and classify its status into three levels:
| Status |
What it means |
| 🟢 Operational |
Responds correctly and with a normal response time. |
| 🟠 Degraded performance |
Still working, but slower than usual (or intermittently). |
| 🔴 Down |
Not responding or returning an error. |
We don't apply a single fixed number for everyone: we compare each service
against its own normal behavior (its usual response time, calculated
over the last few days). We consider there to be degraded performance
when, sustainedly:
- the response time stays above about 3 times its usual value, or
- it exceeds an absolute safety threshold (on the order of 2 seconds),
even if that service is usually slow, or
- the service fails intermittently (some probes succeed and others
don't).
…and only if the situation persists for at least 5 minutes. This avoids
alerts for a one-off spike: a brief network blip or a single slow query
do not count as degradation.
How we detect a service being down
A down status is detected immediately: as soon as a probe fails to get
a correct response, the service switches to 🔴 and the alert is generated
for our team without delay.
Availability (SLA)
In addition to the current status, we show the availability percentage
over different periods (1, 7 and 30 days): the percentage of time the
service responded correctly. A service that is slow but working counts as
available — slowness affects the performance status, not the
availability percentage.
Transparency
You can check the real-time status at
status.dyncloud.es.
Detailed status for customer environments requires authenticated
access.
Operational logs
We keep operational logs of the platform that allow us to diagnose
incidents and understand what happened. These logs are managed in
accordance with our security and data protection policies.
Proactivity
The goal of monitoring is not only to react, but to anticipate:
spotting a trend (for example, growing capacity needs) and acting
before it becomes a problem for you.