When you manage dozens of machines you need a to have one entry-point to monitor all the infrastructure.
Monitoring an infrastructure is not magic, you need probes and data.
With our current infrastructure we gather:
- application logs (apache, tomcat, database, ...) with Logstash
- host server logs (hard disk, memory, cpu, server login...) with MetricBeats
Our monitoring team use all these metrics to build counters, alerts and visualization graphs. See some practical examples below.
Using the apache logs you can extract an approximate origin of the people connecting (or trying to connect) to your solution.
If your apache server handles 80% of its traffic with connections coming from countries where you don't have clients, a simple fail2ban solution will probably be interesting to block this traffic.
The apache server may act as a proxy, so a DDOS attack will also trigger alert if you monitoring your sever wisely.
Hard disk monitoring and memoy consumption are two important metrics for the hosts. Do we need to add more hard disk, do we need more memory ?
Monitoring is very important when setting up a service.
You may have the best software, but if you are not able to monitor it properly, then you will finnaly crash your system.
At NIRVA we have dedicated a team to control all our internal and cloud infrastructures. Every month plans are made to adapt the current monitoring solution.
This is why we decided to focus on setting up dedicated cloud environments for our customers. We take care of the infrastructure by managing the servers, and you take care of developing your business using your dedicated Intelligent Hybrid Mail instance.