Metrics
A system to record in real-time information for resources
On the servers assigned to the Customer, EMENCIA deploys a metrics system to record in real time all information related to system resources and changes thereto over time.
This system allows for proactive advanced planning of certain dysfunctions (e.g. excessive consumption of resources will be identified and can then be corrected before causing the unavailability of the server and its applications). It also allows for a much more detailed diagnosis of complex problems. For example, if an incident occurs, the availability of a precise history of resource consumption allows for verification that the incident is not related to storage occupancy, to the number of sessions recorded in the database, etc.
Deployment of a measuring system.
Ad Hoc Metrics System
The system EMENCIA offers allows, in particular, for the identification and preparation of graphs based on the following information:
- The number of processes assigned to each application (Apache frontends, PHP processing, Django processing, PostgreSQL database backends)
- Resource consumption for each of these process groups
- Precise monitoring of the PostgreSQL database activity (number of queries in read mode, updates, inserts) at a global level, as well as base by base and table by table
- Utilization statistics on the PostgreSQL database indexes allowing developers to ensure they match the queries made
- Number of queries processed using the HTTP cache mechanism (total queries, cached queries, uncached queries)
- Number of queries processed by the Apache frontends (query traffic, data volume and status of the various Apache processes involved)
- Monitoring of the Django application platform (recording of the number of sessions).
Even more specific metrics can be implemented if necessary. For example, for a system that manages a queue (for the transmission of orders or access to a specific service), a measurement of the length of the queue can be added.