fdmon 2020 : new design, new features, improved analytics and data persistency

Until 2019, fdmon database relied on a MariaDB cluster, with a replication also managed by MariaDB. In order to be more robust, available, scalable and distributable, fdmon database is now managed by a proprietary distributed database engine designed especially for high network latencies, allowing large distributions and long distance synchronous replications. One of this new database engine functions is to convert sequential IOs to parallel IOs, in order to mitigate network latencies. The new fdmon database architecture provides better performances, especially when replaying events of the past, and doesn't require any locking mechanism thanks to a partitioning based design. Furthermore, performances are no longer linked to the size of the database.

Replication and failover features have been offloaded to the underlying distributed storage solution (CEPH, Lustre, GlusterFS, ...) and deploying a private fdmon cloud is now much easier.

In this new release, the fdmon black box concept (events recording capabilities) has been dramatically improved thanks to new data persistency features. Henceforth, fdmon can provides specific granular information from the past, such as the following ones, independently from the event management and centralized logs management :

• What was the status of the "ethernet 1/1/22" port on this network router last sunday at 21:13 ?
• Which processes were issuing the most of IOs on this specific disk device at this time ?
• On which physical server this virtual machine or container was running at this time ?
• Which SQL queries this Oracle instance was performing at this time and what were the values of the associated metrics ?

... etc.

Indeed, it is now possible to retain samples of any attribute of any object of a CI or a group of CI (structured data) and any non-structured data associated to any CI resource (HTML) that will, for example, enrich a dynamic graph legend. For that, you just need to add a specific "tag" in your existing analytics code.

So, you can know everything about any object (device, process, flow, kernel parameter, system resource, ...) of any component of your infrastructures or applications at any time in the past (according to the granularity associated to the sampling).

The additional disk capacity required to store what we could call the "Big Data" of your infrastructures and applications is fully managed and optimized by the fdmon Cloud. Nothing to do from the client side !

Two new features have improved the event management : the event categorization and the hashtag. Default and user defined hashtags are associated to each event, incident and log (SYSLOG), so that you can retrieve specific events directly from the fdmon Search Engine and filter the result according to a specific event or log category (security, compliance, performance, ...).

The fdmon Analytics language (fdmon Analytics as a code) provides new functions to create or update recommendations associated to a CI. Then, from a single click, you can check, in real time, the compliance of your infrastructures and applications against rules defined by the monitored technologies, and get the list of current recommendations (globally or for a specific group of CI).

For each CI, recommendations are considered as a resource on which events (status changes, by definition) will be created, acknowledged and historicized.

Finally, the design of the fdmon GUI also has been improved, especially the visibility of resources statuses and alerts. The weight of HTML pages has been reduced again, to give the best user experience, even on very poor internet connections. fdmon uses ONLY HTML5, still does respect and will respect forever the following rules :

• No Java
• No Flash
• No plugin
• No cookie

And let us remind you some of the elementary and originate specifications of fdmon :

• Fully agentless, whatever the monitored technology, what allows a deployment in a few minutes, whatever the size of your IT infrastructures.
• fdmon Cloud : 100% available thanks to its geo-clustered implementation
• ultra-optimized bandwidth between your Cloud or on-premise IT infrastructures and the fdmon Cloud (1 MB/s for 20'000 servers or equipments)
• your fdmon environment is fully customizable (including Analytics code and graphs aspect) allowing creation of high level and business oriented dashboards trees (executive) as well as deep monitoring technical dashboards trees (for operational or support teams).
• IOT : fdmon provides its own MQTT Broker and monitors all MQTT compliant objects

By the way, happy new year and happy monitoring !