No abstract

Complete set of management functionality-each client can be associated with one of the ∼90 boot servers, choosing not only a boot image to be loaded but also sets of kernel boot parameters Special operating system configurations are also available for nodes which need to boot special environments (commissioning of new hardware, diagnostics, etc.) Automated post-boot configurations, in particular for advanced configurations of network interfaces, such as bonding and VLANs

No abstract

A configuration database, complemented by a user interface, pushes the monitoring configuration respectively on Special server nodes that act as Nagios servers.  This was necessary in 2007, when the monitoring system was designed, to distribute the work load of the checks and plotting on the servers  An in-house GUI has been developed to collect and display summaries of the check results performed by each of the ~80 servers  ~3000 hosts: ~100000 checks and history One (or very few) Icinga server(s).  In the first tests conducted it seems to cope better than Nagios with the high number of checks performed:  running ~12000 checks on ~1100 hosts on a single monitoring server results in an average latency* below 1 second * the delay between the scheduled start time of the check and its effective reporting time Eventually the data is stored on the MySQL cluster