Every system that is deployement within an operator:
- Needs to generate alarms for fault management.
- Generates performance management data for software usage, KPI and SLA management.
- Can have interfaces for remote configuration.
- Can be integrated with a central inventory management for simplified asset tracking and audit tracking.
- Generates logs that can be used for auditing.
- Can have an external security system that controls access.
- Can have an external software management system that manages the life-cycle of the software updates.
- Can have a data export for data analytics.
No wonder OSS architectures are complex if in addition to the main functionality all the above need to be managed for every system. Reality is that each system (fault management, performance management, configuration management, inventory management, etc.) has its own standards and expects others to come and integrate with the system.
Can’t this be done simpler?
What if OSS would offer a plug-and-play mechanism? Each new system that is installed would broadcast its availability. Plug-and-play interfaces would be standardized for FM, PM, CM, IM, etc. A plug-and-play interface is a mechanism by which when you add the software, your architecture gets notified of the new system and automatic negotiations happen:
- with the fault management system about what type of traps will be launched and what type of actions need to be taken. Via a Cloud solution, software companies could register once the traps that are launched and the fault management system could “automagically” start monitoring the new system.
- the same is true for the other systems. Performance data configurations are proactively shared with the performance management data. The inventory system receives information about what is happening in the software. The data analytics system receives information about the new system.
- Deployment of the new system should already have happened via a common software management system that also manages updates, high-availability and rollbacks.
Sure there are no standards for this type of semi-intelligence in OSS. However standards in the Web 2.0 era are made by the first dotcom that comes up with a simple system that just does the job.
Software providers would build plug-and-play in their software, hence deployment of a new solution would no longer be traumatic in the OSS area. The idea would be to have a mechanism like a hardware USB plug-and-play. You plug-in a new solution and automatically it configures itself within your architecture…