“I think complexity is mostly sort of crummy stuff that is there because it’s too expensive to change the interface.”
In most cases, OSS projects fail because of complexity. Complexity in any number of layers – service complexity, network complexity, political complexity, operational complexity, etc. Or perhaps a combination of many.
One of the most common forms of complexity is in the interfaces between systems (eg devices, EMS, NMS, OSS, BSS, other).
In an earlier blog post, we discussed the chess-board analogy, whereby the strings, pulleys, rubber-bands, etc represent the interfaces between systems. Even OSS with enormous budgets can become unmanoeuvrable due to interface proliferation and complexity.
For a start, there are many different protocols used – SNMP, CORBA, etc. Then there are different data structures for different device types. Then there are different data sets (eg inventory, alarms, performance, provisioning, administration and so on) on each device type.
Would you believe that for one network device I encountered, the inventory, alarm, performance and provisioning interfaces were all based on different protocols because the interface development had been sub-contracted out to different companies by the device manufacturer? They were even developed in different countries!!
It’s no wonder that interfaces are difficult to build, implement and test, adding significant cost, complexity (risk) and time to any OctopOSS project.
Tomorrow I will describe a dream that I have for making interface development many times easier than with the present approach.