The top-down, bottom-up design process

When planning out a full-stack business / network / services management solution, I tend to follow the top-down, bottom-up design process.

Let’s take the TMN pyramid as a starting point:
TMN Pyramid
Image courtesy of www.researchgate.net

Bottom-up: When designing the assurance stream (eg alarms, performance, etc), I start at the bottom (Network Elements), understanding what devices exist in the network and what events / notifications / monitors they will issue. From there, I seek to understand what tool/s will manage them (Element Management / Network Management), then keep climbing the stack to understand how to present key information that impacts services and the business.

Top-down: When designing the fulfilment stream (eg designs, provisioning, moves/adds/changes, configuration, etc), I generally start at the top* (Business Management), what services are being offered (Service Management) and figure out how those workflows propagate down into the network and associated tools (such as ticketing, workforce management, service order provisioning, etc).

This helps to build a conceptual architecture (ie a layered architecture with a set of building blocks, where the functional building blocks start out blank). From the conceptual architecture, we can then identify the tools / people / processes that will deliver on the functions within each blank box.

This approach ensures we have the big picture in mind before getting bogged down into the minutiae of integrations, data flows, configurations of tools, etc.

To get momentum quickly, I tend to start with the bottom-up side as data flows (eg SNMP traps) are more standardised and the tools tend to need less configuration to get some (but not all) raw alarms / traps into management tools, perhaps just sand-pit versions. For the next step, the top-down part, I tend to create just one simple service scenario, and perhaps even design the front-end to use The Mechanical Turk model described yesterday, and follow the flow manually down through the stack and into element management or network layers. Then grow both streams from there!

* Note that I’m assuming services are already flowing through the network under management and/or the team has already figured out the services they’re offering. In a completely green-fields situation, the capabilities of the network might determine the product set that can be offered to customers (bottom-up), but generally it will be top-down.

Read the Passionate About OSS Blog for more or Subscribe to the Passionate About OSS Blog by Email

Leave a Reply

Your email address will not be published. Required fields are marked *