Network design is not a simple process, unless of course you have only 2 -3 nodes😃. There are a variety of design options for the bigger, more complex networks. Two basic options exist: Ring design or a mesh design, each with a set of pros and cons. So how should you go about designing your network?
Rings are normally the cheapest way to provide resilient connectivity between a set of nodes because they minimise physical connectivity. However all the protected traffic passes through every node, meaning each node needs to be able to handle the full traffic load even though most of the traffic is through traffic. In addition, as traffic is added, the ring rapidly fills.
Mesh networks require considerably more physical connectivity, which makes the network CAPEX heavy. However, once built, nodes can be sized for the traffic they actually process, since through traffic in a full mesh network can be close to zero. As traffic is added only the nodes handling the new traffic need expanding.
The above seems to be common sense. So should you build a ring or a mesh for your needs?
I have had multiple opportunities to analyse this exact question in detail. Looking into the different costs associated with the different designs: dig costs, node count, cost of nodes and traffic distributions.
With respect to traffic distributions, I considered two extreme cases:
With my assumption being that a ‘real’ network would likely see a mix of mesh and head-end type traffic distributions.
After my analysis, I was surprised to realize that, when the number of client services carried are equalized, there is no difference in the network cost for a ring or full mesh design. Thus, I do not need to worry about traffic distribution when considering ring vs. mesh.
Plotting network cost against traffic always resulted in the ring scenario starting low and rising fast, whilst the mesh scenario started high but costs rose more slowly. There was always a point where the ring network cost crossed the mesh network cost. The crossing point depended on distance between nodes (dig and cable cost), as well as the equipment cost.
The simple conclusion is rings are most cost effective to start with while mesh networks are more cost effective at higher traffic levels. The crossover between ring and mesh depends on the physical topology and associated infrastructure build costs.
From this analysis, I would expect to see newer networks are more ring-based, whilst mature networks tend towards mesh – and in fact this is what I observe in practice.
The ideal scenario seems to be one where an operator starts with a ring and evolves into a mesh as traffic grows, adding more connectivity gradually to minimize node through traffic. For large networks, different parts will evolve at different speeds. This approach minimizes investment at each stage.
To support evolution from ring to mesh the nodes need to efficiently support a rising number of line interfaces. This means choosing network elements built to flexibly support evolving line and client interfaces from day one. Building a network with such flexible elements leaves the operator more choices in the future. And more choice means that you are better positioned to react to market changes.
Operators can use a variety of network analysis tools to determine when additional connectivity will provide a more cost effective growth strategy than additional equipment.
ECI's optical and packet transport solutions fully support such an Elastic approach. Our Apollo product family supports: point-to-point, fixed and reconfigurable multiplexing for wavelengths and OTN; interface speeds from 100Mbit/s to 600Gbit/s; full Flexible Grid spectrum; and complete Alien wavelength control.
Our Neptune packet transport family offers: a wide range of: interfaces (sub E1 to 200Gbit/s); matrix (5Gbit/s to 16Tbit/s); and control (Ethernet, MPLS-TP, IP/MPLS and Segment Routing) with in-service upgrade options for each element.
We can help optimize your network for initial deployment, support your strategic growth plans and ensure you are elastic enough to adapt to your rapidly evolving market.
Dr. Antony Thorley C.Eng., FIET has over 35 years of telecommunications experience with vendors, operators and regulators, working on reviews of the business connectivity market. Currently Tony supports pre-sales technical work supporting the UK sales team and the UK market for ECI on DWDM, OTN, packet and network management for customers in the service provider, utility and data centre markets. He started his career with optical component specification and evaluation including early work on coherent transmission and optical amplifiers. Then moved to circuit design, sub-system design, system design and analogue ASIC development. His career has included over time: marketing, customer engineering, network strategy and regulation.