Explore Our Resources | Enterprise IT & Telecom Blog | Tailwind

Why Some Enterprises are Adopting Leaf-Spine Architecture

Written by TailWind | Oct 29, 2018 6:00:51 AM

The traditional architecture for enterprise networks is typically a three-part tiered, or tree, design. It includes a core as well as distribution and access layers. Advances in network technology like software-defined networking (SDN) are making it possible to adopt a new and more effective network using a leaf-spine architecture.

While the initial implementation cost of leaf-spine architecture may be higher than a tiered design, the priorities of the enterprise may quickly outweigh that concern. From more efficient data flows to an elimination of latency, the enterprise will see immediate improvement in transmissions. Take a deeper look at the benefits:

Data shortcuts

A leaf-spine architecture allows the data to take shortcuts from where it is to its destination. In a three-layer, tiered design, the data flows can differ based on the source and the destination devices. While in some cases, the data may only need to travel across one layer, sometimes the data is required to go all the way to the core before moving out again to its destination. This results in differences related to speed and latency, which can cause problems in an enterprise setting where data is flowing more heavily.

Fully-meshed versus partially-meshed networks

All data in a leaf-spine architecture have the same number of moves before getting to a destination, allowing for consistent speed and the removal of latency concerns. The difference is that tiered networks are only partially-meshed, while a leaf-spine network is fully meshed. While the concern is often cited that a fully-meshed network requires too many physical connections, the larger Ethernet links significantly reduce the number of ports required. This is because with a larger link, more data can flow across a single link.

A loop-free environment

Another key benefit to leaf-spine architecture is the truly active-active data flow, with links within the full-mesh load balanced in a loop-free setting. This type of setup works best when managed within an SDN centralized management model, which allows for configuration, management and re-routing of traffic when there is congestion or another performance problem.

The effects of these benefits are most clearly realized in the data center, where efficiencies in east-to-west flows (or server-to-server flows) are most easily found in data.

New developments in WAN (wide area networking) and cloud connectivity are allowing enterprises to increasingly equip their networks with a leaf-spine architecture.

If your cloud adoption strategies are causing you to examine the network architecture in pursuit of better performance, contact us at TailWind. We offer high-speed broadband connectivity to support your cloud migration and assist you in identifying the right solutions for your network.