The blockage-free Spine-Leaf architecture goes Europe!
Today we would like to briefly introduce the concept of Spine-Leaf Architecture. So, what does Spine-Leaf actually mean? Imagine a tree with a stem and leaves. So, this architecture describes the spine of a flat and high-performance 100/200/400G network connection and is based on fully crossed mutual connections of all switches in the data center. It takes advantage of the idea of a blockage-free network because every leaf switch is connected to every spine switch. Such a fully meshed network between the spine and leaf switches is also referred to as "fabric" or "mesh layer". This type of network structure is currently gaining ground in Europe, as the massive increase in data volumes due to virtualization technologies within data centers, among other things, will have to be routed in the future with the help of so-called East-West traffic, without the switches becoming the bottleneck in terms of network speed.
Figure 1: Spine-Leaf Architecture WITHOUT CrossCon®
The advantages of the blockage-free spine-leaf architecture, which requires only one hop from one switch to the other, are therefore obvious:
Figure 2: Spine-Leaf Architecture WITH CrossCon®The CrossCon® is an important building block of the technical spine of our future digital society making it possible to encapsulate the sometimes highly complex cross-connections of the Spine-Leaf Architecture in one device. This means massive savings on individual connections and tranceivers, with the extra advantage of parallel physical redundancy. Would you like to learn more about the CrossCon® in a Spine-Leaf Architecture? Just contact us! We are happy to help you plan your new Spine-Leaf-Project.Sources:https://www.datacenter-insider.de/was-ist-eine-spine-leaf-architektur-a-697000/https://osi.rosenberger.com/de/news-media/herausforderungen-verkabelung-rechenzentren/
- The data path is randomly selected in the fully meshed network.
- Traffic is evenly distributed in the mesh layer.
- In case of impending overload of an existing link from a leaf switch to the spine switch, another path to another spine switch can be used in parallel.
- The traffic between leaf switches always runs over one hop maximum (spine) and, thus, has fast and, above all, predictable latency, which are the prerequisites for increasingly important real-time applications.
- The expandability of the network with the help of additional leaf and, if necessary, additional spine switches ensures demand-based maintenance of blockage-free Spine-Leaf Architecture.