Multicast: How to Overcome the Hidden Problem Waiting to Stall Your Cloud Migration

The Problem:

The industrial IOT (IIOT) industry is built on the ability to connect to and learn from data producing assets. Cloud networks serve the IIOT industry and the billions of associated devices as the current innovation frontier, underpinned by expanding computational performance and networking flexibility. The combination of IIOT technologies and cloud networks is helping companies perform better, innovate faster, create new business models, and jockey for market leadership positions. But here's the thing: not all IIoT workloads can be handled by the cloud, which means some will continue to be relegated to on-premises networks.

Multicast Is the Culprit

Multicast is the biggest roadblock for moving on-prem networks to the cloud. It’s a routing protocol that’s used in highly optimized on-prem systems, and it’s essential to many low-level industrial protocols and their middleware or third-party software. With multicast, you can deliver data packets from one producer to many endpoints at once, or, conversely, packets from many producers to one endpoint. The inconvenient reality, however, is that once a multicast transmission is distributed from an on-prem network to a cloud network, it’s dropped! That’s because cloud networks don’t natively support multicast.

Worse, to the uninitiated system integrators or network engineers, you might not even be aware of this migraine-inducing problem. When engineers do realize what’s happening, they typically scramble to find viable solutions that won’t mess up years of painstaking network customization, built-in dependencies, and third party add-ins. Unfortunately, there’s no easy fix and next-best alternatives stray from “lift and shift” efforts, like rigorous recoding efforts or overly administrative tech re-sourcing efforts.

Multicast in the IIoT World

Multicast is popular in the IIOT industry for many high-performance systems that rely on communicating with distributed endpoints, while minimizing latency and conserving bandwidth. In addition to supporting lots of established workloads, multicast has renewed importance to network architects as teams, databases, and devices become increasingly more remote and distributed, and operators search for improvements in cost and performance.  

One multicast use-case driven by secular trends is the distribution of data for parallel processing of independent tasks across multiple GPU/CPU cores. To illustrate this example, the same raw data would be sent to multiple cores for things like calculating probabilistic AI scenarios, rendering graphic-intensive workloads for different perspectives (L-eye, R-eye), or running CAD design verification. In these examples, workloads distributed to each core are independent of each other and through the parallelizing of the workflow, achieve performance and cost improvements. Parallel computing has uncapped potential across many IIOT applications beyond our examples, including, medical imaging analysis, drug research, oil-well monitoring, and more.

Enter cloudSwXtch to Enable Multicast in Cloud Networks

cloudSwXtch is a virtual switch that adds features to cloud networks for seamless transmission of traffic across hybrid and cross-cloud environments, in ways not previously possible. And, among other benefits, it addresses the issue of multicast not being supported in public cloud networks.

Top features, include:

1.      Multicast and broadcast routing that address the gap in cloud networks for these protocols.

2.      Protocol fanout to distribute unicast streams in the cloud as multicast for synchronized communication to end-points.

3.      High availability via dual path with seamless merge on a packet-by-packet basis.

4.      High-performance with the ability to accommodate +1M packets per second at sub-Ns latency.

5.      Dynamic bridge to cycle between all data-streams and select the relevant ones to send from on-prem to cloud, to optimize ingress/egress costs.

‍

cloudSwXtch is easily deployable as a VM image that creates an overlay network sitting on top of public cloud networks and other systems. Through the addition of networking features, cloudSwXtch normalizes network behavior of clouds and makes workloads look and feel the same in clouds as they do on-prem.