Organizations are undergoing major digital transformation driven by a massive shift in the application paradigm along with modernization of processes and infrastructure.

Evolution of Microservices and Containers

The digital transformation is driven by rapid evolution in:

  • Applications  - new application architecture and agile approach to develop, deliver, and integrate applications
  • Processes  - more agile processes across both IT and business
  • Infrastructure - modernize existing and build new cloud-based infrastructure

With this evolution, telcos and enterprise IT are moving towards a DevOps model by switching application architectures from existing monolithic/n-tier apps to microservices, which are highly distributed, componentized service-based apps. Containers are emerging as an alternative to existing virtualization technologies to enable DevOps and microservices. With a packaging mechanism that typically includes an application and all of its runtime dependencies, including lightweight Linux, containers are easy to deploy and are readily portable across different host systems. Compared to traditional VMs, containers are light-weight and highly elastic, with the ability to be rapidly spun-up or spun-down in sub-seconds and hence have much higher density per host.

Red Hat OpenShift

The Red Hat OpenShift platform includes a container-optimized, minimal footprint OS powered by Red Hat Enterprise Linux, standardized container runtime from Docker, Kubernetes-driven orchestration of multi-container environment, and services such as scheduling, logging and analytics for container management. The OpenShift platform helps accelerate application delivery and increase operational efficiency by providing to developers a platform for self-service, auto-scaling and collaboration, and an enterprise-grade, secure, standards-based and scalable platform to operations teams.

Networking Requirements for Containers

The distributed and dynamic nature of container deployments dramatically increases the volume of east-west traffic (intra-DC traffic between container endpoints), even more than VM-based traffic. While all the typical networking constructs apply (i.e. the physical network connecting servers running container apps, virtual network which resides inside the server hosts and services such as load balancers and security), there are three unique network requirements imposed by container environments:

  • Zero-touch, Automated Networking: As containers are rapidly spun up or down, the network needs to adapt to that rate of change. Static and cumbersome configuration of networks will not work in this environment; what is really needed is a very simple, zero-touch networking environment enabled by configuration automation.
  • Visibility and Troubleshooting: When troubleshooting issues in such a dynamic deployment you will need a way to rapidly identify if an issue is with the physical or virtual network or with the containers/applications. This requires complete end-to-end visibility of traffic flow in the environment across both virtual and physical networking domains.
  • Application-driven Model: Application teams need an application-driven logical model for networking, a uniform model that applies across not only container workloads but also VM-based and bare metal apps in the environment.

Overlay/Underlay vs Unified Approach for Networking

Typical container networking solutions leverage an overlay/underlay approach in which two separate networks are managed by separate teams, and there exists an overlay/underlay visibility gap. With a unified physical + virtual approach, the entire network is managed as a single entity with the same (normal) networking constructs and isolation mechanisms, and you also get the unique advantage of single point of visibility into both environments. While Big Cloud Fabric (BCF) can be deployed as physical-only fabric with an overlay, complete operational benefits and end-to-end visibility can be realized with the BCF unified physical + virtual (P+V) fabric solution.

BCF / OpenShift - Ideal Unified Networking Solution for Containers

The Big Cloud Fabric / Red Hat Openshift solution offers an integrated and unified physical and virtual networking approach for container environments. The two key components which glue the BCF to Red Hat OpenShift are the BCF container orchestration agent and the BCF container networking plugin (CNI-based). The integration of Big Cloud Fabric with Red Hat OpenShift enables simplification and agility of container networking through network automation, deep visibility, and rapid container-to-container troubleshooting.

 

Figure 1: Red Hat OpenShift and Big Cloud Fabric Solution

The Big Cloud Fabric is comprised of a physical leaf-spine Clos fabric of open networking switches running Big Switch Switch Light OS and a virtual network with Big Switch Switch Light VX running on the hosts. A centralized SDN controller acts as a single pane of glass to control the entire physical and virtual network environment.

  • Auto host detection: Hosts running containers, which are connected to the leaf switches, are automatically detected, links are aggregated automatically and the entire fabric is stood up in a zero-touch fashion.
  • Auto scale-out: The fabric scale-out is automated – when new hosts/racks are added, there is no configuration required.
  • Single Pane of Glass: The BCF controller acts as a single point of integration with the Red Hat OpenShift orchestrator. The container network automation happens with an agent that runs on the controller and a CNI-based container plugin that runs on the hosts. As a result, when a container is spawned, the network configuration is automatically taken care of across both the virtual and physical switches.
  • Deep Visibility: The controller-based architecture enables unique visibility and troubleshooting benefits for the network admins as they get a view of all the container endpoints and one-click end-to-end path visibility across both the virtual and physical switches from one container to another. (Figure 2 provides an illustration of this feature).

Figure 2: Single-Click End-to-End Path Visibility with Big Cloud Fabric/OpenShift

In summary, the workflows orchestrated by Red Hat OpenShift with Big Cloud Fabric P+V solution dramatically simplifies network operations by enabling provisioning, management, and troubleshooting of networks at the speed of containers. With a unified approach, it completely eliminates network as a bottleneck and helps application teams realize the true promise of agility of container environment.

To learn more about the solution including a live demo, you can view the recording of a joint webinar by Red Hat and Big Switch Networks: Here

Watch our recent on-demand webinar with Red Hat which addresses best-in-class OpenStack and container networking offerings from Big Switch and Red Hat. Go to: https://vts.inxpo.com/Launch/Event.htm?ShowKey=41381

Bala Ramachandran
Director of Product Management
Solutions & Technology Partnerships