Wherever you have to update a live network with a software or configuration change, how would you easily test beforehand to ensure seamless transition?
Whenever you need to construct complex network topology involving VNFs from different vendors, how would you easily build one such to validate the proposed features and functionalities?
Data networks require thorough testing before deployment. Many times, devices are tested individually. But how would you plan and verify real world scenarios and events against an entire network, especially at scale?
Routers and switches emit a lot of data. But how would analyze them outside production networks without an easy way to build, collect and analyze in a readily constructible lab like environment?
Open Source free software is both a blessing and a curse. So many great applications are readily available individually, but most not easily manageable collectively. How can you easily integrate them into your simulated networks to make best use of them?
In Camphor Networks, we intend to provide a unified platform to address many of these issues from ground up. You may wonder, how?
Let us start from the cloud computing industry leader technology today, the “Kubernetes”. As quoted directly from kubernetes.io website, “It is an open-source system for automating deployment, scaling and management of containerized applications”. In Camphor Networks, we intended to deliver to our customers a Kubernetes based elastic platform, wherein you can seamlessly manage Virtual Machines themselves as containerized applications. This is especially tricky when the VirtualMachines are as complex as so-called Virtual Network Forwards (VNFs).
What really is a VNF? Basically it is a traditional routing/switching data control and forwarding components all packaged inside virtual machine. Hence the name “Virtual Network Forwarder”. The beauty of VNF is that it can be run typically on any general purpose CPU based architecture without the requirement of any other expensive hardware components such as ASICs, switch fabrics, etc. The VNFs, if given enough CPUs, can perform very well, both in the control plane and in the data plane.
Typically VNFs require a custom environment and tuning. Most VNFs support 12 network interfaces. Some support many more. While some VNFs operate in a single Virtual Machine, many require multiple. Imagine a scenario where you are trying to simulate a network involving 10 VNFs each with 10 interraces. That is a total of 100 interfaces. Let us say they are connected in a topology form with 100s of inter connections. Managing those complex sets of network connections can be very difficult to almost impossible if done so manually.
Now, distribute the VNFs across a kubernetes cluster. While Kubernetes brings in a lot of benefits with regards to the elastic scalability and the management of the underlying compute infrastructure, interconnecting various network interfaces of various VNFs can become quite challenging. In Kubernetes, pods (or containers) can get launched anywhere in the cluster, based on resource availability which is extremely dynamic. Camphor Networks platform hides all this complexity from the user and provides seamless ability to instantiate VNFs directly as native Kubernetes resources (such as pods and containers). The network interconnections among the VNFs’ interfaces are also seamlessly managed, thanks to the SDN provided by the platform. Please note that this SDN is a native SDN provided by Camphor Networks. It runs on top of the Kubernetes infrastructure SDN as provided by most Common Network Interface (CNI) implementers such as Calico and WeaveNet.
Another key aspect of Camphor Networks is commoditizing the creation and management of the Kubernetes based clusters themselves. Not only do we provide an easy way to orchestrate and manage VNFs within a cluster, we also provide a similar way to manage Kubernetes based clusters themselves. Using just a few clicks, users can create Kubernetes based Camphor Network Platforms (Clusters) readily available in minutes. This they can do over many of the popular public clouds available today such as AWS, GCP, Azure, Equinix, etc. The same can be readily destroyed as well, thus breaking the notion of getting tied up with any one particular cloud provider.
By providing Camphor Networks Platform features and the platform itself as a service, we intend to enable our customers to achieve better efficiency in their network operations, specifically in the areas of testing, trouble-shooting, training, telemetry and integration with third party applications.