Big Switch Networks Expand Their Reach, Now with HCI Integration

Big Switch Networks, the Santa Clara–based software-defined networking company, has just released a new version of the Big Cloud Fabric product. Big Cloud Fabric, a software-defined networking product that has been on the market for over four years, is heavily integrated into VMware. For the uninitiated, its core pitch is that with its product, you can cut out proprietary networking gear, and that by using its software-based controller, coupled with low-cost white-box servers and switches, networks can be provisioned, orchestrated, and configured programmatically.

Out of the box, it has many advanced features. Unlike NSX, it has a real physical presence. Unlike ACI, it has a real virtual presence. It plays nicely with both. Its data layer can be deployed on Open Networking Dell EMC Edgecore white boxes and the HPE Altoline family of equipment. Its Big Monitoring Fabric product is a Womble product; it monitors “overlay, underlay—so your packets roam free.”

Role-based access can give VM admins and storage admins the ability to push VMs directly on the network. Yes, you can do this with other products, but there are no Band-Aids™ or shoehorning of square pegs into round holes.

BSN has been quietly building up a veritable powerhouse of an SDN product. Interoperability with Citrix XenServer, VMware ESX and NSX, OpenStack, KVM, CloudStack, and many other infrastructure IaaS and PaaS platforms have been built up over the four version releases. However, there has been a small gap in its product set, that being true integration into HCI platforms. Yes, it has had integration with VSAN since the release of version 4.0, but now it has extended that reach into Nutanix, SimpliVity, and Dell EMC ScaleIO.

It does this by providing:

  • High east-west traffic across the server/storage nodes with its high-performance 10G/25G/40G/100G scale-out fabric.
  • Scale-out fabric, which can incrementally provision additional nodes of capacity and performance as the application demands increase.
  • Automated network provisioning of HCI hosts and VMs, which removes bottlenecks typically experienced with proprietary network designs caused by manual, switch-by-switch architectures, based on traditional CLI commands, which impede the desired software-based agility of modern data center operations.
  • HCI-related visibility for network admins to simplify capacity planning and change management and enable rapid troubleshooting.

BCF now has validated design for Nutanix Ready Core for networking status. One thing to note here is that when you deploy BCF on Nutanix with vSphere ESXi as the hypervisor, you gain a fully featured integrated product. The BCF on Nutanix with AHV, although fully supported, is more limited in features.

It has further solidified its relationship with VMware and Dell, having just delivered a validated reference architecture for a VMware SDDC based on BCF, ScaleIO, vSphere, and Dell EMC Open Networking hardware. This is based on its deployment to Tianjin Broadcast and TV Network in China. Jia Zhen, IT Manager at Tianjin, stated that “Big Cloud Fabric was significantly easier and faster to deploy, when compared to [their] prior data center network solution and it has provided ongoing operational efficiencies due to the superior technical architecture. Additionally we have experienced zero network configuration errors since deploying the solution.”

The move into HCI at first seems to be a little bit of a “well, nobody else does it” move, but it actually is a sensible move, especially as HCI vendors increasingly move away from their pure position of one size fits all. VSAN and Nutanix datastores can be presented to nodes that are not participating in the storage cluster, and Nutanix now has storage-only nodes. This breaks its data-locality benefits and introduces latency-reducing performance. But north-south storage traffic across switches is guaranteed to be slower than local storage connected on the local PCI bus. Sorry folks, it’s just physics.

The introduction of an SDN solution into an HCI cluster will not reduce that latency. However, it will improve manageability, scalability, and speed of deployment, as all newly deployed devices are presented to a single logical software-defined switch, regardless of the location or model of the physical network switching device. Jongan Park, network head at Incheon Smart-City, stated that “The software-defined data center is an ideal option for designing a high-tech city, but to achieve the maximum flexibility, scalability and automation promised from SDDC environments, the network, at the core of the SDDC is crucial.” This makes absolute sense. Data is at the core of every company’s crown jewels, and the network is the transportation system. The question is whether you want to drive a modern car on cart tracks or open freeways?