Pluribus Networks will be highlighting how our Netvisor® OS and the Adaptive Cloud Fabric™ interoperates with VMware® NSX to simplify the creation of VXLAN tunnels on whitebox open networking switches at VMworld 2017 in Las Vegas. Our integration and interoperability with VMware is important for environments that have virtualized, bare-metal, and containerized workloads for the following reasons:
- Most customers can’t create a greenfield architecture, and have to deal with a substantial amount of legacy architectures while pieces of the stack are virtualized and containerized
- Despite these complexities, administrators want to have uniform connectivity, and using VXLAN-overlays over a Layer 3 BGP/ECMP is a logical approach to address this need
- Server connectivity is quickly moving to 25Gbps and software-based tunnels can be very expensive and represent heavy I/O workloads
- Combining server-based software overlays with switch-based hardware overlays, while using NSX and vSphere simplifies operations and lowers both CAPEX and OPEX
Top-of-Rack (ToR) switches running the Netvisor OS can perform VXLAN offload at line rates of 25 and 100G, without adding any unwanted CPU loads or additional latency to the flow. This is a big requirement in latency-and cost-sensitive Enterprise and Telco cloud environments where the Netvisor OS already has many production deployments.
Advantages of Netvisor OS and the Adaptive Cloud Fabric coupled with VMware NSX
Most silicon used in ToR switches today can perform VXLAN encapsulation and decapsulation at line rate, without adding latency to the flows. The NSX system can terminate software tunnels on Netvisor OS enabled hardware switches (see figure below) through a hardware accelerated VTEP. The Netvisor OS and the Adaptive Cloud Fabric provides a fabric-wide view, and automates tunnel orchestration between these switches. Netvisor OS uses the distributed vPort table to automatically map destination flows to tunnels, further simplifying and scaling the operation.
For bare metal and containerized workloads, tunnels originate and terminate on a Netvisor enabled ToR switch, creating an overlay over the layer 3 network, or a network stretching across a WAN boundary. The Netvisor OS provides the HW-VTEP L2-Gateway for virtualized workloads and terminates and optimizes the NSX tunnels. In the figure below, each leaf switch pair represents a server rack where the first 2 racks and are fully virtualized to support the software overlays orchestrated by NSX controller while 3rd and 4th rack are containerized and bare-metal respectively in a full HA configuration. In this example, NSX seamlessly orchestrates software tunnels terminating on the Netvisor OS to allow virtualized workload to communicate with bare-metal and containerized workloads. Also, the Netvisor OS automatically creates tunnels and assignes flows to specific tunnel when containerized and bare metal workloads need to communicate.
All this works seamlessly and is orchestrated through vCenter minimizing the need for additional network configurations. The technical and business advantages are:
- The Netvisor OS automates tunnel orchestration between switches and uses its distributed vPort database to assign flows to tunnels without any flood and learn or needing other tools
- The Netvisor OS supports Anycast Gateway and distributed routing that helps avoids east-west hair-pinning in a centralized router environment that interoperates with NSX distributed logical router for virtualized workloads
- The Adaptive Cloud Fabric is automatically provisioned based on vSAN cluster configuration, and simplifies VXLAN implementation, enabling the combination of VSAN and NSX without requiring a complex Multicast design. Since the Netvisor OS supports encapsulated VSAN traffic over a VXLAN segment, it provides a simple layer 2 domain for all vSAN hosts, eliminating the need to design a layer 3 Multicast network specifically to support vSAN implementations
- Containerized and bare metal workloads can co-exist with virtualized workloads seamlessly in an overlay plus layer 3 topology
- Integration between NSX and Netvisor OS reduces human touch points, simplifies deployments, speeding time to deployment and minimizing the chance of configuration errors
- The embedded telemetry of the Netvisor OS provides visibility into every TCP connection and flow inside the NSX tunnel offering full visibility in an overlay and non-overlay environment – Pluribus Insight Analytics™ provides a single-pane-of-glass performance view and eliminates the need for expensive hardware probes
- The Netvisor OS is certified and available on Dell’s Open Networking Switch portfolio, so customers can implement a turn-key solution that integrates the open switch hardware, the Netvisor OS, VMware solutions, servers and storage to reduce operational complexity and speed time-to-deployment
Netvisor and vCenter Integration
The Netvisor vPorts are also deeply integrated with vCenter to facilitate fabric-wide intelligence and improve performance. Key capabilities include:
- VMware ESXi hosts and VMs are detected and automatically configured based on the ESXi network configuration
- vCenter extends the capability of individual vPort based on VMware attributes that become usable and searchable keywords in the Pluribus Insight Analytics performance management platform
- The Adaptive Cloud Fabric also has a vCenter plugin which offers a comprehensive GUI tool for the virtualization admin to get insights on network infrastructure, with role-based access to avoid administrative domain errors
Come see Pluribus Networks at VMworld 2017 in booth 821 to learn more about what we are doing with VMware.
About the Author
Sunay is the CTO and a Co-Founder of Pluribus Networks. Prior to Pluribus, Sunay was a Senior Distinguished Engineer for Sun Microsystems, and was the Chief Architect for Kernel/Network Virtualization in Core Solaris OS. Sunay has an extensive 20+ year software background, and was one of the top code contributors to Solaris. Sunay holds over 50 patents encompassing network and server virtualization.