In this DemoFriday, Brocade’s Robert McBride and Vikram Singh discussed how to simplify vCPE automation and management with open source solutions like OpenStack Tacker and Brocade vRouter. This DemoFriday covered why the automation of the lifecycle management of your virtual network functions (VNFs) for on-premises vCPE is crucial to the success of your deployment and why performance, scale, and density should be a top priority in your VNF decision. Brocade also discussed how it’s working within the open source community to help industry make network functions virtualization (NFV) deployable and automated. After the live presentation, our presenters took questions from the audience. Read the full Q&A below!
Brocade: Yes, Tacker can manage and handle Multi-VM VNFs – a great example of this is the demonstration that Brocade did with the Brocade vEPC.
Does Tacker support SRIOV and policy placement for VNFs?
Brocade: Get up to date information on the Tacker project within the OpenStack community by visiting the Wiki page. Specifically, the capability to utilize SRIOV and implement policy placement for VNFs is currently targeted for the Mitaka release of the Tacker project.
What kind of monitoring capabilities are available for Tacker?
Brocade: Tacker itself provides a VNF KPI Health Monitoring Framework, which provides a pluggable driver to monitor various state and status conditions of the underlying VNFs it is managing and deploying. This health check begins when the VNF becomes in a ready state with continuous network connectivity health checks and an auto-start on failure mechanism is also available based on the VNFD policy. Additionally as well, Tacker can support custom vendor specific monitoring mechanisms through the monitoring framework.
How does the Brocade vRouter provide resource utilization efficiency?
Brocade: The way Brocade vRouter is architected, it is one of the leanest in utilization of compute resources, especially CPU threads. It offers the highest throughput for a given vCPE resource. Comparing performance should also consider PPS, not just throughput. The vRouter resource efficiency also means it can deliver the same bandwidth at much smaller packet sizes. This is the result of architecting vRouter data plane from ground up, using DPDK, for virtualization. One key advantage Brocade vRouter has is that it doesn’t have the legacy of data plane written for proprietary ASICs and doesn’t have to emulate those ASICS.
What are you seeing in terms of deployments? Are you seeing folks that want to have one large vCPE router with different customers sharing it, or do you see that people want a separate VNF for a vCPE?
Brocade: Brocade is seeing a number of different deployments of the Brocade vRouter related to premises or cloud based vCPE. In general there is a hybrid of large, small and simple.
What were the gaps that you saw that made you realize you needed to create Tracker? What is the specific problem you’re trying to solve with it?
Brocade: There are multiple orchestration platforms and solutions available today on the market, and most VNF managers are proprietary and manage vendor specific VNFs. Tacker being an open source project within OpenStack fits into an opportunity to present a path of no vendor lock in and a solution that is multi-tenant aware and common VNF Manager for all VNFs.