Written by Jim Duffy
Jim Duffy is a Senior Analyst for the Networking Channel at 451 Research. He covers enterprise network infrastructure and associated software, and network performance management. Jim has been covering technology for over 30 years, including 25 at Network World. His coverage focused predominantly on enterprise networking infrastructure, including routers, switches, and associated software.

The adoption of software-defined networking and disaggregation of switch hardware and operating system software promises much in the way of simplicity, automation, orchestration, rapid provisioning and customer choice. Indeed, recent surveys conducted by 451 Research indicate that a chief driver of software enablement in networking is faster response to business needs, followed by improved availability and reliability, decreased cost and enhanced security.

Software-driven networks breed complexity

But for all of the operational ease that the software-driven disaggregation promises, there’s still some complexity inherent in scaling out the technology across the enterprise. This is readily evident in one particular networking discipline: visibility.

Creating a visibility infrastructure for a large-scale network can be just as complex as the network itself. Depending on the size of the network, the visibility infrastructure may require hundreds of packet brokers and switches to instrument the network for monitoring, management and insight into performance, and to forward packets to the appropriate analysis and security tools when necessary.

The need for visibility fabric management

Just as a large-scale network can benefit from a fabric-based architecture of finely orchestrated elements, so too can a large-scale visibility infrastructure. In an ideal world, by weaving packet brokers into a software-driven fabric, individual or groups of packet brokers could be deployed and scaled rapidly in just a few mouse clicks from the GUI of a single central management station.

This central fabric manager would easily organize complex resources and configurations – such as switches, ports and filters – provide intuitive configuration of traffic flow topologies, and allow for deployment of new switches or topology designs in minutes.  It would provide a user--based easy and natural logical grouping of the major functions in building and operating the visibility infrastructure: configuration, deployment and monitoring. This manager would allow for simple definition and deployment of filters, topologies, topology changes, and granular monitoring of the status of physical components and their attributes, and of traffic flows.

Key requirements

Consider the following questions when planning future visibility deployments:

  • How would the fabric management system help you with triage and troubleshooting?
  • What will the time to deployment be?
  • What is the process of creating the network visibility topologies?
  • How would the packet broker infrastructure scale?

In short, the fabric manager should drive out operational complexity and reduce time so that visibility into the network itself can be achieved in short order. The same uniform workflow to operate a large-scale, software-driven network can also benefit a large-scale visibility infrastructure comprised of hundreds of disaggregated, software-driven packet brokers. This is the power of software. And it’s not just for networks anymore.

Applying the same principles of network disaggregation to network visibility should reap the same benefits – simplicity, automation, orchestration, rapid provisioning and customer choice. This, in turn, should produce the same results: faster response to business needs, improved availability and reliability, decreased cost and enhanced security. The more agile and responsive the visibility infrastructure, the more agile, responsive and secure the network. And the benefits to the overall business are obvious.

~ Jim Duffy, Senior Analyst for the Networking Channel, 451 Research

Subscribe to Our Blog