Find a clue, solve the case, fix the problem. Getting the truth out of network with Insight Architecture!

16 December 2019

Crime story aficionados are often hung up on one simple question: „who did it?”. And sometimes their gut is right as they’re able to identify the killer halfway or even at the very beginning of the story. However, problems with computer networks usually prove a tougher nut to crack and even a hard boiled engineer may lose his way among false leads and dead ends. Luckily, not every mystery calls for Sherlock Holmes. With Insight Architecture from Extreme Networks your trail will never get cold!

Your typical data center administrator is a law-abiding, well-organized, calm person. He likes his rack cabinets clean and tidy, working like a clockwork. But what does he do when a problem emerges in the middle of the night and the data center is more than 100 kilometers away? Often the network is first to take the blame and the „manhunt” carries on, inevitably reaching other departments, with the client breathing down our neck and constant requests for ETA. Because no one thought that the real issue lies within the server…

From the network administrator’s perspective, Insight Architecture, as the name would suggest, allows for a shorter mean-time-to-innocence, i.e. time it takes to prove that certain problem does not lie within the network infrastructure, but instead is caused by other component – a virtualization, storage or a specific application, for instance. Simultaneously, the tools that are at our disposal thanks to Insight Architecture can help us locate the actual source of the problem and, eventually, eliminate it. So how does it work in practice?

Picking up the trail

Due to the evolution of modern switch technology, the operating system is often run by powerful CPUs from the Intel XEON family. Their massive computing power and additional cores allow us to allocate resources necessary to run Guest VMs. However, Extreme Networks went not one but many steps ahead, while designing its switches. Our devices have several Data Plane interfaces directly connected to the aforementioned virtual machines by 10 or even 25 Gb/s links. This unmediated connection breaks new ground as it allows the properly adjusted VM to monitor or generate and “inject” traffic directly into the Data Plane.

Logical View of Insight Architecture on the SLX Platform

The Control Path between system VM and packet processor is an IP-based link. It is possible to run another VM on each switch. From the administrator’s point of view, it works like a physical server connected directly to one of the access ports on the switch. From here we can generate and receive a high-bandwidth transmission between the packet processor (Data Plane) and Guest VM. The resources themselves are allocated and isolated so as not to disrupt the OS of the switch itself.

Depending on the switch model, we have 2 or 4 CPU cores, 8 GB of RAM and an SSD ranging from 80 to 256 GB at our disposal.

Tools of trade for a network detective

A self-contained, dedicated Analytics Path serves as an additional communication channel. Due to the fact that it remains completely independent from the Control Path, we are able to see clearly what kind of traffic is passing through a specific networking device at any given moment. Such a physical separation allows you to transfer large amounts of traffic with minimal latency, without causing degradation of performance neither in Control nor Data Plane. Thanks to the KVM-based virtualization, the administrator is free to use custom tools and services such as TCDump, Data Dog, Splunk, Remedy, perfSONAR, New Relic, Ansible, Spirent or other applications normally run in the Docker environment. Directly on the device, without deploying additional network infrastructure.

With the Insight Architecture’s help, the administrator can create his own, personal toolbox which provides useful solutions for monitoring and, especially, troubleshooting. But we’re also free to lean towards microservices and run them in the Docker environment, supported by Guest VM. The customer can lay down a custom set of his favorite tools to keep a watchful eye over the DC’s network infrastructure. Even with the finite amount of physical memory, the only limitation is our imagination. In that way our partners may sell Extreme switches like it was their own, original product – all they need to do is set up a combination of their own tools!

Piotr Kędra, Systems Engineer, Extreme Networks Poland

Solving problems… just like you were there!

The single, most important benefit that comes out of using devices based on Insight Architecture is moving the analytics and diagnostics closer to the problem.

I think we will appreciate Insight Architecture the most when we are woken up in the middle of the night, right after the problem has been reported to the server room that is 300 kilometers away. Instead of packing frantically and rushing out the door, right after the call we can analyze the problem calmly, just like we were there. All we need is our little “Swiss Army knife” which will help us make a proper diagnosis in no time.

Piotr Kędra, Systems Engineer, Extreme Networks Poland

Insight Architecture was first introduced in the SLX family of switches. VSP7400 devices brought the technology to the VOSS system, and x465 is the first switch from the EXOS family to utilize Insight Architecture as well. So if you are a data center, campus or access network specialist, it is well worth the effort to get acquainted with the Insight Architecture and the possibilities it offers in terms of performance monitoring, troubleshooting and security.