Data Center

Where Are Your Traffic Flows, and Can You Redirect Them?

by asardell on ‎10-12-2016 11:00 AM - last edited on ‎11-01-2016 04:25 PM by Community Manager (999 Views)

For my twenty or so years in computer networking, the promise of a network that is sensitive to the applications running on it has been put forth almost continuously.  And to some extent, constructs such as Quality of Service (QoS), policy-based routing, and even simple logical domain separation through subnetworks and VLANs have provided the support to improve application performance.

 

As part of the digital transformation, however, applications are becoming much more complex, and traffic growth is increasing tremendously. Data center applications are composed of many components, which creates horizontal traffic (east-west traffic) within the data center. Furthermore, microservices have become more important across application development.

Even as data center networks scale to support this growth, operators of these networks need new tools to predict and adjust to traffic patterns with flows.

 

So questions as to how to:

 

  • Bring the network closer to applications,
  • Understand how applications are using the network, and
  • Identify which applications are of primary interest to customers

Are more critical than ever.

 

What is Needed for Optimized DC Applications in the Digital Era?

 

A key requirement is visibility. A few weeks back, Sanjay Khanna wrote a very popular blog called I Can See (My Network) Clearly Now, which illustrated the value of visibility.  Brocade executives Jason Nolet and Dan DeBacker have taken the discussion a step further in a new video called Optimize Application Performance with Advanced Brocade Network Visibility.  

 

The latter video describes how Brocade worked with many accounts across network, security, and application teams to understand the direction that application development and architecture would take over time.  We discovered that applications have largely become disaggregated with the advent of microservices and componentized, distributed development.

 

Accordingly, customers need to know about their traffic flows, and how much traffic traverses between application components, servers, and other networks. Customers typically do not know this and they can benefit hugely by finding out. 

 

How Do We Gain This Visibility? 

 

A visibility architecture, such as the SLX Insight Architecture (Figure 1), helps operators understand what the nature of the traffic between the components that make up an application.

 

insight.png

 

Figure 1: SLX Insight Architecture

 

The key components and points are as follows:

 

  • A guest VM runs in a KVM environment enabling visibility for network-wide insight
  • A dedicated 10GE analytics path ensures no disruption to network traffic
  • Flexible streaming can be in-band and out-of-band to analytics applications
  • A dedicated SSD storage provides real-time capture for easy and fast access

The visibility data can then be sent to analytics engines of your choice (such as Splunk, New Relic, or Data Dog) for processing. These applications, and others, can also optionally be on-box in the Guest VM, although if a large scale-out is required it may typically make more sense to keep them physically separate.

 

How Can We Optimize Applications with This Information? 

 

Now, if we take this a step further, we can look at Brocade Workflow Composer, powered by StackStorm technology, as a way to add DevOps-style automation to the mix (Figure 2).

 

BWC.png

Figure 2: Brocade Workflow Composer, Powered by StackStorm Technology 

 

In addition to remediation and troubleshooting, Brocade Workflow Composer can used to reconfigure the network through Netconf interfaces to configure IP routing and thus redirect traffic flows.

 

Recommended Links