Editor's note: Ixia is the winner of the May SearchNetworking Network Innovation Award for its IxNetwork 7.0 network topology and testing assessment platform. The product, released last year, permits administrators to emulate how various applications and their respective protocols might affect network performance prior to their introduction.
SearchNetworking site editor Chuck Moozakis spoke to Michael Haugh, Ixia's senior manager, market development, about the metamorphosis of IxNetwork 7.0.
What was the genesis behind IxNetwork 7.0?
Michael Haugh: Ultimately, IxNetwork had become our single product [assessment] framework that allowed us to add in protocols that spanned all of the layer 2 and layer 3 market segments, from broadband to data centers and from routing and switching to core MPLS and carrier Ethernet, as well as timing protocols that played into mobile backhaul. Even five years ago, it was a story of adding more protocols into the framework.
A few years ago, we redesigned the data plane and, with that, introduced ViperCore. We changed the front-end, we changed the middleware and we changed the underlying firmware. The big buzz then was QoS. How do you handle the prioritization of traffic and having multiple levels of service? That ultimately is what ViperCore enabled [the assessment and consequential impact of multiple traffic flows].
IxNetwork 7.0 is the end vision of what we are trying to get to with the product. As I mentioned, over the years we kept adding new protocols and, ultimately, we had two different workflows. We had a workflow for routing and switching and a different protocol for broadband access. When those sets of protocols were migrated in, they were migrated in under a different framework and different code path. While it was good to unify those protocols under a common interface, you had test cases when you needed a broadband protocol, like point-to-point protocol and you wanted to run it under MPLS, and we were challenged to do that.
How did 7.0 overcome those limitations?
Haugh: IxNetwork 7.0 introduced what we refer to as our common protocol framework. With CPF, all of the protocols are migrated into this framework. It provides consistent workflows as you move from one protocol to the next, and it provides the ability for protocol stacking. Any protocol within CPF allows you to stack one protocol over the other.
The product also introduced a scenarios view within the graphical user interface (GUI). And under that, users see a graphical display of ports and they see a graphical display of the stack itself. The stack is color-coded to measure the health of the stack. So it gives customers a visual topology of the network --everything from the physical port, the emulated router, the protocol stack running and the emulated posts or traffic tied to that. Customers wanted tools for more complicated testing and protocol stacking, but they also wanted to do that through a nicer user experience.
What were some of the challenges Ixia faced when upgrading IxNetwork to 7.0?
Haugh: First, it was to determine which protocols to be migrated into the new framework and, second, to determine what the most demanding [emulation] scenarios to support were. In addition, we had new protocols to consider. For example, OpenFlow. We placed it in the legacy protocol model, and the reason was a time-to-market issue. It would have been ideal to put it into CPF, but we wouldn't have been able to do that for six to eight months.
The other challenge was to keep the legacy code path and workflow completely intact, to permit users to have backwards capability and the ability to configure through the legacy mode.
What are IT managers saying about the strategies they need to employ when emulation testing?
Haugh: In the layer 2 and layer 3 space, some of the hottest things we've been looking at include protocol stacking complexity. Manufacturers are providing a lot of features and functions in their devices. They are consolidating the network, they are offering a variety of services, they are connecting their mobile backhaul networks. So you are adding to the protocol soup, and managers have to try to ensure they have the scale and performance needed.
At the same time, operators and service providers are dealing with the proliferation you are seeing with social media and BitTorrent and all these other challenges. We've added the App Library (Editor's note: the library is included within Ixia's release of IxNetwork 7.10), so you can simulate the effect users might have on a network as they access BitTorrent, Facebook, Google [and other social media sites].
This is certainly an area of interest where they are distributing policy all the way down to edge, even within the broadband space.
senior manager, Ixia
What our customers are trying to test is to ensure that their policies for dealing with traffic are applicable to the applications that are stressing their networks, perhaps rate-limiting for BitTorrent, or for certain schools and companies in certain hours blocking access to Facebook. They are becoming more policy oriented in the way they are treating explicit traffic and applications.
This is certainly an area of interest where they are distributing policy all the way down to edge, even within the broadband space. They want to make sure that they emulate these protocols and understand [how these applications] affect performance.
How is virtualization affecting emulation and testing?
Haugh: There has been an explosion of server virtualization, and we've been well-positioned [to serve that market] with IxVM. It's basically a port that has been virtualized and can be installed as a virtual machine. It extends testing from layer 2 through layer 7 and right into virtual machines so we can test the switches, virtual devices and other components.
What trends are you seeing in the data center?
Haugh: Data centers have been a hot area. Carriers are building out data centers; large enterprises are really looking to what I will jokingly call "kill the spanning tree." So what are they moving to next? Do they move to a layer 2 protocol like transparent interconnection of lots of links (TRILL)? Or shortest-path bridging? Or do they move to layer 3 in the data center? Or could they potentially move to something new, and a lot of them are doing early evaluations of protocols like OpenFlow. As these protocols are developed and introduced, we are adding these protocols and helping customers test in these spaces.
How has IxNetwork evolved in order to emulate all of these protocols?
Haugh: It's optimization at a hardware level, where we've migrated the architecture. We've gone from single CPU and single field-programmable gate array (FPGA) per port to an architecture where we still have FPGA, which allows us to deal with high level flows and measurements in real time by leveraging that FPGA. Now we are leveraging several multicore CPUs and we can use the horsepower we need as we get to higher scales or running higher app traffic. We can be more flexible in terms of addressing the scale and need. The other challenge always facing us is keeping up with equipment manufacturers in terms of density.
What are some of the key areas Ixia is examining for future development?
Haugh: We've expanded from layer 2 and layer 3 to layer 7. With the acquisition of Breaking Point Systems [purchased in 2012], we are addressing a key point that customers are engaging us about, which is security. We have a portfolio in the security space. There is also significant need and requirement in the Wi-Fi space to ensure that applications are going to work over Wi-Fi. Through the Anue acquisition, we have essentially a monitoring switch, and that allows customers to optimize the tools they are using to monitor the network.
What we're learning from the monitor side we are feeding back into our emulation [product]. Another key area or initiative for us is to expand in the enterprise. To expand in the enterprise, you need a much more user-friendly user interface and it needs to be a lot more canned, so what we've rebuilt and what we're going to release are HTML5 overlay interfaces that have very specific test capabilities. We are beginning to take our core underlying intellectual property and overlay it with a nice and easy-to-use interface that applies to an enterprise's specific challenge, be it security, be it storage performance or be it call center verification.
What are some of the emerging technologies Ixia is keeping an eye on and how will those affect future strategies?
Haugh: On the layer 2 and layer 3 side, we have a very good handle on the protocols that are coming in. We take part in all the standards bodies and we work very closely with equipment manufacturers. We are already beginning to look at far-out things. We are continuing to look at higher densities for higher speed Ethernet. I think the challenge and the opportunity for us is to really capitalize on the explosion of virtualization and virtual support for network function virtualization. So as things move into the server environment, how can Ixia be there and be spun up and tested right in line with the deployment of a service or function. I think this is where we are beginning to move toward.