Avatar

With several key applications moving to the clouds, how do our customers ensure application performance? What if they deploy for instance Public or Private Hosted solutions or hybrid WAN, how do they ensure application experience?

Today, we all see more and more new delivery models such as Private Cloud, Hybrid Cloud or Public Cloud, new Hybrid WAN deployment replacing or complementing MPLS by internet link to reduce cost and enhancing application delivery, Audio and Video applications deployed in enterprise, applications going HTTP or HTTPS making them more complex to detect.

How does the network play a critical role in the application experience that enterprises need to provide today to their users? How can you rely on the network to provide enough agility, flexibility and control with so much new applications, deployment models and delivery methods.

1

Enterprise today need to be able to rely on the network as well to handle all those new challenges. IT organizations need the ability to identify, monitor applications running on the network, define policies and better control and classify those applications, to provide the best end user experience, and keep up with new request growth but without having to replace all the actual infrastructure.

Patrick Charretour, Consulting System Engineer in my team has been focusing on Application Awareness and Performance for some time now, including Cisco Medianet and Cisco Application Visibility and Control Architecture, and his working with many enterprises on this strategy today.

Q: Patrick, Applications have been running on the network for a long time, what has changed?

A: You’re right, an application running on the top of the network is not new at all. But today, just being able to detect and know what are the applications running on the network is becoming more and more complex. We used to be able to detect application based on precise and unique information such as port type (for example, we all know HTTP is port 80). However, now a lot of applications are HTTP/HTTPS, including business critical applications on which the enterprise success is relying. Other applications may have different components in the same flow such as Audio, video and more. Finally, we have cloud applications and even Cloud VxI client delivery.

Detecting is just the enabler. As soon as you are able to detect, you may want to monitor how those applications behave and based on this monitoring you may need to apply policy and control in order to provide the best quality of experience to the end users.

Last, users are more and more mixing professional and personal applications in their day-to-day work, which is bringing a risk in term of security of course but can, also, consume network resources for non-critical business applications.

2

Q: Often when we speak about providing better quality to an application, we think Quality of Service (QoS)?

A: True. QoS is a way to provide (or not) bandwidth to applications especially where the bandwidth availability is the most critical, in the WAN.

However, QoS needs to evolve to become application detection and application monitoring oriented. QoS is not anymore about providing bandwidth; it is about providing priority or not to an application based on application priority and behavior. It also needs to be complemented by intelligent path selection (PFR).  Service Providers and Managed Service Providers are in fact already working on those changes today.

I want also to mention that we often think that the problem is in the WAN as in the LAN we have enough bandwidth. Well, it may be true, but it is taking a risk.

As an example, when enterprise used to deploy Videoconferencing or even Telepresence on their network, it was easy, or at least not too difficult, to quantify the impact of those applications on the network. First of all, number of units was not so important and second, most of the organizations were running the service in reservation mode allowing them to know and better control the impact on the network.

Today we see fast growing deployment of Unified Communications Clients onto the user desktop, including in Cisco. Those client or application allows the user to communicate by chat, audio or even video. But those clients bring also new challenges as they operate in on demand mode, so it is far more complex to know what will be the impact in real time. Last but not least they are an application in the middle of the other applications, so how do you differentiate them?

What is true in this example with audio and video clients is also true for cloud applications, leisure applications such as YouTube, encrypted application and even homemade enterprise applications.

 

Q: At Cisco, we seem to have several solutions looking after Application Awareness and Performance? Application Visibility and Control, Medianet are two examples. Why?

A: Yes we have several initiatives in this area. While Application Visibility and Control (AVC) is embracing the whole application challenge, it quickly becomes obvious that we also needed to work on a more specific solution to help our customers with their audio and video applications. But both are solutions relying on common technologies working together to identify, monitor and control applications.

 3

Medianet provides visibility and monitoring of audio and video applications flow but also of the clients, and end points. All of them are audio and video application, but all of them are different and have different requirements. Last, in most of the enterprises, audio and video applications are handle by a different group than the one dealing with business applications.

4

Both solutions complement each other, allowing an enterprise to start with a generic application performance solution such as AVC (including a view of audio and video flow) and then go with more granularity in audio and video application performance and control with Medianet, without having to implement new technologies. And the reverse is true as well.

Both AVC and Medianet complement each other and are key components of new initiatives such as IWAN for new Enterprise Deployment model or Unified Access for converged Wired Wireless.

Q: Speaking about Technologies and impact, we hear often that Deep Packet Inspection methods are great to detect applications but they can also have an impact on the performance. Constant monitoring of the performance can also have an impact. What is your view?

A: You are right. DPI techniques such as NBAR2 are very powerful and let the network detect what applications are in use. Today we are able to detect and differentiate more than a thousand applications. Yes, there is an impact but we constantly improve it. Last, NBAR 2 is embedded in the customer network, no need to add a new box with all maintenance and separate configuration associated.

Going further, we are also starting to have more and more possibility of not having the network detect the application, but to have the application to let the network know it is there. We started with Medianet MSI and Metadata. Our audio and video applications now have the ability to provide to the network a library of information. Who knows the application better than the application itself? The network can then take decision in policing, control and monitoring.

I want to make clear that it is not about replacing one solution with another, but about having the flexibility to adapt to each different situation.  To be able to react tothe fast growing number of new application and complexity, especially with more and more applications going to the cloud and being encrypted.

On the monitoring aspect, we made a lot of progress in making the deployment of those monitoring solutions less complex and with less impact. The same Management management applications used to configure and monitor the network, can now be used to deploy and operate those solutions easily, as they are in fact already in the network.

Do you see the same trends in your own organization? Which challenges are you facing? Please tell us.



Authors

Eric Marin

CTO

Borderless Network Architecture, EMEAR