Cisco NetFlow is a must have for managing virtual applications in the cloud

Q & A with Jagan Jagannathan, the founder and CTO of application management 2.0 solution vendor Xangati.

Tech visionaries from all over the world will be visiting the Cisco booth #2118 in San Francisco's Moscone Center while attending VMworld 2009 next week from August 31 up until my birthday on September 3 (BTW: I'll be turning 54, but often it seems, Network World readers tell me I act like I'm 24). Additionally, VMworld attendees may also wish to take note that earlier this week application management 2.0 solution vendor Xangati, announced it would be demonstrating its AppMonitor for Virtualization Management at the upcoming VMworld 2009 conference in the New Innovators Pavilion pod #1438F (see the Xangati pod location map). Last month readers may recall, I did the first Cisco NetFlow Q & A with Xangati founder and CTO Jagan Jagannathan and so for readers who will be attending next week's VMworld, the Q & A followup below might be of interest too:

1. Let's get to the point, why is Cisco NetFlow a must have for managing virtual applications in the cloud?

Jagan Jagannathan: The answer is simple: Each router/L3 switch is capable of being a NetFlow source that is aware of each and every IP-based conversation that traverses it. And, given that there are routers present at all locations talking directly to the leveraging one’s pre-existing L3 infrastructure, they can be aware of all activity that their end-users are having with the cloud and plug the visibility gaps immediately. For more details on the value of NetFlow, one can refer back to the previous Q&A we did together.

2. What happens to one's existing management framework when their applications move into the cloud?

Jagan Jagannathan: Moving applications out of one’s network and into the cloud is highly disruptive for conventional application management systems, which were essentially architected for client/server deployments within the confines of an enterprise. The issue at hand is that for every application moved into the cloud, an instant visibility gap is created—how big that gap is depends on the pre-existing data collection model. If the application was previously managed by agents and now the enterprise is leveraging a new but equivalent application in the cloud, then IT operations loses all understanding of that application’s activity. If IT was using probes, the visibility coverage was just for a few key locations, which at the time was fine. However, cloud migration leads to a big de-centralization effort where any remote office can speak directly to the cloud through their own Internet connection, which means visibility is lost for the application activity of all these users. To make up that visibility gap would require a comprehensive and costly deployment of probes, which in and of itself would result in a manageability headache.

3. What's Xangati offering with Cisco NetFlow that helps with visibility gaps into the cloud?

Jagan Jagannathan: As we noted previously, part of Xangati’s differentiation is the use of NetFlow for Application Management. NetFlow enables Xangati to track all applications (both enterprise resident and cloud-based) in a virtual manner, whereas probes and agents are tied to the physicality of the enterprise. And specific to cloud apps, an important aspect of the solution is the granularity to track the application activity of every end-user down to the second. Given the dynamic nature of cloud environments where performance issues can come and go in the "blink of an eye," this level of specificity is paramount in being able to see problems as they unfold. Moreover, it is important to underscore the need to have an identity of the end-user tracked in real-time through a linkage to a directory system as opposed to looking at an ephemeral IP address.

NetFlow for Application Management 4. Doesn’t moving an application into the cloud cause the user-level application experience to vary considerably?

Jagan Jagannathan: It absolutely does. An end-user’s experience with an application can potentially vary minute-by-minute and be different depending on how they are accessing the application. Additionally, each end-user’s experience will be different than their co-workers. After all, or WebEx et. al, do not have their applications served by one giant monolithic centralized server. All of the above, simply adds to burdens of IT operations even though some have thought of the cloud as a way to take things off of IT’s plate.

5. Xangati made an announcement earlier in the week about its AppMonitor for Virtualization Management being demonstrated at VMworld, does this help address the user-level application experience challenge?

Jagan Jagannathan: The AppMonitor was specifically designed to provide the end-user directly with a window into what their experience is with an application at any given moment. It is the first time in one product (a thin-client linked back to the core Xangati appliance) that an end-user can view their complete application activity, while at the same time track their response time to the specific cloud server they are trying to communicate with. In short, it is not enough to know that you are talking to WebEx; you have to know where specifically in the cloud you are communicating. The AppMonitor is the first self-help solution for a business end-user for any and all of their network-based applications. And IT likes it because it helps establish a level of transparency between themselves and their business users. In addition, it has our Visual Trouble Ticket capability embedded in it, so an end-user can share their exact experience with IT at any point they are experiencing a performance problem.

View demo unframed

View demo unframed 6. Given the variability in end-user performance, is it fair to ask a service desk to support applications that are not even on their network?

Jagan Jagannathan: It is not really a question of fairness. Like it or not, when there is a performance problem with a cloud application, a call is going to be launched into the IT’s front-line—the service desk—and not the cloud provider. Therefore, to quote one of our customers who is an IT architect at a F500 manufacturing company: "It is great that IT is saving big money on this cloud project, but that doesn’t mean a thing if the cloud just blew up IT’s incident management process." That is why they are leveraging the Xangati Visual Trouble Ticket framework; because instead of "flying blind," the end-user collaborates to help them get a DVR-style recording of their actual application experience at the exact instant they were having a problem. And with actual, granular information on where the response time breakdown is, there is an opportunity to define where the ownership of the problem resides. Not only is the incident management process maintained for cloud applications, it becomes enhanced.

7. How should an enterprise go about working with their cloud provider to properly assign ownership of a problem?

Jagan Jagannathan: The reality is that the vast majority of cloud application providers are not going to offer IT any visibility into what’s going on within their cloud and with their specific servers. That is why IT needs to establish a framework that provides transparency on its own terms. That is where the Xangati VTT framework with response time measurements comes into play. The IT organization will be able to look at the replay of the issue and isolate the problem area to something within their enterprise boundaries, within their ISPs network or within their cloud provider’s infrastructure. If the recording points to an issue with the cloud provider, the IT point person can forward on a copy of that recording to support at the cloud provider. That recording will isolate the individual user’s experience with an explicit server at an explicit point in time. With that level of specificity, the cloud provider should be able to use its own instrumentation to diagnose the circumstances around the end-user’s sub-optimal performance with the application. We see this framework as a way to literally close the loop on addressing performance issues.

8. How does an application owner engage with a solution like Xangati if he doesn’t understand the ins-and-outs of NetFlow?

Jagan Jagannathan: An application owner shouldn’t care about the technology behind their application management solution. What they should care about is being presented with comprehensive (yet detailed) information about what is going on with their application in real-time. The main Xangati UI can be customized by the application owner to track their application in aggregate and from all their critical regional offices. If that level of detail is considered too much, we have also developed an AppMonitor for App Owners that provides them a quick reference dashboard for their application.

9. Many cloud providers charge based upon a data usage model, how can NetFlow help with that?

Jagan Jagannathan: You are correct that part of the bill one would receive from Amazon Web Services, as an example, would be usage based—usage-based in terms of how much data was exchanged between the enterprise and the Amazon cloud. NetFlow provides the means to track the cumulative usage of a given application in aggregate and down to specific regions, if necessary. As a result, the NetFlow information might be the only way to actually verify that the bill you have received maps back to actual activity between the cloud and the network. This information is available in the main Xangati UI as well as through the AppMonitor for App Owners.

10. August 31 is the beginning day of VMworld 2009, what will Xangati be demonstrating there?

Jagan Jagannathan: We are going to be in the New Innovators Pavilion (pod # 1438F, see map to our pod location) demonstrating our new AppMonitor functionality. For those who happen to be there, we welcome the opportunity to walk you through all that we covered above in the actual product on a production network.

What's your take, do agree with Jagan that Cisco NetFlow is a must have for managing virtual applications in the cloud?

BradReese.Com Cisco Refurbished - Services that protect, maintain and optimize Cisco hardware Contact: Brad Reese | Twitter:

  1. Plixer offers free tool that brings Netflow analysis to Cisco ASA firewall
  2. Cisco's commercial business council cochair bolts
  3. Cisco adds lawyers to its cost-cutting list
  4. Is the fair value of Nortel's metro Ethernet and 40Gig product line only $600 million?
  5. Former Cisco star joins Arista Networks
  6. Ericsson President and CEO has no fear of competing with mighty Cisco
  7. Wall Street Analyst: IBM is hedging its bets on Ethernet switches
  8. Cutting out the channel, are we seeing Cisco's future with the launch of Eos?
  9. Exactly how new is Cisco's organizational structure?
  10. Management vision of Cisco CEO John Chambers under fire
  11. Cisco says there's no CCIE amnesty program
  12. Employee reviews on slam Cisco
  13. In-line monitoring vs. NetFlow or sFlow
  14. Cisco board directors are at the epicenter of the Microsoft and Yahoo! deal
  15. What business is Cisco in today?
  16. HP scolds the Cisco unified computing system
  17. Cisco CEO John Chambers to stay at Cisco for another decade or more?
  18. Cisco buying Dell: Allan Leinwand's idea is brilliant
  19. Chocolate icon Hershey fails at online retailing
  20. View Brad Reese on Cisco Story Archives
Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2009 IDG Communications, Inc.