TMCnews Featured Article

 

My last Service Provider Views column featured an interview with Praveen Kumar (News - Alert), CEO of Packet Island, which was recently acquired by BroadSoft. The focus was on QoS and QoE, not just for voice, but now video and the broader Unified Communications (News - Alert) umbrella. With BroadSoft, they are integrating these capabilities into a hosted solution, which many – myself included – view as the next big wave in services, especially for SMBs.

 

There’s a lot to talk about here, and the overall interview has been broken up into two parts. This column is Part 2, and picks up where Part 1 finished off, namely why hosted and cloud services are gaining so much traction now. The interview continues below.

 

PK: QoS can mean different things to different people. I often explain QoS as representing the set of technologies that help deliver high quality IP-based services – particularly for real-time services, like VoIP and video. There are a variety of QoS technologies that have been invented to improve service quality of IP-based services. However, the issue with QoS is that it requires end-to-end enforcement and, therefore, end-to-end control of the service delivery network. This is not practical in today’s cloud-based services world, where the service often traverses multiple service provider domains before ending up in the customer’s LAN.

 

To make matters worse, most last-mile access providers are locked in price wars, where the business objective is to provision as many broadband users as possible into their access networks to provide reasonable data service quality, but poor real-time service quality. In fact, we have observed several cases where last-mile access networks turn off QoS tagging in IP packets as soon as the packets enter the networks.

 

Therefore, ensuring end-to-end QoS is not very practical in today’s world of heterogeneous networks. This is one of the main reasons why a lifecycle management approach is required so that you can determine issues before deployment and manage the service quality over the lifecycle of the service as the network evolves.

 

QoE is often defined as the quality of the service as experienced by the end-customer. QoE depends on the type of application service that is being delivered and is influenced by the end-to-end QoS that is in place. 

 

PK: VoIP uses approximately ten times more packets than data for the same bandwidth use – this means that the typical low-end routers and firewalls in SMB and branch office networks have to potentially do ten times more work once VoIP is deployed. Video makes this problem more intense, since now, not only are there a lot more packets, but they are large packets as well. Furthermore, video degrades quicker than VoIP.

 

So, lifecycle management is critical for successful video deployments. We have a solution for video deployments where we can pre-assess networks. Our assessment technology embedded in our micro-appliances allows us to test networks for low-end video all the way up to high-end telepresence video deployments. We also have continuous active monitoring technology built into our appliances that allows us to monitor the real-time performance of video links while monitoring who is causing congestion spikes. We plan to continue to expand on this capability over the next 12-24 months.

 

PK: Our DPI function is used for ongoing monitoring and diagnosis of application flows, with a specific focus on VoIP. Our devices are passive monitoring appliances that sniff VoIP packets to characterize problems on a per-second basis. We also do active route analysis for each call to determine WAN congestion. When congestion is caused by the LAN, our data flow analysis helps identify who is causing the congestion. All of this data is collected on a per-second granularity and can be used to characterize the nature of the degradation, with the objective of leading to a quick resolution.

 

We are able to do this because of our cloud-based service model, with which we can store and correlate such fine grained DPI data at a carrier scale. We store this data for several weeks allowing for trend analysis, capacity planning, and SLA reporting. So, our competency is really in the cloud, where we can collect, process, and correlate to enable quick problem resolution across a large number of customer sites.

 

SBCs have traditionally focused on admission control and the DPI capabilities that are embedded in SBCs are not focused on data collection for scalable problem diagnosis across a large number of customer sites.

 

PK: You are absolutely correct. When we started our company, we wanted our solution to not be just a pure cost item for service providers, but a revenue generator as well. This makes ROI determination a much simpler exercise. For example, we have several customers who use network assessments to demonstrate to customer prospects that they need a new broadband T1 line, or need to upgrade their firewall, etc. to ensure business quality VoIP.

 

This creates new up-sell opportunities for product/service sales, in addition to being able to charge for the assessment as a professional service engagement. Once DPI-based monitoring is deployed, service providers can offer SLA reporting services, network behavior analysis services, remote troubleshooting services, and more, to help customers enjoy good QoE. These services can be charged on a subscription basis or on an on-demand per-use basis.

 

In addition to generating new revenues, our solution provides cost savings on the support side. Given the high volume, low margin economics of cloud-based business models, high support costs can kill the business. A basic truck roll to a customer site typically costs $300, and keeps going up the longer it takes to troubleshoot the issue. Given that typical cloud-service customers are small customers, it is just not practical to do truck rolls whenever transient quality problems are reported.

 

We have had many service providers share horror stories about how their engineers have spent many hours trying to diagnose transient issues at customer sites that only generate a few hundred dollars of annual revenue. We offer DPI-based monitoring services with annual pricing that costs less than a single truck roll providing a powerful ROI based on support cost savings.

 

PK: We have been fortunate to have had good success at the BroadSoft Connections conference every year we have exhibited. This is the first time we have exhibited as part of the BroadSoft family, and we are thrilled about the future prospects. We have always seen our solution as a catalyst for the cloud-based VoIP/video market. Over the past few years, service providers have gained much more clarity on the need for QoS/QoE management through a lifecycle management approach. With BroadSoft’s installed base of over 450 carriers, Packet Island now gains customer and partner reach, and has the opportunity to accelerate the growth of cloud-based VoIP and video services market.

 

PK: Much of the functionality we will be adding in 2010 will be focused on simplifying end-user problem resolution by adding deeper problem detection and characterization capabilities along with automation. You will also see us partner with several device vendors to embed our technology in CPE devices and IP PBXs to further catalyze the growth of the cloud-based market for converged media services.