infoTECH Feature

December 01, 2009

The Importance of QoS for IP Communications - Interview with Packet Island, Part 1

During the recent BroadSoft Connections event, I was struck by how much attention service providers were paying to the Packet Island (News - Alert) display. They aren’t the only company focused on QoS and QoE, but it was a great opportunity to see first-hand how important this is becoming now, especially for service providers offering converged services to their business customers.
I think Packet Island was a great addition for BroadSoft, and having met their CEO, Praveen Kumar at Connections, it was a good time to get his thoughts on the topic. Being part of BroadSoft (News - Alert) now, I should clarify that his title has changed to BroadSoft PacketSmart Chief Technologist. BroadSoft PacketSmart is currently a standalone SaaS-based offering, but is in the process of being integrated into the BroadWorks platform. Praveen has a lot to say about the challenges for addressing QoS and QoE, along with the value of providing this as a hosted solution, and this article is Part 1 of my two part interview with him.
JA: Prior to joining the BroadSoft fold, Packet Island was a small, but highly focused company that addressed some important challenges around the delivery of IP communications in a converged network environment. To familiarize our readers a bit more, please tell us a bit about the issues you were seeing when the company started in 2004.
PK: During 2004, my co-founder and I were part of a company that bought the VoIP chip business from Intel. As we folded that business into our company, we saw signs that VoIP was finally starting to take hold after many years of promise. However, much of the early success we saw seemed to be on the consumer side, where Vonage was waging a battle to take ground in the traditional telecom space while Skype (News - Alert) was using P2P to simplify PC-based VoIP calling.
However, on the business VoIP side, we saw the market having very slow take rates. Much of this slow growth had to do with convincing business customers that Hosted VoIP was as good as traditional PSTN voice. Many of the early VoIP providers assumed that more bandwidth meant good VoIP quality, but failed to realize that bandwidth alone does not ensure real-time service quality. So, as many of us now joke about in the VoIP space, the expectation of “plug-and-play” VoIP deployments soon became “plug-and-pray” VoIP deployments.
Some service providers worked around this challenge by deploying dedicated end-to-end networks for VoIP delivery. However, this approach often meant higher costs in terms of new wiring, switches, and management, and worked against the value proposition of VoIP being a less expensive but more powerful alternative to PSTN due to the benefits of converged media networking.
It was clear to us that a new approach was required to make cloud-based VoIP delivery meet the quality expectations of business customers.  This is where the concept of VoIP Lifecycle Management (VLM) made perfect sense. VLM is based on the idea that for a successful VoIP deployment, you have to look at the VoIP service lifecycle in a holistic manner and address the entire lifecycle.
In 2004, the VLM concept was in its infancy and therefore new to most network managers. Data networks typically do not need pre-assessment or verification since provisioning sufficient bandwidth solved most data network quality issues – and SNMP-based polling was sufficient to monitor the health of data networks, since transient congestion issues and occasional packet loss did not really degrade data services. With VoIP, it became essential to monitor for transient congestion, packet loss, and other real-time issues that the SNMP polling-based approach misses out on. Thus, we saw the VLM approach as being crucial for successful VoIP deployments.
JA:  IP communications has come a long way since then. To what extent did you feel that Packet Island was pioneering solutions that seem so necessary today?
PK: VLM first started gaining traction in the enterprise VoIP space. As hosted VoIP started ramping up, the enterprise VLM vendors tried to apply the same enterprise-based VLM approach to the hosted world. However, such enterprise-born solutions were complex and designed for expensive and time-consuming on-site professional services engagements and, therefore, did not scale from operational and economic perspectives for the hosted VoIP market.
At Packet Island, we created an architecture and a business model to deliver VLM using a SaaS-based model that was more appropriate for cloud-based or hosted VoIP delivery. Service providers typically sold their services using both direct and indirect models, and did service deployments using a large network of service technicians, who are often not VoIP experts. We needed to create a multi-tenant solution that a service provider could use to have a handful of experts directing a distributed field force of service technicians to apply lifecycle management techniques to ensure successful VoIP deployments. We also needed to create a solution that could scale to monitor a large number of customer networks on a continuous basis using Deep Packet Inspection technologies to detect and characterize transient issues and help isolate root cause without truck rolls.
Though we started our company with an exclusive focus on the cloud-based VoIP market, we now see our technology applicable to all cloud-based services. The reason is that IT and telecom are now converging within the cloud. This is evident from many of the data-only providers adding VoIP and video to their IT application capabilities to deliver rich UC applications from the cloud. When cloud-based apps were just delivering data, adding bandwidth was the easy answer to most quality issues. As real-time VoIP and video are now being added to the mix of cloud-based services, lifecycle management is required. Since this has to be done using a cost model that matches the economics of today’s cloud-based service models, the SaaS (News - Alert)-based model we pioneered in 2004 is more relevant than ever.  
JA:  Related to that, what are the typical trigger points that get you on the radar of service providers, and where do these issues figure in the overall scheme of their priorities?
PK: Unlike data-only cloud-based services, when it comes to delivering VoIP or video from the cloud, the customer expectation for quality is quite high. When it comes to ensuring service quality, hosted VoIP providers need to worry about ensuring quality in several parts of the delivery network, including the core application infrastructure, core service network, service delivery network, customer LAN, CPE devices, and more.  Among all of these elements, the service delivery network and customer LAN are elements that most service providers don’t have control over. Even those service providers that do control the end-to-end service delivery network don’t really have end-to-end control under a single organizational entity, making problem resolution a complex process.
Evidence we have from leading VoIP providers suggests that the cost of on-site service technician time resulting from quality issues is the second-highest cost item after the cost of the equipment used to deliver the service. Given the high customer acquisition cost, and the impact of service quality on customer churn, we have seen service providers prioritize operational support systems like what we provide right after the core infrastructure service components.
JA:  Let’s focus more on the momentum we’re seeing around cloud services and the SaaS model. Why is this happening, and why should service providers in particular be embracing this?
PK: There are two trends over the past decade that is enabling this massive shift to cloud-based services.
The first trend is on the enterprise side of the market, where there has been a shift from large homogenous private corporate networks to highly distributed multi-site networks that use the Internet as an interconnect. The broad availability of inexpensive broadband and VPN technologies has played a major role in this shift by eliminating the need to build large private networks using dedicated leased lines. Globalization has also played a role by leading corporations to decentralize their operations into smaller regional branch offices to enable efficient engagement with a global market place.
As large enterprises get highly distributed, it makes economic sense to centralize enterprise-IT services into “the cloud,” hosted either at the corporate data center or by third-party SaaS providers. Furthermore, the current global economic crunch is forcing many enterprises to rethink their IT strategies in terms of “own vs. rent.”  So, from a large enterprise perspective, my view is that IT has moved from its “strategic, and therefore need to own” status to an “essential utility that can be rented” status, leading to adoption of cloud-based services.
The second trend is in the SMB market, where inexpensive broadband has enabled a large number of SMBs to get connected to the Internet and IT-enable their business operations. For these cost-conscious SMBs, cloud-based services are very attractive, since they often cost one-tenth of equivalent in-house IT product options.
These two trends have played a major role in the momentum we are seeing in cloud-based services, and it is going to make many of today’s IT product vendors look irrelevant five to seven years from now.
This industry shift is an opportunity for service providers that have been fighting pricing wars in selling bandwidth or voice minutes. It enables them to expand into more value-added services and create new revenues streams. However, service providers need to take note that today’s emerging cloud-based providers are delivering services at price points that require economies of scale and, therefore, a fast ramp-up. Competing in this emerging cloud services space requires service providers to be nimble and shake loose from some of the traditional paradigms that have slowed them in productizing and supporting new services. The ones that move quickly can leverage their installed base to establish leadership positions and reap rich rewards.
JA:  Hosted services have been slower to gain adoption than most of us expected, but that is definitely changing now. Momentum for the adoption of both hosted and managed services has been very strong in 2009, and I expect to see more of the same in 2010. One driver would be the emergence of cloud services and SaaS-based offerings. How does the cloud change the value proposition for going hosted, and what are you seeing as the ideal scenarios for businesses to go this route?
PK: As you’ll recall, hosted services were a mini-bubble during the dotcom days, when the players were referred to as Application Service Providers (ASPs). However, a majority of them failed to gain traction since the last mile bandwidth was just not there due to lack of broadband availability. After 2000, as broadband penetration took off, the ASPs that remained in the field (e.g., (News - Alert)) went on to do extremely well.
Device/protocol interoperability issues have now pretty much been resolved, but network quality issues still remain. This is where we expect to catalyze the market to higher growth rates being part of BroadSoft.  As a result, we are observing accelerated growth now in hosted VoIP and SIP trunking as the economy forces companies to move to cloud-based services.

Jon Arnold, Principal at J Arnold & Associates, writes the Service Provider Views column for TMCnet. To read more of Jon’s articles, please visit his columnist page.

Edited by Erik Linask

Subscribe to InfoTECH Spotlight eNews

InfoTECH Spotlight eNews delivers the latest news impacting technology in the IT industry each week. Sign up to receive FREE breaking news today!
FREE eNewsletter

infoTECH Whitepapers