EC13 Report Part 3: Mobility

Mobility for UC comes in many shapes. A soft client on iOS and Android is the entry ticket, but the sky is the limit after that.

This is the third in a four-part series covering some of the major trends I saw at Enterprise Connect Orlando. Friday I covered Virtualization, and yesterday Cloud Services; tomorrow's topic will be the Conference Room Experience.

Mobility

UC clients on mobile devices have quickly become the entry ticket to mobility. Every vendor I talked to--voice, video, or other--has soft client(s). iOS and Android are the preferred OS for smart phones and tablets, followed by Windows for laptops and a mention or two of BlackBerry. However, a good mobility story rarely stops at just mobility.

Extensions to the call processing software allow switching (or shifting) sessions across any number of devices associated with a user. Alcatel-Lucent's "rapid session shift" switches between mobile phone, wired phone, soft client, or a video endpoint from partner LifeSize. There is no auto detection, and the changeover is initiated by a push of a key or tap on a screen.

Thrupoint takes session management to an entirely different level. Their Session Broker--which originates from Ubiquity--allows applications to execute during session setup, and for example, to check permissions, apply policies, and re-route sessions to another location/device. Thrupoint therefore provides a scalable platform for mobility applications.

I came across several mobility servers that allow additional automation of the call handoff process and focus on the Wi-Fi/3G/4G/GSM roaming scenario. The idea is that when the user leaves Wi-Fi coverage, the call automatically switches to VoIP over 3G/4G. If the 3G/4G service deteriorates, the call automatically switches to basic TDM GSM voice.

Aastra demonstrated the Aastra Mobility Controller (AMC) from the acquisition of Munich-based Comdasys. The mobility server is offered with the Aastra's MX-1 system (acquired from Ericsson) or as as standalone product with third-party systems.

ShoreTel has fully integrated the mobility server from the Agito acquisition, and offers it with ShoreTel IP Phone System or as a standalone product in third-party environments.

Conclusion

Mobility comes in many shapes. A soft client on iOS and Android is the entry ticket, but the sky is the limit after that. Advanced mobility functions such as session shifting and session/call handoff may become important differentiators for vendors. Reliability is still an issue, since mobility touches on several networks, and network timers do not always work exactly as expected.

EC13 Report Part 2: Cloud Services

EC13 Report: Cloud ServicesVendors are rushing into creating cloud services to reach new customers--but there are concerns to address.

Vendors are rushing into creating cloud services to reach new customers--but there are concerns to address.

This is the second in a four-part series covering some of the major trends I saw at Enterprise Connect Orlando. On Friday I covered Virtualization; tomorrow's topic will be Mobility, and Wednesday's will be the Conference Room Experience.

Cloud Services

Cloud services are popping up everywhere, reflecting the lower barrier to entry for those who want to become a service provider. There are currently several hundred Managed Service Providers (MSPs) that provide voice and video conferencing services. The market is very fragmented and the barrier to entry is relatively low, so players are trying to differentiate themselves by supporting multiple vendors, solving interoperability issues, and gaining critical mass, In addition to AVI-SPL, I met with AGT, which differentiates itself through offering software-based MCU in the cloud and offering management of video endpoints.

More and more vendors across the industry are trying to make it even easier to become an MSP--by building infrastructure for running their applications in the cloud, and by encouraging their distributors to become MSPs and resell the service. As bonus, the vendors provide management tools so that the freshly baked MSPs can measure usage and charge end users for services month by month.

The cloud allows some traditional on-premise vendors to address very small customers that cannot really afford an on-premise solution. Interactive Intelligence announced its Communications as a Service (CaaS) offering called Small Center, targeting 10-50 agents--and promising unrestricted growth up to 5,000 agents without platform change. This is adding fresh competition to Five 9, a pure-play cloud call center offering.

On the video side, Magor is repositioning itself away from telepresence and towards being a visual communication infrastructure vendor. Its new Aerus cloud service leverages sophisticated routing algorithms to enable new collaboration models, away from the traditional "endpoints calling into the bridge" model.

Since these offerings rely on Internet best effort service, some industry analysts warn against it, while others embrace the Internet quality. Unified Office is taking a different approach, focusing on engineered quality for customers (small businesses with 5-75 employees) that need more than best-effort. Unified Office installs a TCN (Total Connect Now service) box on customer premise to measure QOS and analyze IP networking problems; then works with IP networking service providers to resolve issues. Unified Office also uses an open source SBC based on Asterisk that connects to multiple SIP trunking providers and routes calls based on network quality. This reminds me of the Least Cost Routing function in PBXs; however, Unified Office does not look at cost, just network quality.

The big question is "Why do vendors roll out their own cloud services?" One school of thought is that vendors waited for service providers to develop scalable services and since this did not happen, decided to try it themselves (although it is clear that running a service is very different from making products). Another school of thought is that it is just everyone in the industry looking for room for growth by experimenting with new business models, even risking competition with SP customers.

Finally, the term "hybrid cloud" was often used in conversations at Enterprise Connect, always with different meaning. Some picture "hybrid" as a media server on premise with application server in the cloud. Other imagine some services implemented in private cloud (essentially a hosted outsourced data center) while other come from the public cloud. BroadSoft's definition, for example, is having BroadWorks system in the SP data center while getting UC services from the BroadCloud.

Conclusion

Vendors are rushing into creating cloud services to reach new customers. Small enterprise companies will benefit the most, as cloud services would allow them to use platforms they cannot afford to install on premise. One concern is the best effort quality of the Internet that most cloud services rely on. The second concern is how well vendors will perform as service providers. The third concern is about vendors competing with their service provider customers.

EC13 Report Part 1: Virtualization

EC13 Report: VirtualizationMost of the core call control vendors are virtualizing at least someof their systems; now virtualization is spreading to video elements, contact centers, SBCs and more.

Most of the core call control vendors are virtualizing at least someof their systems; now virtualization is spreading to video elements, contact centers, SBCs and more.

I have been attending Enterprise Connect and its predecessors since 1998, and my focus gradually shifted from PBXs to IP-PBXs to call centers to video conferencing and most recently to cloud services. Enterprise Connect 2013 was a unique opportunity to meet with vendors from across these communications industry segments and search for commonalities and trends.

The sheer number of vendors participating in EC requires discipline and excellent planning before the event. To get a comprehensive view of the industry developments, I met with 28 vendors in back-to-back briefings, visited the exhibits, and attended a few targeted conference sessions. The most powerful trends I discovered during the event are virtualization, cloud services, mobility, and redefining of the conference room experience. I have written a post on each of these topics, and these will run on No Jitter over the next 4 days, beginning with virtualization.

Virtualization

Virtualization--once the exclusive domain of non-real-time applications--is now becoming a cornerstone for scalable and redundant deployments in voice and video communications. Here are some of the vendors' strategies:

* Trailblazer Mitel developed a strategic relationship with VMware early on, and systematically virtualized its entire portfolio. For them, virtualization is not a feature but rather a strategy for developing new channels and reaching new customers.

* Avaya has virtualized its Avaya Communications Manager and Contact Center application as well as several other elements of the Aura architecture.

* ShoreTel has virtualized its IP Phone System and Contact Center application, but their approach to virtualization is less radical: virtual machines are positioned as a solution for larger offices, while appliances continue to be the preferred option for branch offices.

* Alcatel-Lucent OpenTouch has been virtualized and is used by partners to offer cloud services. Its release included a new licensing model and a new tool to monitor licenses.

* Siemens Enterprise Communications is in the process of certifying its OpenScape portfolio with VMware. * NEC has already virtualized UNIVERGE 3C, its Windows-based UC system from the Sphere acquisition.

So virtualization is becoming a part of most vendors' strategies for pure-IP systems. When it comes to hybrid TDM/IP, Avaya, Alcatel-Lucent, Siemens, and NEC have such hybrid systems in their portfolio but are not rushing into virtualizing them.

Virtualizing hybrid systems is not easy, and even if part of the system is virtualized, there must be a TDM/IP gateway somewhere in the network to support TDM extensions and trunks, which negates some of the virtualization benefits. Aastra MX-ONE has been virtualized, although the virtual edition is only offered on the European market for now. Aastra's mobility server (Aastra Mobility Controller = AMC) is also virtualized.

In the video conferencing world, Polycom is preparing its RMX (now RealPresence Collaboration Server) platform for deployment in virtual environments. Since server virtualization is based on x86 CPU architecture, Polycom's first step is moving from Digital Signaling Processor (DSP) to x86 CPU hardware; the result is Polycom RealPresence Collaboration Server 800s, Virtual Edition, a physical server that runs the conferencing application on x86 CPU. The resulting scalability decrease (20 transcoded HD video ports, as compared to 80 in the DSP version) is compensated by offering 60 Scalable Video Coding HD ports that require no transcoding.  While the commercial success of such a solution may be limited, it is a necessary step towards running the Polycom media processing engine on generic hypervisor. Eventually, this is the architecture required to power Polycom's Cloud Axis service.

Radvision (now part of Avaya) took a different approach with Scopia Elite 6000. While the conferencing application in the appliance runs on x86 CPU, an accelerator board handles some of the video processing. That allows the scalability to be kept up to 80 transcoded HD calls and is a small step towards the cloud future. Other solution elements such as Scopia Desktop are fully virtualized.

AVI-SPL deploys Cisco video infrastructure in virtualized environments, including applications such as VCS. It would be interesting to track if Cisco will take the Polycom or Radvision approach in getting its Codian MCUs from purpose-built DSP hardware to virtual environments.

Looking across the contact center industry to test the hypothesis that virtualization is an universal trend, I came across Five9, a cloud call center provider that uses virtualization in its network of leased data centers, and Calabrio, whose Calabrio ONE suite of contact center workforce optimization software includes call recording, quality assurance, workforce management, speech analytics, and performance-based management. For Calabrio, the only real technology challenge is virtualizing the call recording part, which requires consistent high performance and dedicated HW resources.  This was a consistent theme in all conversations around virtualization: functions/applications that require dedicated resources are hard to virtualize.

In the voice networking segment, Session Border Controller (SBC) vendors are beginning to think about virtualization, with Sonus having specific plans to virtualize its SBC products before the end of 2013. They recognize the virtualization trend but want to avoid performance limitations, especially with transcoding in SBCs.

I was not sure I would find virtualization in Adtran's portfolio but it turns out Adtran Wireless Access Controller is fully virtualized on VMware. Replacing controller hardware every time the Wi-Fi standard gets updated (remember 802.11 a, b, g, n) is not efficient and the company moved to a software-only controller that they later virtualized.

Virtualization also found its way in Extreme Networks' management application, which now runs on VMware. Extreme's policy management engine retrieves user information from the enterprise directory (Active Directory) and applies policies to each user, no matter where and how the user is connected to the network. With virtualization, Extreme has carried this concept over to virtual machines: the management application retrieves information about VMs from the VMware hypervisor and allows the administrator to create profiles per VM. As the VM moves from one server to another or from one data center to another (using the vMotion feature, for example), Extreme tracks the VM and keeps applying the same policy to it no matter the location.

Conclusion

Virtualization is not a religion, and vendors have to pick and choose which elements to virtualize in their portfolios. Media processing functions are the most challenging, since they require dedicated resources--going against the fundamentals of virtualization.

 

Controlling the Borders

After Oracle announced the acquisition of Acme Packet, a lot of people asked me about the role Acme Packet plays in the network. I started piecing together bits of information to understand the tumultuous history of session border controllers that led to the acquisition. The result is this article that hopefully explains the value of Session Border Controllers (SBCs) in the network and Acme Packet’s role in this market segment.

SBC Origin

SBCs have been around since the early days of Voice over Internet Protocol (VoIP) and have always been in the center of heated debates in the industry. The VoIP purists believed that the VoIP client should communicate directly with the VoIP proxy, so that the session can be established end-to-end, that is, between the two communication parties (clients). The model first broke when telecom-style soft switches started terminating sessions to clients and messing with the session parameters; they became so-called Back to Back User Agents (B2BUA). The benefit of B2BUAs is that they can implement telephony style features like forwarding, transfer, conference, and the “holy grail” of enterprise telephony - multiple line appearance.

Once the end-to-end session was broken however, folks decided that it can be broken several times, and put another type of session breakers – session border controllers – between VoIP clients and soft switches. VoIP purists were enraged but, as time passed, the practicality of SBCs won over the argument for clean architecture. At the end of the day, SBCs solved the most difficult problem for VoIP, firewall/NAT traversal, and the choice was really between successful call based on “bad” architecture vs. failed call following the “good” VoIP blueprint.

Over the years, I have attended a lot of IETF meetings and the issue of bad SBCs messing up the clean VoIP architecture was frequently discussed. Since IETF created SIP as a signaling standard for VoIP, the discussions there were always about SIP clients, SIP SBCs, and SIP servers. In the parallel universe of video conferencing, the ITU standard H.323 was in use long before SIP emerged. Firewall traversal was as much an issue in the H.323 world as it was in SIP; therefore, H.323 SBCs started appearing in product portfolios, mostly based on the ITU-T H.460 standard. 

SBC Functionality

The original firewall traversal function in SBCs was later enhanced with other connectivity services such as manipulation of SIP messages (for example to hide the senders address), IPv4 to IPv6 interworking (for connecting VoIP clients on IPv4 networks with VoIP clients in IPv6 networks), and SIP - H.323 translation (for example, to connect H.323 video systems with SIP video clients). Additional security services were introduced to protect the network from Denial-of-Service (DoS) and other attacks and from malformed IP packets.

More powerful SBCs could also encrypt/decrypt signaling (via TLS and IPSec) and media (SRTP). SBCs gradually evolved to enforce Quality of Service (QoS) policies (for example, to limit bandwidth or limit number of calls), perform regulatory compliance functions (emergency calls prioritization, lawful interception, etc.)

Many of the new generation of SBCs also provide built-in digital signal processors (DSPs) to provide media services, such as media transcoding. The bottom line is that SBCs started as firewall traversal tools and ended up as complex elements of the VoIP network doing pretty much everything but call control (that was prerogative of the soft switch/application server).

SBC Market Dynamic

Acme Packet was at the forefront of the SBC market, some may say they defined the market segment. In the early 2000s, service providers started offering hosted VoIP services and needed reliable and scalable SBCs for their new networks. Since tier 1 service providers usually buy from a small group of vetted vendors, Acme Packet built partnerships with Ericsson and others, and became very successful selling through such channels to service providers. Two other SBC vendors - Kagoor Networks and Jasomi Networks – were less successful. I remember meeting with these companies in the early 2000. At the time I was leading the Product Management team for Devices at Siemens, and our SIP phones always went through some kind of SBCs before connecting to SIP servers / soft switches; we therefore tested interoperability. Around 2005, the SBC market became very hot and big players started buying SBC startups. Juniper Networks acquired Kagoor in 2005. Ditech Communications acquired Jasomi in 2005 (Nuance Communications acquired Ditech in 2012). Nextone was swallowed by Nextpoint which Genband acquired in 2008. AudioCodes acquired Netrake in 2006.

Big VOIP vendors took various approaches to adding SBC functionality to their portfolios. Cisco developed its own Cisco Unified Border Element (CUBE) while Avaya acquired Sipera in 2011. In the service provider market, BroadSoft relied on Acme Packet (in fact that is how I first came across Acme Packet years ago) while others like 8x8 developed their own SBC technology.

The SBC market attracted competition. In 2010, Sonus Networks newly hired CEO Ray Dolan set the goal to make Sonus “strong #2 in the SBC space”. Sonus quickly transitioned from making soft switches to making high-end SBCs but needed medium and small sized models to complete its portfolio. In August 2012, Sonus acquired Network Equipment Technologies (NET) of Fremont, CA and – with now complete line of SBC products – is positioning itself to directly compete with Acme Packet in both the enterprise and SP markets. Sansay, the #3 in SBC market, emerged from the former Nuera Communications team in San Diego in 2002. At ITexpo 2013, Sansay won the “Best Service Provider Solution” award for their transcoding technology, integrated into the SBC product.

Through the years, I got to know two other players in the SBC market. Edgewater Networks focuses on SIP SBCs for service providers and H.323 SBCs for Polycom (Video Border Proxy = VBP portfolio), although recent version of the Polycom branded product support SIP as well. Ingate Systems focuses on SIP trunking and hosts its SIP Trunking Summit / Workshop at every IT Expo event; this event evolved into the SIP Trunking - UC Seminars. I had the pleasure to present at several Ingate events, and can say that they are very well-organized, well-attended, and very educational.

SBC’ Role in the Enterprise

While many speculate about Oracle using Acme Packet to build a stronger position in the service provider market, I see a lot of applications for SBCs in the enterprise environment.

Enterprises increasingly use services from the cloud, including cloud-based voice/video services, and need tools to control the traffic to and from the cloud. SBCs provide this session control functionality. In the new architecture, enterprise users do not access the cloud services directly but through the enterprise SBC which provides security, accounting, etc. services and, if necessary, protocol translation and media transcoding. The SBC therefore becomes the official entrance for real-time communication to the enterprise. 

In addition, SBCs are critical for enterprise VoIP deployments that require connections among existing IP-PBX systems, new deployments (such as Microsoft Lync), and other enterprise applications (such as Contact Centers). When I think about it, SBCs with H.323 support can also connect the installed base of H.323 video conferencing equipment in the enterprise. In this scenario, the SBC becomes the universal connector, the glue that connects all parts of the currently disparate enterprise communication infrastructure.   

Will WebRTC Change The Communications Industry?

Back in April 2011, I wrote about the WebRTC discussion at the 80th IETF meeting in Prague. The idea to add real-time communications (RTC), that is voice and video capabilities, to the web browser is a natural progression of Google’s “everything should be done in the browser” philosophy, so there is no surprise that Google has been driving WebRTC. One and a half years later, WebRTC has created quite a buzz in the industry and a wave of startups has developed clients and services based on WebRTC technology. I was therefore excited to meet key players at the first WebRTC Conference and Expo in San Francisco that gathered about 300 participants, and focused on two questions: a) is the WebRTC technology ready for deployment and b) what is its impact on the communications industry?

The Surprise Sponsor
No one was surprised that Google sponsored the WebRTC conference.  But pretty much everyone was surprised to see Plantronics as the second sponsor of the event. Plantronics keynote speech by Joe Burton explained why... The digital divide is getting wider – while billions of devices are connected to the Internet, many people struggle with the technology. People ask Plantronics to solve the bigger communication problem: bend the web to meet the needs of the average person. It is more than headsets; it is about wearable technology that uses sensors to measure direction, temperature, moisture, etc. parameters and adjust the communication environment. Plantronics sees WebRTC as the communication infrastructure to support wearable devices because WebRTC allows much bigger pool of web developers to build communication apps much faster and leads to democratization of communication technology.

WebRTC Vision And Realty 
WebRTC’s vision is to provide interoperable royalty-free high-quality communication between browsers. To deliver on this vision, Google acquired On2 Technologies and made the VP8 video codec available royalty-free. There are many comparisons of VP8 and H.264 video codecs and the jury is still out on which one is better but one thing is for sure: transcoding between VP8 and H.264 leads to lower video quality than any of the two codes on their own.

On the audio side, the preferred codec for WebRTC is Opus, a super-wideband codec that delivers better quality than G.711 and G.722 used in IP telephony applications today. And while there are other super-wideband codecs, including AAC-LD and G.719 used in video conferencing systems, having Opus audio in the web browser will definitely drive web developers towards using it in voice/video applications. 

Unfortunately, the codec situation has not improved since early 2011. Google still insists on royalty-free VP8 and Opus while the communication industry is still based on G. and H. codecs. Some speculate that if H.264 becomes royalty-free, the argument for VP8 would be removed, and the world could happily converge around H.264. However, with the next video codec H.265 already on the horizon, the vision of using one video codec across applications and markets sure seems unrealistic.   

WebRTC Status
WebRTC today has fairly simple functionality. First, it allows the web application to get access to the user’s microphone and video camera (using getUserMedia API in JavaScript). This is critical step since JavaScript is downloaded from the web server and the user has to explicitly allow it to interact with the camera and microphone in the user’s device. Current implementations basically ask the user for permission for every WebRTC call, and there is still no good solution that combines ease of use and privacy. The second key WebRTC function is establishing a connection between two browsers (using PeerConnection API in JavaScript). This is the equivalent of “basic call” in voice communications and is therefore pretty straightforward. The third function is data transfer (using DataChannel API in JavaScript) – this function is designed for sharing data/screen as part of a collaboration session but it has not been implemented yet. In addition, WebRTC offers noise suppression, automatic gain control, echo control, etc. functions to complete the toolbox for web developers and enable creation of fully functional web based communication clients.

Google would like to see WebRTC in all web browsers but today only Google Chrome and Mozilla Firefox have the latest WebRTC functionality. Opera browser provides partial support of WebRTC (e.g., Opera v12 supports getUserMedia). Internet Explorer does not support WebRTC; the functionality can be accessed through a Chrome frame (a Google plugin for Internet Explorer). Realizing that not all devices can run voice and video in the browser (most mobile devices simply do not have the performance to do that) Google and other vendors provide native C++ versions of WebRTC for mobile platforms. In general, developers do not want to develop separate mobile apps, so as soon as the performance is available in mobile devices, the apps will run in the browser.

Mobile
One of the biggest challenges to WebRTC is addressing mobile platforms. In general, mobile devices have lower performance than wired devices and there are concerns that running software video codec in the browser will drain the mobile device battery. Others argue that the screen and wireless radio consume so much power during a video chat that the incremental power for running a software codec is negligible. Another reason for concern is Safari's ~99% share of iOS devices. Apple supports H.264 (and not VP8) and has not yet announced any plans for WebRTC. 

In mobile networks, there are several new communication methods that are competing for attention and resources. Similar to WebRTC, Voice over LTE is a technology that enables voice and video over (mobile/4G/LTE) IP packet networks. Real-time Communication Suite (RCS) on the other hand seems to be compatible with WebRTC. RCS defines what the client should look like and what functions it should support in a SIP call. It is therefore possible to create a RCS-compliant client based on WebRTC – as demonstrated by Crocodile.

Demos
The WebRTC conference included a lot of demos: some were part of the keynotes while other ran in the expo area throughout the event. 13 demos officially competed for best demo awards. Among the winners were Vidtel (Best Conferencing), Zingaya (Ready Now Award), Acme Paket (WebRTC to IMS gateway), and Crocodile (WebRTC/RCS client). Plantronics received the Visionary Award while PubNub (a San Francisco startup) won “Best WebRTC Demo” award and was also audience favorite.

Industry Impact
The current consensus is that while WebRTC clients will become very cheap or free there will be business opportunities around infrastructure and services.

Applications for WebRTC include telephony, video conferencing, collaboration solutions with gaming, etc. The current hype is around video but most benefits are in voice apps, where WebRTC can clearly cut the cost of communication and impacts existing business models. Right now, few people see WebRTC as money making opportunity and many say that asking how to make money with WebRTC is the same as asking how to make money with Flash and JavaScript.

Several companies make money connecting video conferencing traffic from Skype and GoogleTalk to enterprise video conferencing systems and this model can be extended to new WebRTC applications. But traffic among free consumer apps (like Skype, GoogleTalk, and future WebRTC apps) does not generate a lot of revenue. There is also a concern that cheap services can completely disrupt the subscription model used for voice and video services today.

Takeaways
In many ways, the WebRTC conference reminded me of the early SIP days (10-12 years ago). Many speeches predicted that WebRTC will replace PSTN but the logic question is "Why would it be different this time?" Some cited the better starting position the industry has today - with a lot of expertise about voice and video in IETF and development community. Increasing the developer pool is indeed a strong argument for WebRTC. While moving from SS7 to SIP in the 2000’s increased the developer pool from few hundreds to few thousands (the number 6,000 SIP developers was mentioned), moving to web applications increases the developer pool to several millions (potentially 10-20 million Java developers worldwide).

On the flip side, the WebRTC functionality is still quite rudimentary – mostly basic call, not even data sharing. The demos worked quite well but interoperability is still an issue due to the lack of WebRTC support in many browsers and due to incompatibility of voice and video codecs.

For service providers, WebRTC is yet another way for third-parties to develop Over The Top (OTT) services inexpensively, and compete with alternative SP-backed approaches such as VoLTE and IMS.

Enterprises are looking for creative ideas what to do with WebRTC, and most think of contact centers. Similar to the "chat" buttons and pop-up windows that have started appearing on web pages, I expect to see "call" buttons based on WebRTC technology in 2014-15. I have some ideas about WebRTC applications behind the corporate firewall and am sure that more ideas will emerge by the next WebRTC conference in June 2013 in Atlanta, Georgia. 

To come back to the original question, the impact of the WebRTC on the communications industry will depend on finding the right applications, adding the right functionality to the toolbox, and completing the standardization activities in IETF and W3C. Only then will WebRTC gain sufficient momentum and disrupt the communications market.