IP Everywhere

FORGET

ABOUT THE ATTENTION-GETTING BUT ULTIMATELY UNDERWHELMING VOICE-OVER-IP SOFTSWITCH OVERLAYS OR IPTV SERVICE DEPLOYMENTS OF YESTERYEAR.

In today's carrier networks, IP may not always be hyped or even seen, but it is indeed everywhere — and in 2010 it's only going deeper and making an even bigger impact.

Today, the wireline network core is largely based on Internet Protocol — or IP, for short, the packet-based protocol that most famously handles Internet traffic. Long-haul transport is becoming increasingly so as well, with not just IP but Ethernet — which began its life on the LAN but whose simplicity and compatibility with IP has made it the network of choice almost everywhere — becoming largely predominant. Next up: the access network, where fiber to the node and a focus on network/protocol convergence — not to mention the proliferation of IP applications — are pushing IP into tomorrow's connected homes (and pockets). Also on tap: greater levels of IP interconnect, in which carriers pass traffic from network to network as pure IP.

On the wireless network front, meanwhile, long-term evolution (LTE) and other 4G technologies are driving IP into tomorrow's mobile packet core. Not to be outdone, IP is moving into the mobile backhaul network as well, where IP and Ethernet are replacing yesterday's leased T-1s, as bandwidth requirements boom and timing issues challenge backhaul providers mixing voice and data packets on new converged networks.

Finally, there's a world of progress being made with IP at the application layer, with deep packet inspection (DPI), policy servers and other IP elements providing a new network intelligence layer, while IP-based architectures such as IP multimedia subsystem (IMS) and evolving application approaches such as rich communications suite (RCS) aim to drive IP throughout the network and out to the client level — not to mention the proliferation of over-the-top IP services and applications straining carrier networks.

“We live in a world that is truly evolving to unified communications — brought together by IP,” said Seamus Hourihan, vice president of marketing and product management for Acme Packet. “In all areas of the network — enterprise to cable to wireless to traditional fixed line — IP is happening, and more than that the convergence of data and voice and video over IP is happening.”

Not that the acceleration of IP isn't causing new challenges: Carrier-grade security and service assurance remain omnipresent concerns, as does address exhaustion, which may be finally heralding the long-awaited move to IPv6, the next generation of the core Internet protocol.

So as a new decade dawns and carrier networks stand ready to face new challenges and take on new opportunities, we decided to take an in-depth look at the “State of IP,” circa 2010, focusing on two key areas: the network and application layers. Here's what we found.

THE NETWORK: IP IN THE CORE — AND BEYOND

Both traditional incumbent carriers and newer upstarts are betting their businesses on IP everywhere-style strategies.

For instance, in December nationwide CLEC Bandwidth.com formally launched FlexNetwork, an all-IP network that it says enables IP-based voice origination for a variety of “voice 2.0” players, like Voxeo and Ifbyphone — not to mention Google Voice. By not only being all-IP but also fully exposing external voice application programming interfaces, Bandwidth.com is basing its entire business model on the idea “that voice is simply an IP application that can be mashed up in many new and interesting ways,” said T.R. Missner, chief technology officer for Bandwidth.com and architect of its FlexNetwork.

Not to be outdone, incumbents are systematically moving IP into all corners of their existing networks — made all the more daunting, of course, by having to retire, or simultaneously manage, their TDM circuit-based networks.

Today, for most global, Tier 1 carriers, IP already rules the day in the core of their networks. But the transition to IP isn't stopping there. Consider Verizon, for example.

“It's clearly not going to happen overnight, but we're always looking for areas and ideas for how to transition our voice infrastructure. Internally, a lot of our traffic is already on IP,” said Prodip Sen, director of packet network architecture for Verizon. Beyond that, Verizon has announced and is busy working on an IMS architecture that will underpin all its future services — including voice. Meanwhile, Verizon's much-touted FiOS today delivers its video-on-demand service via switched IP, and while its vanilla cable channels run over cable-style QAM, Verizon has set an all-IPTV migration path, Sen said. Converging all services onto IP allows for new blended services, he said, as well as underlying cost savings due to decreased operations and maintenance expenses.

Despite the costs and technical challenges associated with retiring TDM networks, the time has come when the advantages of IP are beginning to outweigh those concerns.

“Efficiencies and efficiencies of scale are the primary thing providers are looking for,” said Jay Wilson, senior vice president and general manager for Adtran's carrier division. “Even when we're doing some things like TDM-to-IP legacy service migration for them, the faster we can get it to IP the happier they are. They want to push that to the edge as fast as they can. From a carrier perspective, the biggest thing is efficiencies, not necessarily just efficiencies of scale but efficiencies of bandwidth and ease of provisioning, improved visibility.”

On the residential side, driving IP deeper into the network is part of an ongoing “major refresh of the subscriber edge,” said Lindsay Newell, vice president of marketing for Alcatel-Lucent's IP activities. Specifically, legacy broadband remote access platforms, implemented largely to drive basic DSL service rollouts — and supporting ATM aggregation and ADSL termination — are being replaced by new broadband network gateway access technologies optimized around IP, Ethernet, VDSL and the delivery of high-definition-capable, more real-time triple-play services, Newell said. On a box level, that means a lot more multiservice routers on the network edge; on a protocol level, it involves moving to more IP-native DHCP tunnels and IP traffic management techniques focused on a new class of multiservice metrics, such as latency and jitter — in short, “really a much more sophisticated set of higher-capacity, more real-time IP traffic control,” Newell said.

If anything, the move to IP — and IP-centric technologies such as MPLS and Ethernet — is happening even more quickly in the enterprise, where businesses “are trading out T-1, ATM and frame relay as fast as they can,” said one source. In the enterprise, the move to IP is both a basic network connectivity change as well as an enabler of an added level of intelligence in the network cloud (more on that later) that ultimately gives businesses a better, more managed level of communications services and applications.

One of the next big steps is pure IP-to-IP interconnection between networks. In the enterprise, this idea of keeping IP as IP is leading to a boom in business session initiation protocol (SIP) trunking services, where former islands of IP PBX connectivity are now connecting to PSTN IP backbones via IP and SIP, improving the economics of enterprise voice and laying the groundwork for more advanced, end-to-end IP voice features and services.

What is more traditionally thought of as interconnection — peering between carriers — is also moving to an all-IP environment, albeit a bit more slowly. Next-generation wireless operators may end up leading the way in IP peering, but traditional wireline carriers and, especially, cable operators also see value in passing VoIP traffic onto peers as pure IP. “What is lagging, quite frankly, among all this talk of peering and interconnect at the IP level is that Tier 1s aren't doing it in any big way right now,” said Acme Packet's Hourihan. One initiative in this area is the GSMA IPX, which aims to create a standard, global way to exchange IP traffic among mobile and fixed operators — though it remains largely in the trial phase.

For IP interconnect, it may be the case of too much of a good thing, said Jim McEachern, manager of service enabler standards for Nortel Networks. “There are core standards for [IP interconnect] in place, but there are too many of them,” he said, adding that there are activities underway, at standards group ATIS and elsewhere, to profile IP interconnect points and better define how IP traffic can be exchanged. One challenge to IP interconnect is that while TDM settlement fees are well established, many carriers today peer IP traffic with no settlement fees at all — one back simply scratches the other. That must change before IP interconnect becomes the norm.

Finally, wireless. Today, mobile services such as voice and SMS are delivered almost exclusively via circuit-switched networks — in the radio access, mobile switching center (MSC) and between networks and backhaul points. With 4G technologies like LTE, all that begins to change. Not only do mobile switching centers get IP-enabled, but traffic exchanged between MSC locations, not to mention between networks, begins to move to IP as well. IP even starts to penetrate radio access networks and clients. The move to IP in the mobile network demands an opus all its own, and luckily in this issue we have one (see “Core Wars” on page 20).

Just as big is mobile backhaul, where specialty backhaul providers and wireline carriers providing backhaul services are beginning to move to IP and Ethernet to deliver not only greater levels of bandwidth, but more flexible bandwidth as well. Backhaul provider FiberTower, for instance, is working with Adtran to move its backhaul services from TDM to Ethernet, said Vijay Lewis, director of engineering for FiberTower. “Ethernet today has very low penetration at cell sites; the reason is very cheap T-1s bought on five-to-seven-year terms,” he said, adding, however, that growth in data services means that “T-1s are a quick fix but not the long-term solution. 2010 is shaping up to be the year of Ethernet backhaul. I said that in 2007, but I was wrong. But now is the time. Mobile service operators have to add smart capacity. [For backhaul provides], that means it's not about adding T-1s, but about having Ethernet and being able to change capacity on a dime.”

The major challenge for backhaul is the timing requirements to put voice traffic on Ethernet. For that reason, many backhaul operators are keeping T-1s in place to handle voice and moving in Ethernet to support booming data traffic. Another option is to integrate both by running Ethernet over Sonet where possible, with the synchronous nature of Sonet transport mitigating those voice-timing issues.

For vendor Adtran, IP and Ethernet in mobile backhaul is all about helping carriers move to an integrated access infrastructure serving residential, business and backhaul businesses as efficiently and with as few boxes as possible. “You don't want to build out an expensive fiber-to-the-node network into a neighborhood and not take care of the business across the street,” said Mike Martin, director of product management for carrier solutions for Adtran. “If you have a cell tower right there, you don't want to run a separate fiber to it, but use the [gigabit passive optical network] you already have and mix residential and business and mobile traffic. You've got to make all these services work and play together.”

Indeed, IP in the carrier network is ultimately about convergence, a long-term “wish list” goal that may be beginning to come true. While some service providers view it as risky, most fixed-line carriers — residential, mobile and wholesale backhaul — are “very willing and able to carry all that traffic in a single aggregation network,” said Alcatel-Lucent's Newell, adding that the ultimate goal is for a variety of access technologies — especially wide-area wireless like LTE and femtocells in the local area — to bring traffic onto carrier IP networks and ultimately the public Internet as quickly as possible, drastically reducing costs and radically flattening carrier networks. “It's quite a hot topic, the proposition of a converged IP network capable of handling one, two or all three service types — it brings huge benefits in provisioning, integration, training, operations and maintenance,” Newell said. “You could well start seeing that type of integration, with LTE in particular as a trigger for these converged [IP] networks.”

APPLICATIONS: NETWORK INTELLIGENCE AND KILLER APPS

Such all-IP converged networks, trafficking solely in data packets and running on ever-cheaper, more standards-based software and hardware, turn tomorrow's central offices into something more like IT data centers. The next step is to apply data center-centric ideas such as virtualization to the carrier network, providing even greater levels of flexibility. “One of the things we can begin to do [is help the carrier] carve our resources using virtualization within a secure domain router,” said Suraj Shetty, vice president of service provider marketing for Cisco Systems. “You get one set of resources supporting wholesale IP, another set of resources carved out to support a Tier 1 enterprise customer and another set for services on the public network. In a virtual machine, data center kind of approach, it's a way to bring more network resources to bear on the fly.”

Cisco competitor Juniper Networks has talked about the future IP-based, virtualized central office in much the same way — and traditional network equipment vendors are moving in that direction as well.

Working hand-in-hand with those greater network efficiencies, IP changes the carrier application equation. Technologies such as DPI help carriers better manage the IP packet flow over their networks, while policy servers let them construct business rules to better manage those networks, the applications that run over them and the customers that access them. In short, with IP “the network is becoming more application-aware, and applications are becoming more network-aware,” said Jonathon Gordon, director of marketing for DPI vendor Allot Communications.

These new IP tools of the trade are already being used in the mobile arena to help operators get a handle on exploding mobile data growth by implementing more flexible and real-time billing and charging schemes (see “Mobile Billing Blows Up” on page 28). That same infrastructure can be leveraged in a variety of additional ways — most notably to deliver and sell (if net neutrality regulations allow it) quality of service guarantees to enterprises, consumers, and third-party content and application providers. “What you end up with is a lot more transactional decision-making happening in real time on the edge of the carrier network,” said Joe Hogan, chief technology officer for IP policy and transaction platform vendor Openet. “This is all going to evolve quite rapidly and drive carriers into what amounts to a very large-scale transactional policy business. If service providers don't have a platform that is attractive [to partner with] over-the-top providers, they are going to become the victim of them.”

The interesting thing about the recent explosion in DPI- and policy-based transaction platform deployments is that such elements are well within the landscape of big-picture IP application architectures like IMS, which otherwise has taken eons to move into service provider networks. But while the smart-IP-box approach is about quick wins and fast deployments, IMS remains about building out an entirely new software-based and service-driven architecture for carrier IP networks — a goal as necessary, and as challenging, as ever. The fact that IMS isn't simple to understand or implement is both its strength (in that complexity lies its promise and potential power) and one of its obvious weaknesses, said Elisabeth Rainge, analyst with IDC. In 2010, however, the ultimate promise of IMS remains essentially unchanged: it's an IP-based core network made up of service, control and transport layers; it's a standard (or standards); it's a next-generation network architecture; and it's an enabler of new services. Even today, as IMS becomes more real every day, “when we think about what IMS is, it's by no means a specific class of products,” Rainge said. “It remains a concept that takes on many different flavors as it has evolved and is adopted.”

Whatever the specific deployment approach, IMS is largely boiling down to two critical pieces: the call session control function, enabling signaling and transport; and the home subscriber server (HSS), for dealing with customers on the network. In emerging LTE architectures in particular, “HSS deployments and investments are already very clearly underway,” Rainge said. Meanwhile, exactly how providers deal with signaling and control layer issues in their IMS deployments has emerged as a clear differentiator between various provider implementations of IMS.

In the end, the outlook for IMS is good, Rainge said. The architecture aligns with important service provider goals, including green/sustainability (IMS runs on cheaper, cooler servers for the most part), system administration improvements, and the ability to deliver better services cheaper and faster — not to mention it supports the convergence move toward IP. “There's not a Tier 1 out there not implementing IMS,” said Acme Packet's Hourihan.

Adding a new twist to the IMS story is recent momentum around RCS, a key companion technology for delivering new services via IMS. In some quarters, RCS has been met with a fair amount of skepticism, as critics wonder if a lengthy, specification-by-consensus approach to new service development can compete with the rapid-fire mobile services world driven by the iPhone, Android and other more developer-centric platforms.

Yet for mobile operators to compete they need some competitive advantage. Mobile device- and operating system-makers benefit from being able to target a global audience; in comparison, mobile operators are typically regional-based. The counter is to develop a suite of RCS services that is able to operate between carrier networks — something that companies like Apple can't guarantee. It's just that level of interoperability on a network/service level that drove mobile's biggest success story, SMS. RCS services center around a networked address book profile based around the customer's mobile phone, as well as services such as mobile instant messaging, which take advantage of that greater knowledge of the customer profile, location and state. The recently debuted RCS Release 2 adds some key enhancements to the specs, most notably support for delivering services to broadband access clients as well as support for multidevice environments. Both of those capabilities potentially make such RCS-based services as in-call multimedia sharing, conversational messaging and presence-enhanced contact management all the more powerful.

In the end, IP in the carrier network has the potential to be greater than the sum of its parts, bringing efficiencies to the core, convergence to the access edge, and new application control and capabilities out to developers and customers.

And that's what makes “the State of IP,” circa 2010, so promising, indeed.

THE IP BILL OF RIGHTS — AND WRONGS

These IP truths we hold to be self-evident: The migration to a (mostly) IP-based PSTN brings with it a lot of uncertainty - and a few truths for those brave enough to put a stake in the ground. Acme Packet's Seamus Hourihan, an IP veteran, shared with us nine “rules“ that he said sum up his view of how yesterday's public circuit-switched networks will collide with tomorrow's IP networks to create something new altogether - warts, blemishes, the promise of a new network, and all.

• In IP, we trust no one.
• Addresses will forever be a collection of heterogeneous schemes.
• SIP is not the only, nor a single signaling protocol.
• Codecs will never converge to a couple - audio & video.
• Unlimited bandwidth, QOS and signaling resources will forever be a myth.
• Some sessions are more important/valuable than others.
• No one can do anything they want.
• IP regulatory compliance requirements will increase.
• Service provider business models will never be homogenous.

IP-BASED IMS DRIVES INTELLIGENCE, APPS INTO THE NETWORK

IP multimedia subsystem architectures are moving from novelty to must-have, as IP drives deeper into the network. Carriers have moved from viewing IMS as a would-be standard to a next-generation network architecture to an enabling platform for new services.

RACING TO (IP) EXHAUSTION

IPv4 may finally have seen its last day — no really, we mean it this time.

For almost as long as Internet Protocol packets have been finding their way into service provider networks, the next generation of IP — IPv6 — has been said to also be right around the corner.

Created more than a decade ago to address the so-called “exhaustion” of available IPv4 addresses, IPv6 features not only a larger address space (128 bits versus 32 bits for IPv4), but simplified management, better integrated security features and more. Network address translation, or NAT, in the enterprise has extended the use of IPv4, and more recent carrier-grade and network-embedded NAT solutions added even more life.

“For years, vendors were told if you didn't support IPv6, you couldn't get in the network,” said Keith Landau, chief product officer for Genband. “But it wasn't really true. It was like when they'd say you couldn't get into college if you didn't know a foreign language. There were always ways around it.”

But with mobile and smartphone use exploding right now — not to mention machine-to-machine IP connections proliferating even more rapidly — the time may have at last come for telecom network operators to take the IPv6 plunge, experts say. By most accounts, available IPv4 addresses are likely to run out sometime in 2011.

“With IPv6, it's not a standards or technology issue; it's a question of timing and investment and when exactly do you pull the trigger,” said Jim McEachern, manager of service enabler standards for Nortel Networks.

While diminishing-return NAT tricks and exploding device usage are impacting the supply side of the IP equation, the demand for IPv6 is finally starting to grow as well. Several governments, most notably in Japan and Korea, have put strong incentives in place to drive the move to IPv6, while the U.S. Department of Defense has mandated its use moving forward. Meanwhile, Google IPv6-enabled some of its Web apps last year, while Comcast is planning a major IPv6 trial in 2010.

“There aren't going to be flash cuts” from IPv4 to IPV6 networks overnight, McEachern said. “You're going to have interworking between IPv4 and IPv6, for practice purposes, forever. Initially, you'll have islands of IPV6 in a sea of IPV4 and then vice versa.”

One of the challenges in putting off the move to IPv6 for so long is that you essentially today “have a bigger [global IP] network, so there are more things you have to touch to upgrade from v4 to v6,” said Lindsay Newell, vice president of marketing for Alcatel-Lucent's IP activities. “But the time has finally come.” — Rich Karpinski

COUNTING DOWN TO IPv4's END

The number of IPv4 addresses is steadily exhausting, driving the need for enterprises and carriers to move to IPv6. Counting down the days (and address spaces) are a variety of meters, counters and — fittingly enough, given that the mobile device proliferation is speeding up the migration — iPhone apps.