Click here to Subscribe

The NGN conferences, the Venture version in the spring and the “ordinary” conference in the fall, are always a sanity check on the industry. For the last three years it has been bleak. John McQuillan did his usual good job of laying out the stark reality. The message that came through today, in these, the darkest days of this industry, is that the brave are blazing new trails on the backs of failures. Certainly one that stood out is Cogent. In only three years since it began, it is handling 7% of all the US Internet traffic. It has 5,000 customers in 52 markets in the US and Canada using a 12,500 route mile inter-city fiber network coupled with a 7,400 fiber mile intra-city network on 139 metro fiber rings in 23 cities. It could be said that Cogent is a scavenger, having acquired the assets of PSInet for less than a penny on the dollar, but the customers of Cogent are not complaining about getting 100Mb/s Ethernet connectivity for $1,000 per month. The Cogent CEO, Dave Schaeffer, was proud to state “We have only one product.”

http://www.cogentco.com/

But even Ralph Ballart, VP, Broadband, SBC Technology Resources, began his talk by stating, “I still have a job.” There is no escaping the reality that telecommunications is undergoing a massive overhaul. The success of optical transmission technology is a deflationary technology. The revenue source remains voice but this market is in turmoil as competition arises with cellular, voice on cable, and VoIP. As the rate of voice disconnects increases, the ability of this cash cow to sustain the margins of the business declines. The traffic growth is in data but the cost per bit is a fraction of what is paid for voice. All of this creates massive forces of upheaval. The telecommunications industry of 5 or 10 years ago just does not exist anymore.

At 30,000 feet one sees the fault lines on the ground, but it is the different plate tectonic settings below the surface that causes the earthquake. In telecommunications the analogy is that optical, data and the economics of both are changing the landscape of the business while the legacy companies hold on to voice and their 1,000’s of products. There was $3.1 trillion spent in the last decade in telecommunications and most of that has been lost. Even at WorldCom, which wrote off $80b in assets, $40b of that was in property, plant and equipment which was now valued at $10b. This meant that the company paid $1 for what is now worth 25¢. With this overhang of over-valued equipment, distressed companies, and technology without anyplace to get used, are stranded. The ILECs hold on to old business models and regulatory protection while new companies desperately seek the first occurrence of cash flow and to reach a cash positive situation. Even Cogent has most of its Cisco equipment based on loans that Cisco has made to the company for the equipment purchase. The earthquake analogy holds – the ground is moving and no one knows what will remain when it is over.

Industry Overview

It is bleak. There is no other way to describe the status of telecommunications, especially venture investing. John McQuillan provided his overview, which included the following:

In terms of the IPO market, 2002 was awful and 2003 has been worse. The IPOs in 2002, for this sector, were only $266m. In 2000 the total IPOs raised $18.7B while this was only $1.5B, less than 10%. There were 8 weeks in early 2003 without an IPO, the worst gap since 1975.

In terms of M&A in 2000 the total, in this sector, was $101B, $21.5B in 2001 and $9.6B in 2002. Another 1/10 factor.

In 2002, $106B was venture invested, which dropped to $41B in 2001 and went to $21B in 2002. In the last quarter of 2002 it was only $4B. This marks 11 consecutive quarters of declining investment. Equally as important, 80% of the money is going into follow-on rounds. Only $800m went into new deals.

The venture investment today looks like 1995. The investment rates are similar to 1995 and the number of companies invested in is close to the number on 1996.

One of the most difficult situations is the capital overhang. That is, the venture capitalist have money which has nowhere to go. In 2001 there was $105B. In 2002 only $6B was raised and there was $6B in fund reductions. In 2002 $21B was invested but $75B remained, which is called the overhang. The lack of the IPO market and no M&A’s mean there is no movement of money while large pools remain stagnant.

Overview of the Presentations

On the first day the sessions included:

Next Generation First Mile Solutions: Broadband Access Products
New Directions in Metropolitan Networking: Meeting Real Needs
Emerging Optical Architectures: Enhancing or Replacing SONET/SDH
Service Provider Customer Roundtable

The talks can be summarized by the following:

Our equipment is much less than what you paid before;
We will save you on operational expenses; and
Our equipment enables you to offer new services that generate revenue quickly.

Yet, in the Service Provider panel it was clear this was not enough. The service providers expanded on what is missing:

We need system solutions not boxes;
We will buy from start-ups but we need assurance of long term support; and
Who are you partnered with, in terms of a long term and established supplier, that will provide us with assurance of support?

Thus, in spite of high bars being set for these companies to sell to the incumbents, they need to be a part of the overall system and supplier chain.

High Points

Here are some threads we picked up from the presentations.

There are two camps: Old and New.

The Old is represented by the ILECs, IXCs and MSOs. If the Old is an ILEC they have the advantage of cash flow from voice. AT&T is an IXC but with UNE-P they are a “new” entrant into (local) voice. From a business perspective, the Old players are burdened with two components: their networks and the product line. The Old continually seek to make use of the large capital investments they have made, almost no matter how inefficient. Certainly one of the best examples is the use of SONET for metro Ethernet. The burden of the product line is also a retardant for innovation. Hardly anyone buys Frame Relay but there is still a large installed base of users. Thus, it is an oxymoron to think of a “legacy free ILEC.”

The New is represented by Covad and Cogent. These companies do not care about anything but just their product, which is usually one deep. This allows for focus, driving cost out and very selective marketing to achieve an ROI. From a user standpoint, if one intersects with these companies, such as being in the right building or geography, the gains are large. In the case of Covad, it means getting DSL from a CLEC, such as AT&T. In this case, AT&T has a huge value in offering DSL and bundling all the services it can to lower churn. For Cogent, it is being close to a fiber ring or in a building that has fiber. If this is the case, the gain is also huge. Your connection to the Internet or network, is just like the rest of the LAN.

A shift in optical networking is underway. The optical network is becoming protocol agnostic. That is, optics provides transport only. The only purpose of switching is for restoration. Thus, mesh topologies are valuable, even on a national scale in that they allow for redundancy and rapid restoration. A major cost driver is in OEO, where the digital signals emerge. GMPLS has made significant headway in adoption. GMPLS has an advantage for optical transport in that it can serve as the control plane for the network. The technology which is gaining share for optical switching is MEMS. This has the advantage of being simple, in much the same way as TI’s DLP is to displays. MEMS switches do not have to operate on the content of the pipes.

AT&T presented a different view of the network. They described a network which is PnP by the user. This fits their concept of blending the consumer and business with the devices they have and the network. As a result of this, AT&T showed a network which had the users at the center, not on the edge. Discussion ensued on the parallel between this and the PC industry. For example, the need was cited for a Dell, as a systems integrator, in the telecommunications industry. As more of the network surfaces to the end users, there will increasingly be a need for a company to “pull it all together.” As with the PC some companies and individuals will be able to manage their network needs personally and others will need help. I was impressed with the depth that AT&T is bringing to the definition of where their network will evolve to.

SBC stated that >90% of its DSL installations are self install. That is, SBC mails the customer a DSL modem and the consumer brings up the service.

SBC has its first deployment of FTTH. This uses PON in the Mission Bay planned community in SFO. There are actually two channels on the fiber: 622Mb/s for data and a 800MHz channel for cable emulation and video transport. SBC was also singing its request for regulatory relief and waiting the final rule making on UNE-P.

The Challenge of SAN

SAN is one element of the network that got a bump up from 9/11. Storage back-up and recovery is now much more important to the enterprise. But, the market and technology issues are deeper than the after effects of these horrific events. In the larger context of computing, as the network performance approaches the bus bandwidths in the computer, the computing components disaggregate. No longer is there a question “Do you know where your disk drive is?” but “…is my data safe in multiple sites so that continuity of the enterprise assured?”

John McQuillan began by characterizing the challenges which include:

Need to: reduce complexity and inflexibility of configuration, improve performance and lower the cost of SAN;

Need for low latency, guaranteed and deterministic performance and high availability in SAN; and

Need the ability to support longer distances and features such as mirroring and back-up.

Candera stated its product, in beta testing now, is a network storage controller that works with all the data, switches, storage and servers. The result is a single interface to all the storage on the system. The objective is to achieve an SLA storage utility model.

Akara took a very different approach in that they are building a hardware software solution of carriers, assumed to be ILECs and IXCs which allows for SAN over SONET. This seemed odd but the speaker stated that SONET’s advantages of high bandwidth, low latency, high availability and security made SONET a logical choice. Contrary to much of the current thinking, the comparison was with IP based SAN, which the speaker described as ineffective compared to SONET based SAN.

Lastly ONStor described an approach which uses a box called a SAN filer that lies between the customer equipment and the SAN. These filers essentially normalize the SAN storage so that a single point of management contact is enabled along with unified access control.

The board level discussion, which included VCs that have invested in this space and these companies, was interesting.

The dominant company in storage is EMC. The opportunity is based on the speed with which these start-up companies can move into this space before EMC does.

One of the major challenges is that EMC is very customer focused. A parallel was drawn with Cisco, which is also customer focused. Thus, it is seen that these companies dominate their respective markets because they are agile to the demands of the customers. For a challenger in SAN the new companies must find niches in the market not being exploited and drive a solution faster than the dominant players.

Given that the only exit model for VCs today is M&A this gave rise to a discussion on the climate for M&As.

For a long period the M&A environment was poisoned by the poor success of acquiring companies. An example held out for how not to do this was IBM’s acquisition of Rolm, a PBX company, in 1984.

Yet, the view of acquisitions has changed with the success of 3Com in acquiring companies and most recently Cisco. Thus, the ability to manage an acquisition is seen as an important strategic advantage.

The role of acquisitions, as a means of acquiring business, talent and core competency, has become more favorable based on the qualities of the acquiring company and how well they manage the process.

VoIP

Based on the VoIP session today one would conclude this is a done deal – VoIP has won. However, other conferences we have gone to indicate quite the opposite as VoIP struggles in making a compelling ROI case to the enterprise. Converged voice and LAN networks spell a disaster waiting to happen for many IT managers and thus the only way to implement them is another wired LAN infrastructure. Hardly cost effective.

Yet, in our discussion with the President and CEO of Netrake, it is clear that VoIP is fundamentally an arbitrage play which goes well beyond today’s implementation. If data traffic costs 1/10 of voice traffic, the arbitrage argument goes, why not use data pipes for voice also? It remains to be seen if these economics preserve themselves when the data network is used for voice. From a market perspective, the ILECs live in fear of a VoIP service provider rolling into a Fortune 100 enterprise with a compelling service proposition, especially for CENTREX users. When I asked “…who is pushing this bypass strategy” the answer was: both emerging VoIP carriers and some enterprises. Thus, the potential is for VoIP to disrupt the existing revenue streams of the ILECs from some of their major customers. Now that is real competition.

The industry is facing the prospect of making IP a telephony transport, which it was never designed to do. It is one thing to make a SIP phone call and something much more expansive to handle 10,000 phones in an enterprise over the LAN infrastructure and handle all the media and calling functions. From this need has arisen the requirements for a session controller. The board level discussion which followed provided context. The market question is:

Can these start-up companies define a market and execute on that market faster than the companies supplying VoIP products today? For example, Cisco is a major player in the VoIP enterprise space and they certainly see the need for session controllers. Thus, to many it is a race to market.

A session controller is basically a Layer 5 router. In the IP network, session communications takes place in Layer 5, such SIP and H.323, while the media which includes voice, video and date happens at Layer 3. The session controller provides QoS management, security management for both SIP and H.323, protocol conversion and interoperability between vendors. As we have heard at VON, VoIP has problems passing firewalls and the session controllers are to address this. Further, due to the characteristics of the internet and peering it is possible for one VoIP service provider to look into the network structure of a competitor to see what network that competitor has built – a session controller blocks this. Another function of the session controller is to circumvent DoS attacks that could block telephone service.

Because the session controller must do deep packet analysis and billing on a micro scale it has the potential to do much more. Some of the applications include gaming support, VoIP push to talk (VoIP IM), and multimedia security.

Netrake has a carrier class session controller which handles Carrier to Carrier VoIP and Carrier to Enterprise VoIP. It claims to have a very high call rate and thus fits the carrier requirements.

NexTone has a session controller that supports multi-protocol signaling. It uses the SIP “b2bua” or back to back user agent. This allows for gateway call signaling, which are channelized through one or more centralized b2bua’s. In order to provide isolation, this session controller actually terminates the call and then regenerates it on the output. NexTone makes the distinction between a Core Session controller and a Border Controller.

I asked the CEO of Netrake why the term was border controller and not edge, and in response what I got was – new terms get defined as markets evolve. Put in another way a border controller is an edge device.

Kagoor Networks suggested that its border controller replace a route controller, NAT, QoS probe, firewall, and traffic shaper. Ambitious.

We take the emergence of these new devices as a sign that the VoIP market is getting serious. Yet, in the follow-on enterprise roundtable it was clear that the enterprises still have major questions on the ROI of VoIP. The recurring issues of enterprise transport still are of concern. One speaker spoke of the need to completely redo the LAN infrastructure of the enterprise in order to avoid performance degradations to either data or voice traffic when they are both supported.

Wireless Enthusiasm Abounds

It is incredible what a 30%+ growth market will spawn. In a conference where survival is a daily activity, the emergence of an exciting market in wireless, including Wi-Fi and 802.11, has the entrepreneurs coming out of the walls. 3 months ago there were 7 wireless switch companies and now there are 10.

The four talks we heard are but a sample of what is taking place in this market and technology.

Flarion

The CEO, Ray Dolan, gave the presentation. Flarion is seeking to fill the data gap left by the poor performance, both market and technology, of 3G as a data delivery medium. Flarion has a proprietary chip technology for mobile devices, specifically targeted to carriers to offer wireless IP data services. It uses 1.25MHz of FDD radio spectrum and OFDM modulation, which they call OFDM-flash. Downlink is claimed to be 3.4Mb/s with the uplink at 1Mb/s. It was interesting that Flarion called voice just a thin application on the network and radio.

Today there are four products: RadioRouter, a base station, FlashView, the element management system, PC 1000, a wireless network card, and FLR 1500, the Mobil Baseband Chipset. The system is claimed to operate in any spectrum up to 3.5GHz.

Flarion touted its 40/40 rule. The users pays $40 for unlimited use and all the players in the food chain get 40% gross margins – OEMs and operators – on a nationwide network. The $40 unlimited usage was limited in the total data that could be sent in a month, however, the amount seemed reasonable. They claim the system is in 3 trials in Korea and all the major wireless operators in the US are looking at it.

Apparently the 802.20 working group has been launched which addresses this technology. However, Ray stated that at the last meeting the 3G crowd showed up and this may slow the works.

Aruba Networks

Pankaj Manglik, President and CEO of Araba, stated that the future of Wi-Fi was based on his wireless switch. This switch is to be the parallel to the first RJ-45 Ethernet multi-port hub, which allowed a hub-and-spoke implementation of Ethernet--the foundation of today’s successful implementation of plug and operate Ethernet. The Aruba switch offers security, detection of rogue access points, self-healing, and load balancing. The functions claimed to be inside the box are: site survey, PoE switch, intrusion detection, mobile security and terminal server.

At 802.11 Planet we found that the needs of the enterprise in 802.11 would drive their own requirements and thus the equipment suite. This wireless switch is the first example of that. During the boardroom discussion several points were raised.
It is not clear the switch will exist as a stand alone box or as a feature in other Ethernet switches, in fact, Nortel claims it has such an enhanced switch which supports wireless needs; and

In order to be effective all the AP must be plugged into the same switch which could mandate wiring changes.

In conclusion, Pankaj showed the mobile network world and at the center was his switch – not surprising. The points of service include: enterprise mobile networks, public WLAN hot spots, mobile applications and then public mobile networks.

Vivato

The CTO, William “Skip” Crilly gave the presentation on this innovative antenna technology. This is called a switch but is really a beam forming phased array antenna. The key is that the antenna is used to form beams on a packet by packet basis. As a result the effective power to a device is higher. Vivato states that the FCC has approved this and it is the basis for their first products. The advantages include: multiple Wi-Fi transmissions simultaneously, ranges in the km, no changes in the clients and the ability of one switch to serve a building from the outside.

The antenna has 3 beams of Wi-Fi that are 7 to 9 degrees horizontal and 12 degrees vertical, which are distributed over 100 degrees of horizontal FOV. There are 128 elements in the phased array antenna. The gain is 25dBi. Input to the antenna is 2 1000BaseT and 2 10/100 BaseT ports. Range for the indoor antenna is 300m and the outdoor unit up to 4,000m. The antenna handles rogue detection and multiple VLAN security including PPTP and IPSEC,

The two products are the indoor unit for $9,000 and the outdoor unit for $14,000. An outdoor deployment can be used to radiate a building and thus get building coverage. In a building one antenna is claimed to cover a whole floor.

This product is an interesting combination of antenna design and underlying switch silicon to handle the switching. This is a competitor to Aruba but it also carves out a unique space by implementing the switch within the antenna.

Airgo

It was not clear exactly what the product is they are offering. However, the technical thrust is MIMO antennas with DSP to negate multipath effects in Wi-Fi for buildings. This is apparently also a switched beam approach. The client does not change. However, they claim 18Mb/s at 100m in a mini-PCI card.

Boardroom Discussion

This was interesting and with the following points:

When asked why none of these speakers discussed mesh networks it was stated that the success of meshing is open. For example, 4 years ago Nokia bought a mesh topology company called RoofTop and they worked some 3 years on the technology, ran tests in California and released the product in 2001. It appears to have gone nowhere.

It was strongly suggested that an implementation with one antenna, such as what Vivato has, is not the best way to go. The key disadvantage is the lack of redundancy.

From a carrier perspective, the value of Wi-Fi comes not in hot spots but geographic areas – a complete shopping mall or a full campus. We are just now seeing the technologies that could make this happen.

Nortel: At Nortel we intend to use mesh networks in our products. There is high value in having intelligence in the AP. This means that the network is self configuring and self healing. Networks must be able to adjust themselves.

Headend Approach to the Last Mile with DSL

Celite Systems takes a very different approach to broadband to the home. Basically they use DSL as a delivery technology to the node near the home, in this case the SAI (Serving Area Interface) box. With this they claim to provide 10Mb/s Ethernet to the home. Note that this is shared bandwidth. The panel needs to be within 6,000’ from the home. The striking element is that the CPE, which is an Ethernet bridge costs only $50. The technology details were light, however.

What was quite important is that Celite claims to address the price problem with DSL. Only by driving the prices for DSL service significantly lower, to $30 and less, will the penetration continue to rise. It is their premise that the early adopters have been reached and only with price reductions will penetration continue. Right on.

However, we wonder if the message that Celite has will ring with the ILECs. The reason being, is that the incumbents must change the way they do business. Just plugging DSLAMs into a rack is not adequate to increase penetration. To think that another approach would be required may not fly. It should be noted that Celite appears to have the early customers overseas, possibly in Asia.

Comments?
E-mail webmaster
Page updated 1/24/07
Copyright 4th Wave Inc, 2007