Cable IP Networks, VoIP and the Enterprise Challenge

MSOs have been upgrading their networks and building out to new customers for years. The result is a network footprint that today has excellent coverage and a technology that is superior to the basic two-wire loop currently deployed by the incumbent local exchange companies (ILECS).

MSOs have done a great job of leveraging this buildout and of competing with ILECs for basic residential telephone services. Using hybrid fiber coax (HFC) technology, MSOs have been able to duplicate the basic two-wire, residential, plain old telephone service (POTS) offered by the ILECs. Yet, also on that same coaxial cable, they are offering high-speed broadband Internet services.

Conversely, the telephone companies have been at a distinct disadvantage, as their two-wire loop digital subscriber line (DSL) technology simply cannot compete with HFC broadband IP. The speed just isn’t there. That puts the MSO in an interesting position, and it begs the question “so what’s next for the cable companies?”

An ILEC’s highest-margin customer is its business client, a market cable has yet to succeed in capturing on a large scale. But it could. MSO networks have the bandwidth and the core technology in place to compete aggressively with an ILEC for that business user. Before that takes place, some key issues must be addressed.

Cable’s Next Step

Business telephony users are different from residential telephony users. Organizations larger then eight employees generally use either a key system or a PBX. This allows the enterprise to take advantage of trunking ratios, whereby there is a ratio of trunk circuits to the telephones that is less than one; i.e., there are fewer trunks to the phone company than there are numbers of telephone stations deployed inside the organization. This ratio varies depending on the business, but ratios as low as 4:1 stations per trunk or as high as 16:1 stations per trunk are common.

In traditional telephony, these trunking circuits generally are brought in via T1 circuit interfaces and then are connected to the PBX or key system. Here is where the business user differs significantly from the residential user.

The MSOs for the most part cannot deliver digital time division multiplexing (TDM) T1 trunks per se to end users. Rather, except for the larger cablecos, to date, MSOs only have succeeded in delivering two-wire POTS circuits, intended to connect directly to a telephone station and, thus, giving a station-to-trunk ratio of 1:1.

So the MSO must charge for that circuit as if it were a single two-wire-station telephone line because it ties up bandwidth space on the cable network and a station side port on its switch, regardless of whether it’s being used in a call or not — just like a telco’s real two-wire loop pair out on the telephone pole.

This 1:1 ratio is unattractive to the enterprise business user. It’s simply too expensive and doesn’t offer the features needed by PBX users to conduct business, i.e., Answer Number Information (ANI) and Dialed Number Information Services (DNIS) along with the ability to hunt as a trunk group into a PBX.

The CableLabs DOCSIS Solution

CableLabs has attempted to put forth a T1/ISDN solution using HFC, and some of the larger MSOs have experimented with it, but the technology is plagued with serious shortfalls.

Problem number one is that it’s expensive. T1 over DOCSIS HFC requires new (and expensive) hardware at the headend office to deal with T1/TDM technology. Further, it starts putting significant pressure on the cable network itself because it ties up bandwidth. T1 over HFC is nothing more then a circuit of another kind, regardless of whether it is being used to carry voice or not.

Perhaps the biggest drawback to T1 over HFC is that it’s already obsolete. Enterprise users are upgrading their PBXs to interface directly to voice over IP (VoIP) and specifically to session initiation protocol (SIP). Taking that into account, a T1 (as a telephone interconnect for the enterprise PBX) rapidly is becoming irrelevant. In its place, the enterprise PBX will want to take a direct RJ45 Ethernet connection and do all telephony natively via IP.

So, with these two facts — it’s an expensive proposition to deploy DOCSIS T1 over HFC and the T1 interface already is irrelevant — it would seem unlikely that this would be a technology the cable industry would embrace on a large scale to conquer the enterprise voice market going forward.

What About Hosted VoIP?

The IP segment of the MSO network is a perfect transport medium for VoIP and, to date, many cable providers have begun offering hosted VoIP services. Hosted VoIP is similar in nature to the old Centrex service, whereby the carrier creates a virtual PBX domain for the enterprise off its core end-office switch. Hosted VoIP does the same thing and thus negates the need to have a PBX on location at the customer premise. Nonetheless, there are significant issues with offering business users hosted VoIP services over cable IP networks.

The first issue is quality of service (QoS). Getting VoIP correctly over the last mile to the end user is tricky and requires VoIP engineering if that end user plans to merge bulk enterprise data with VoIP traffic over the same cable drop. If a QoS solution is not deployed, the voice calls will be pre-empted by the bulk enterprise data, giving the user an unsatisfactory experience, i.e., choppy, noisy calls.

Another major issue is how to traverse the corporate firewall. Hosted VoIP services must penetrate the corporate firewall but all firewalls are different when it comes to VoIP support, presenting immense technical and even political problems for the MSO to overcome.

Firewalls for the most part will require re-configuration by the end user to accept VoIP traffic. In many cases, firewalls either won’t support VoIP or the owner/administrator has no idea how to make that reconfiguration.

Finally, hosted VoIP simply doesn’t scale. Beyond eight users, hosted VoIP services lose their attraction. These products only offer a 1:1 ratio between trunk and station so, in the end, they are expensive services that also can tie-up corporate bandwidth quickly. Having said all this, hosted VoIP may not be a long-term solution for the enterprise; it’s simply too costly. Couple that with free PBXs (like Asterisk) running on almost free PC hardware the size of a toaster, and enterprises ultimately will realize there is no substitute for a PBX on the premises. Further, they will demand a PBX that can handle VoIP as a native signaling medium.

How About SIP Trunking?

The other alternative to HFC T1 and hosted VoIP technologies is SIP trunking. SIP trunking basically is a better virtual T1 over IP, with none of the channel limitations found in a T1. It can provide as many trunks on which to talk as there is bandwidth to handle them.

This non-channelized nature makes SIP trunking an attractive technology for the MSO because there are no physical hardware changes needed to handle small or large customers. All the operator needs do is to tailor the bandwidth to the end user to give the enterprise the voice capacity it needs.

Exo-Firewall QoS/Security VoIP Solution

Again, there are significant issues. Offering SIP-trunking services to customers over cable IP networks has some considerable technical challenges that must be addressed.

To offer SIP trunking services, the MSO’s back office still must be provisioned with switches. In the case of VoIP, these are soft switches as opposed to the traditional Class-5 end-office TDM switch technology used by POTS and T1 telephony. Granted, these devices are far less costly to deploy than are their TDM counterparts, but they must be deployed nonetheless. Further, there are billing systems that must be provisioned and interfaced to this core VoIP switching infrastructure.

These back-office infrastructure buildouts can be daunting for a small or medium-sized MSO, especially in light of the fact that many MSOs simply are outsourcing their current residential telephone needs to outside PSTN carriers. As such, many cablecos supply the base customer loop transport over cable and then let a third-party PSTN telco handle all telephony needs at the backend. This may be a major hurdle to clear because SIP trunking still requires the supplying organization to be fluent in telecommunications expertise. ( Note: To date, only a few of the larger MSOs even have experimented with this sort of a SIP infrastructure buildout, because it’s still a costly and complex undertaking.)

Perhaps the biggest obstacle to SIP-trunking deployment by the cable industry can be found at the customer premise. Just like other hosted services, SIP trunking still has the problems of:

Corporate Firewalls. All firewalls are different, are usually difficult to set up to handle VoIP, may not even support VoIP and/or its administrator may have no idea how to set it up for VoIP.

QoS. QoS issues only become worse with the higher call volumes SIP-trunking services invite.

Security. When offering SIP-trunking services, MSOs must protect IP Layer 5 to help ensure hackers and thieves don’t hijack the service and run up fraudulent bills on a customer’s account.

Enterprise Solutions

A simple solution to the corporate firewall and QoS issues is to avoid them all together by provisioning a dedicated IP cable drop just to service VoIP. That eliminates any contention between the enterprise bulk data and corporate firewall issues. It does nothing, however, to resolve security risks.

Separated dedicated drops for VoIP unfortunately are a costly solution, and one that small and medium-sized clients likely will reject. Not merging services over a single drop defeats one of the core economic reasons for moving to VoIP technology in the first place: converging communications networks down to one network to handle all media.

So what can the MSO do? To meet the challenges of QoS, dissimilar corporate firewalls and security issues, all working on a merged VoIP/bulk-IP drop, an “Exo-Firewall solution” (EFS) is needed at the customer premise (see Figure 1).

Ideally, the EFS solution would be a router positioned between the MSO’s IP cable modem and the corporate firewall. Its attributes would be as follows:

EFS in Bridged Mode. The router would be deployed in a Layer-3 bridged mode, unseen by the corporate firewall. In so doing, the corporate firewall would have no need for re-configuration or a change to accept VoIP SIP trunking on the same IP drop.

EFS QoS Solution. The EFS would mitigate and manage real-time transport protocol (RTP) voice streams and bulk corporate data such that real-time VoIP is given priority over all other traffic.

Security at Layer 5. The EFS must offer security at IP Layer 5 and be “SIP aware.” At a minimum, it must only authenticate, and allow traffic to flow to and from the MSO’s SIP switches while disallowing call signaling and traffic to anything outside that established SIP domain.

EFS Must Segregate VoIP Traffic. Ideally, the VoIP office stations (VoIP telephones) should be segregated onto their own IP local area network (LAN) segment and separated from enterprise LAN traffic. This is needed for QoS concerns as for security concerns. The EFS would be the mechanism to tie this private VoIP LAN domain back to the main public wide area network (WAN) via the shared IP drop.

In so doing, VoIP traffic never would enter the enterprise LAN and so never present a potential security threat. This is critical. At all costs, the MSO must avoid getting involved in any way with the enterprise LAN infrastructure behind the corporate firewall, both for political as well as technical reasons.

The MSO’s Clear Choice

If the major issues of QoS, disparate firewall interfaces and VoIP security can be resolved at the customer demark, SIP trunking clearly will be the technology of choice for the MSO and its enterprise customer. Some of the SIP-trunking advantages for the MSO operator include:

A path to the future. With VoIP and SIP trunking, expensive and antiquated circuit-based T1 equipment isn’t needed at the customer premise or at the MSO headend. This lowers costs for deployment radically and provides an excellent migration path for the future.

No bandwidth waste. Bandwidth over cable is better utilized with VoIP than it is with HFC voice; VoIP only ties up bandwidth when it’s being used. This has significant (positive) ramifications as more cable customers switch over to the MSO for telephone services.

CPE deployment is simplified. SIP trunking allows the MSO to handle any number of SIP trunks at the customer premise, up to the limits of the bandwidth available. This radically simplifies and streamlines installations.

VoIP handles the legacy PBX. SIP trunking easily can handle today’s legacy TDM PBX with SIP-to-T1 or SIP-to-analog conversion gateways.

SIP trunking is ready for the next PBX wave. SIP trunking instantly readies the MSO for native SIP/VoIP interfaces, such as those found on modern PBXs like those made by Asterisk.

What About Enterprises?

For the enterprise user, the advantages of SIP trunking are numerous:

Total Direct Inward Dial (DID) Freedom (Global Number Portability/Global Number Services). With VoIP, the end user is freed from the PSTN rate center and so can have any DID number terminate to any customer premise equipment (CPE) site, regardless of where that customer site sits geographically.

Low-cost, off-premise extensions. By leveraging the Internet, VoIP allows off-premise extensions to remote employees at almost no cost.

Low-cost, multi-site connectivity. SIP trunking allows the user to leverage the Internet and totally bypass the PSTN for communications between multi-site organizations, significantly lowering communications costs between sites.

Voice redundancy. With VoIP’s ability to allow users to ring DIDs into any location, one can realize redundant enterprise VoIP services. Two ingress points of presence can be established into the enterprise. The incoming calls then can be load-shared across those two or more points, resulting in voice redundancy for the enterprise. This is almost impossible to do with traditional PSTN technologies due to cost and PSTN rate-center limitations.

True multi-site enterprise unification. VoIP communications integrates seamlessly with other IP services and IP networks, allowing true unified-communications possibilities for the enterprise.

Converged networks and technologies. Converging multiple networks with technologies down to a single network and a single underlying technology offers significant cost-savings possibilities.

Cost-savings in support personnel. Multiple support groups can be collapsed down to a single group to handle all data and voice needs. The cost-savings realized from that sort of merger can be considerable.

Cost savings in monthly recurring circuits. IP communications from the enterprise to the outside world generally is done over IP drops that are, in many cases, partly under-utilized. By moving voice service to that resource and removing the stranded costs of TDM T1 circuits, more significant savings are at hand.

Cost-savings on usage. VoIP carriers in general bypass the ILECs and go directly to wholesale national carriers for national and international destinations. That results in much lower usage costs, which in turn get passed through to the end user.

What It All Means

IP networks will be the dominant transport means for most media types (including voice) for the foreseeable future. The reasons for this are clear: IP networks are easier to deploy, less costly to manage and offer far more flexibility in how they handle and transport various media types.

Granted, VoIP offers some unique challenges, due to its real-time nature, but they can be handled. Therefore, IP — being ready for voice transport — puts the MSO in an interesting position.

So will CableLabs sometime in the near future offer a standard SIP-trunking solution that will allow the cable industry to move forward? Probably not. Focusing on larger-capacity circuit technologies over HFC (DOCSIS T1) is not the solution. Rather, SIP trunking over IP is the solution because of:

VoIP bandwidth allocation on the last mille;

The VoIP CPE network interface (the firewall problem); and

VoIP security for, and VoIP separation from, the corporate LAN.

If these challenges are resolved and the MSO figures out how to market the service, cable IP networks stand a good chance of becoming the dominant transport of choice for small and medium-sized enterprise voice users as the future unfolds.

Peter Sandstrom is CTO at BandTel. Contact him at [email protected].