For the past couple of years, open access was mostly a political matter. Would multiple system operators (MSOs) accept outsiders into their networks and their headends? Or would they battle the issue on a multitude of technical and organizational grounds? It seems likely that unless the industry ignores the logic applied to the AOL/Time Warner merger, many of the organizational and political hurdles will eventually be put to rest. MSOs will be required to open their networks to third-party Internet service providers (ISPs), giving them the ability to compete openly for the keyboards and minds of the millions of end users expected to be connected through broadband networks.

MSOs have historically viewed open access warily–figuring it would reduce revenue by cannibalizing what would have been a monopoly business. Another way to look at open access is as an opportunity to expand overall broadband market share. By expanding distribution through a selection of ISPs, customers can be lured away from DSL, satellite or broadband wireless providers.

The direction it takes depends on how well services are delivered and managed. The relative simplicity of HFC networks could yield tremendous management advantages, giving cable a clear advantage in the broadband market. Delivering on the potential requires automating the provisioning processes that will now straddle two providers' back offices.

Figure 1: Today's architecture.

Service provisioning challenges

"Provisioning" refers to the sequence of many complex events that take place in order for a service to reach an end user. This service is a full product, not simple connectivity–the entire range of things (connectivity, e-mail accounts, disk space, browser, etc.) that a customer would receive upon signing up for a particular package (see Figure 1). For many network service providers, delivering these complex services is a difficult, error-prone undertaking. The so-called "provisioning bottleneck" in current systems is a result of having to manage interactions between several traditionally separate pieces (see Figure 2). Automating service provisioning in an open access environment adds significantly to the level of complexity because, first and foremost, new pieces must be added to the provisioning puzzle.

Figure 2:
 Provisioning gap-current systems hold back automation.

The first step in developing any solid provisioning function is to map out the business process of delivering services. There are a number of ways in which open access might be implemented. Many of the major cable companies have open access trials underway, and there appear to be a number of different approaches to the business model. Some providers believe that they will be able to tightly regulate the number and kinds of ISPs they will allow to operate inside their networks. This is being dubbed managed access, rather than true open access.

Others are building HFC networks specifically for the purpose of selling wholesale transport to service providers, bypassing the video and telephony services of traditional MSOs. Some providers will provide a portal for users to select an ISP, while others will require signup through the ISP (as is common with DSL access today). Any provisioning infrastructure should allow for this variability. Interestingly, the key technical challenges to provision services are similar, regardless of the business model chosen. Unless the right tools are put in place to assure rapid, error-free management, increased profits will likely elude providers.

From the network service provider perspective, there is a need to associate and route network traffic from individual subscribers to their particular Internet service provider. In traditional cable offerings where the network operator is also the service provider, there is no need for this function. Any subscriber attached to the network is also a customer of that network's services. But in an open access environment, there are a number of possible ISPs, so something in the network must have the capability to identify, route, and aggregate traffic according to the ISP (See Figure 3).

Figure 3: Open access network architecture.

There are a number of alternatives to solving this new problem, but "source-based routing" is emerging as the industry standard. This refers to the use of pools of IP addresses for customers of specific ISPs so that routing can be used to direct traffic to and from the proper location.

The primary reasons are the relative ease of implementation (cable high-speed data networks are DHCP-based) and the natural fit with cable's IP-based, class-of-service strategy for advanced services.

The primary changes required to support open access provisioning using source-based routing are in the area of IP address management. Address pools must be established and managed over potentially widely distributed networks. Most DHCP servers are not well designed for this task. The other change is that the signup process must be modified to allow some users to follow one business process (use of the network provider's ISP), while other users follow another (use of a third party ISP).

There is a new communication requirement between the ISP and the network provider to set up and manage services. Potential elements include service establishment, bandwidth modification, billing notification and a wide variety of other entities. Security is also a critical issue, given the attraction that free broadband service has to the hacker community. Note that this communication can originate on either side. That is, the ISP could signal the network provider to establish service, or vice versa.

The early open access trials have approached the communication requirements by considering the potential interaction needs and developing custom message formats and protocols for those needs. The MSO is typically in the driver's seat and dictates specifications to any ISP that wants access to its network. There are some obvious technology candidates (XML, SSL), which seem to be taking hold. There have also been some industry discussions about the need to standardize some or all of these communications.

While standardization makes sense, the primary obstacle that must be overcome is the class of service variations utilized for value-added services.

Figure 4: 
Provisioning evolution: Any access, any application, any environment.

Note that as the market evolves from basic "connectivity" services to value-added, on-demand services, the need for communication becomes frequent and ongoing, and the nature of the communication becomes more complex. Rather than providing static "packages" (basic/premium/movie channels, etc.), services of the future will be available a la carte, allowing subscribers to receive content and applications when they want it. These on-demand, bandwidth-sensitive services (e.g., voice or video-conferencing) require on-demand coordination between the network service provider and its ISPs.

There is an analog to this provider-to-provider communication–the Access Service Request (ASR) and Local Service Request (LSR) of the U.S. incumbent local exchange carriers (ILECs). When a CLEC wishes to resell a line, trunk or entire service from an ILEC, it uses the appropriate ASR or LSR format. There is even a forum (the Ordering and Billing Forum or OBF) where new versions of the format are discussed and voted on. Despite the provisioning difficulties associated with CLEC to ILEC provisioning, the ASR and LSR are considered successful.

Addressing these challenges has been made more complicated by the frenzy of franchise swapping and the resulting multi-vendor nature of the resulting networks. While single-source networks of straightforward devices can be configured using the element management software (EMS) of the vendor, multisourcing the network or requiring more advanced application server configuration has significant service provisioning implications. Adding multiple service providers to this increasing network complexity leads to a situation where the most pressing issues in equipment selection may not revolve around shelf space or port density, but around the ability to flexibly provision services on-demand.

Solution: Automated Service Provisioning

Many cable operators and ISPs will address these issues with new "provisioning engines" or next-generation provisioning systems to automate all necessary functions and keep service provisioning flowing smoothly. The provisioning engine's ability to respond to many end user requests, talk to different components, or alert administrators to problems can solve challenges such as these for both network providers and ISPs.

These provisioning systems will also provide the foundation for growth as the networks evolve to deliver services over multiple transports (see Figure 4).

To understand the power of these next-generation provisioning systems, take a look at the following potential scenario.

After enduring disappointing connection speeds from standard dial-up Internet service, Jane Consumer is looking for high-speed access via cable modem.

  • She hooks up the cable modem (PC came configured with an Ethernet card and TCP/IP already) and is presented with a choice of competitive ISPs (supplied by the local network provider) and the services they offer.
  • After studying, she picks ISP X and clicks on its logo. Immediately, Jane is taken to a sign-up screen, where she enters relevant data (name, credit card number, etc.).
  • Clicking "Finish," she finds herself immediately able to access the Internet via her new high-bandwidth connection. Later, Jane decides to order a pay-per-view concert that's going to be broadcast soon.
  • Via her ISP's customer interface, she clicks on the concert and puts the charge on her stored credit card information.
  • She also takes the opportunity to change her contact information and set up an extra e-mail account for her daughter.
  • Soon afterwards, she learns of another ISP that offers superior functionality at a slightly lower price.
  • She visits the network provider's interface that she initially used to select her ISP and switches to ISP Z by clicking on a single button to initiate the change.

This process is complicated enough when the network provider and ISP are the same company. Providing a seamless, error-free experience becomes that much more challenging when you consider that in these examples, at least five messages must be transmitted between the two providers. Open access will only be a success when everything that Jane, the ISP, and the network provider need is provided with minimal wait and maximum efficiency and transparency.

Consider what happens during provisioning to handle these details:

  • When Jane installs her modem and powers it up, the DHCP server of the MSO auto-discovers the physical address of the modem and allows limited network access to a registration page with an up-to-date list of available ISPs for Jane to choose.
  • Her choice prompts a signup page from the selected ISP where Jane inputs her personal data. The network provider's provisioning system, in conjunction with the ISP provisioning system, enters data in the ISP database, performs a check on her credit card information, sets up her e-mail addresses and any special features ordered.
  • When signup is completed, the provisioning system configures the DHCP server to allow network access via the selected ISP's backbone by assigning an IP address from the appropriate pool and resetting the modem to the selected bandwidth.
  • Later, when Jane orders a concert or makes other changes, the ISP provisioning system signals the network provider to allocate the appropriate amount of bandwidth for the required amount of time, bills Jane's credit card for the extra bandwidth, changes the information it stores about Jane, and sets up a new e-mail account. The network provider's provisioning system must respond appropriately to the ISP's request to change bandwidth.
  • Finally, when Jane changes ISPs, the network provider's provisioning system sends "cancel" information to the old ISP, which in turn deletes her account. The network provider's system then resets network access to the new ISP's registration page by assigning a new IP address and resetting the modem, and initiates the above process with the new ISP.
  • There is one additional provisioning step that Jane will never be aware of. The network provider's provisioning engine must assure that the ISP is billed for the wholesale use of its network.

In the end, the two provisioning systems in tandem managed all the details of Jane Customer's cable modem experience. Jane dealt with only the interfaces and forms relevant to a customer; she was never made aware of the complex series of commands and actions that led to her successful service provisioning. All processes took place automatically, which eliminated the need for human intervention and sped the results of Jane's various requests.

What has to change in MSOs' infrastructures to achieve this seamless service delivery in an open access environment? Three major things.

The first is new IP address management platforms. The IP address management platforms currently used in cable networks are typically based on those used in commercial office environments. Cable high-speed data service, particularly in an open environment, requires additional functionality. Issues such as distributed address pool management, modem reset speed and the ability to create individual profiles for specific modems become much more critical.

An application programmer interface (API) must be defined between the back-office systems and ISPs. The above example shows the number of interactions between the MSO and the ISP. Open access will break down if these interactions have to be handled on a case-by-case basis, so MSOs need to define an API so that ISPs can check for service availability, order and change service, inquire status and so on. The relative simplicity and flexibility of this API offers an opportunity for an MSO to differentiate itself based on service–the better the API, the more attractive the MSO will look to ISPs shopping for a network service provider of choice.

MSOs must upgrade and modify their existing provisioning infrastructure. The MSOs' current provisioning processes need to be modified to allow two branches of work; one if the customer chooses the MSO as the ISP (the MSO then provides retail services like e-mail, IP video, etc.), and the other if the customer chooses another ISP. This latter branch has several new steps, including a request for an IP address from the proper pool, and the addition of a new user to the ISP's wholesale bill.

The difficult reality is that many MSOs do not currently have the provisioning infrastructure to achieve the level of automation in the example above even for their own customers. The benefits of streamlined provisioning–reduced operational cost, faster subscriber activation, faster rollout of new products and services–is often worth a significant investment regardless of whether an MSO provides open access or not.

A glance at the future

Despite the recent market downturn, there is no question that in the future, broadband access will be commonplace. In many of the desirable markets, it will also be competitive.

Users will have a choice of networks over which to access their desired content. Mobility also introduces a separation between network and services as anytime, anywhere access to the same content requires multiple networks. Service providers are likely to become super-communities of interests, and transport mechanisms will be transparent to end users.

In this open access world, the competitive advantage for network providers is ease of use. Cable networks are considerably easier to manage than traditional circuit-switched networks, so the MSOs have the potential to succeed where DSL has stumbled. By providing ISPs with network and provisioning APIs that will make it easy for ISPs and their hungry subscribers to sign up, use, upgrade, downgrade, modify and order data products, cable MSOs have an opportunity to be the broadband access method of choice.