After discussing the various aspects of accounting and performance management, now we want to relate the information to the business. How can the two areas help you design and operate a productive, reliable, and cost-efficient network? During the Internet hype, revenue generation and assurance were certainly not the center of attention, but this changed quickly when the global economic slowdown started. Today, if a network architect develops a new design or extends the existing design, a solid business case needs to be in place to rationalize the investment and calculate a quick return on investment, or the project might never kick off. Therefore, you need answers to the following questions:
How can accounting and performance management increase revenue for a service provider?
Can it help an enterprise to better utilize the network equipment and identify bottlenecks in the network, so that only these bottlenecks are resolved by an upgrade instead of upgrading the entire network?
Can you justify an investment in accounting to increase revenue assurance?
Will accounting and performance management help reduce the operational expenses?
Can you classify the most- and least-profitable areas and access points of a service provider network?
The related costs of usage reporting are a major deterrent that might stop administrators from considering accounting a valuable solution for their network. Three main areas are related to these concerns:
Collection overhead— Based on the selected technique, generating usage data records at the device level can create a significant resource impact on the device CPU utilization and can require additional memory and bandwidth utilization when exporting the data records to an external collector. This needs to be considered in the proposed accounting and performance management architecture.
Mediation, processing, and reporting overhead— Additional resources are required to collect the records and aggregate and mediate them afterwards. At this point, the collection contains only data (collected usage records) and not processed information, such as end-to-end performance records, a list of security attacks during the last month, or a printable invoice for a customer. The next step is processing the data at a performance, security, planning, or billing application, which creates useful information for the NOC personnel. The costs of these applications cannot be neglected; even open-source software has related costs of server hardware and administration expenditures.
Security overhead— A huge benefit of collecting accounting records is its inherent multipurpose functionality. As mentioned, multiple applications can benefit from data collection, which results in a "collect once and reuse multiple times" proposal. Imagine the disaster that happens when data sets have been manipulated. If multiple applications base their results on wrong data sets, the whole network operation is jeopardized. This is not limited to a customer's invoice, but also to network planning and especially security analysis. As soon as an attacker detects the corporate accounting architecture, he or she can manipulate data sets and bypass the traffic meters that were put in place to identify security threats. It is critical to implement mechanisms to protect accounting information from unauthorized access and manipulation.
A possible approach to address the collection overhead and security overhead issues just mentioned is to deploy a Data Communications Network (DCN). The DCN concept was developed in the service provider context, but it can also be useful in an enterprise environment, where it is usually referred to as out-of-band (OOB) management. What is a DCN? A DCN is a standalone out-of-band network providing connectivity between the network operations center (NOC) and the network elements, remotely and independently of the user traffic. The opposite approach is called in-band-management, where management traffic and user traffic share the same infrastructure. The DCN can be used for all network management operations, such as network surveillance, provisioning, testing, alarm monitoring, service enablement and restoration, and collecting accounting and performance details. Figure 1-25 illustrates a DCN.
The costs related to accounting solutions—both CAPEX (capital expenditures) and OPEX (operational expenditures)—can become an impediment that the network planner might not be able to eliminate. We propose building a strong business case before considering the technology and implementation details. The balance between costs and benefits can be harmonized by the granularity of the collected accounting records. Again, the level of required details is determined by the business case, not just the technical feasibility. If you can come up with good arguments to justify a short return on investment for the solution, we assume that the chances of getting project approval are much greater.
The following questions should be considered first:
What is the benefit for the cooperation of collecting usage and performance data records?
Which existing processes are impacted by the proposal?
Can processes be optimized by leveraging results from an accounting and performance management application?
Which new processes can be created that were not possible in the past?
Does the proposal generate additional revenue? What is the ratio of investment (CAPEX) to revenue?
Which additional operational costs (OPEX) are caused by the proposal?
Can the solution achieve a competitive advantage?
Does the project have an executive sponsor?
Has outsourcing been considered?
What are the low-cost alternatives? Which functions cannot be implemented by a low-cost design? Is the additional functionality worth the premium price?
What is the price of not implementing anything? Keep in mind that not doing anything also has a related price, which can be monetary (such as losing potential revenue) or nonmonetary (such as decreased customer satisfaction or missing future opportunities).
Following these questions, we develop some suggestions that might apply to the situation of individual readers. This list is not complete, but it tries to give a comprehensive set of ideas that network operators can use for further consideration.
Nowadays, service providers are faced with increasing bandwidth demands from customers without a corresponding increase in revenue. All of us want flat-rate DSL or cable at home, while at the same time paying less than we did when using the dialup or ISDN network. This requires careful planning at the SP to ensure that only the necessary parts of the network are updated to meet customer needs. The days of a fully overprovisioned network are gone. Service providers need to find a balance between over- and underprovisioning the network. If the traffic that is placed on a link or circuit is greater than the bandwidth, it is underprovisioned (or over-subscribed). If less traffic is placed on the link than it can transport, it is overprovisioned. Making appropriate business decisions about network provisioning requires details that only network usage analysis (such as accounting and performance management) can facilitate.
We should take the time to define "service provider" in this context. Sometimes, people interpret service provider as an "incumbent" or PTT (Postal, Telegraph, and Telephone) company, which was the case several years ago. Due to the deregulation of the telecom market, there are incumbent (or telecom) providers (the classic ex-PTTs), challengers, Internet Service Providers (ISPs), Application Service Providers (ASP), and so on. Multiple large enterprises have outsourced their IT departments, and these have transitioned from cost centers to profit centers. National Research and Education Networks (NRN or NREN) provide networking services to the national universities but also have international agreements in place. All of these "service providers" have a common denominator—offering services to users. The user definition varies and so does the service, but thanks to competition, users have some level of freedom to select another provider, and that should be enough motivation to provide good quality.
So far, we have considered so-called "service provider"-specific benefits of accounting and performance management, but we don't stop at this point. Enterprises can benefit from the same functionality, except for billing. Instead, some use it for department charge-back and monitoring usage policies. Deploying accounting for charge-back per department might not reduce the network's operational costs, but it can lead to better assignment of the costs. Assuming you work in a group that requires high-speed networking access (for example, an automotive engineering team, spread across multiple sites and heavily using computer-aided design technology), it would probably be okay for you to get charged for an upgrade of the links you are using. On the other hand, if you lead the human resources department and have no high demands for the network, you would probably not be willing to support the complete corporate network upgrade just because of another department's request.
Accounting solutions can help assign the costs to the originator instead of an equal cost distribution; consequently, enterprises should consider accounting as a mechanism to enforce rational cost allocation and recovery.
Monitoring usage policies also falls in the same category. If the enterprise network was designed to meet the normal traffic demand during business hours, a policy could be established to run backup processes during the night to better utilize the network infrastructure. If some users decide to ignore the policy and run large backups during peak hours, all users would suffer and would have to pay the extra price for a network upgrade. Accounting and performance management can supply beneficial information.
Both enterprises and service providers use accounting and performance management to baseline and plan the network. We already explored several scenarios where a solid baseline is the foundation for higher-level services as well as planning. A service provider might offer baselining as a service to the enterprise, potentially without additional investments. If the provider already collects accounting data for billing purposes, these records could be used as input for a monitoring and planning application. Either the service provider offers the raw data collection to the customer, and he runs his own application, or the provider offers this service in addition and delivers the results as planning proposals to the customer.
Another common area is security management. Identifying and stopping security attacks in the network is one of a network operator's most critical tasks. Viruses and other malicious code attacks are growing in number, and so is the cost incurred by companies, government organizations, and private individuals to clean up systems and get them back in working order. Malicious code attacks include trojans, worms, and viruses.
Service providers can benefit by protecting their customers from DoS and worm attacks. Accounting techniques, such as Cisco NetFlow services, can be used to detect attacks, isolate attackers, and identify attack propagation. ISPs have the opportunity to offer security reporting and protection services to their customers and generate additional revenue. In the future, we expect that security parameters will be included in SLAs. Enterprise networks have had issues with the flood of worms, including SQL slammer, MS slammer, Nachi, and others. Security services that prevent attacks can significantly reduce enterprises' OPEX.
QoS and SLA management are two other areas for collecting performance and accounting usage records. The combination of QoS and SLA can provide a competitive advantage for an ISP, even though QoS deployment is not a necessary condition for offering SLAs. Some ISPs offer different service classes and SLAs already; unfortunately, no standardized SLA definitions and QoS settings are deployed across multiple ISPs today. Users notice this when using IP telephony in the Internet and the two parties are connected via different carriers. Unless SLAs between the ISPs themselves and toward the customer are defined, the traffic is treated as best-effort, with the result that quality of voice calls is not predictable.
The traditional models of carrier interconnection offered no end-to-end service level, which was a limiting factor for business-critical applications that require QoS. IP Telephony and video will most probably become the driving application for QoS in the Internet, and in contrast to current Internet applications, voice is the first real application that requires some level of service quality. Internet QoS might also become a revenue generator for ISPs, which is relevant when legacy telephony, which was the PTT's source of income for decades, will no longer be available. Of course, some users want IP telephony for free, but others are willing to pay for excellent voice quality, even though the price will still be drastically below the previous prices for long-distance legacy voice calls. SLAs across carriers will become available, and customers can select carriers based on SLA contracts and guarantees.
Finally, yet importantly, we want to address peering agreements and billing.
Peering agreements are mutual contracts of two providers to transfer each other's traffic free of charge. Transit agreements represent a service in which one provider charges the other for data transmission. In both cases, accurate collection of exchanged traffic is vital to both parties. Usually each of the two contractors measures both ingress and egress traffic, but obviously it would be more efficient to collect the data only once—which requires a level of trust between the two parties. Because there is not one, but multiple industry standards for collecting accounting information, the worst case happens if the parties use different accounting collection methods that are not interoperable. This leads to different measurements at the end of the month and can be a starting place for conflicts.
It gets even more complex when billing scenarios are addressed.
Managing large volumes of complex voice call-data-records and Internet usage-data-records is a challenge for most organizations. A list of challenges includes the following:
There can be hundreds, or thousands, of accounts across the organization (depending on the granularity of the billing records).
Different accounting records from various sources increase complexity and require mediation.
Billing dates may not be aligned where more than one supplier is providing services.
Some divisions may require electronic billing; others may prefer printed invoices.
In spite of these issues, billing can be a great source for identifying future growth opportunities and customer satisfaction (or dissatisfaction). A provider can identify profitable areas that show accelerated growth and increase bandwidth to maintain and increase customer satisfaction. Unprofitable areas could be examined to identify competitive strategies and launch marketing campaigns to gain new customers. Even though flat-rate billing sounds desirable to all customers, only heavy Internet users really benefit from it. An average individual user might not be able to generate the same volume of a family, where parents and kids use the Internet extensively. In this scenario, the low-volume users pay the bill for the high-volume users. This is exactly the situation we faced with enterprise networks in the past, where the costs were shared equally across all users. Nowadays, corporate users expect a fair distribution of costs and compliance with corporate policies instead of flat-rate distribution. The same might happen to commercial users when considering the cost of Internet access; thus, we expect usage-sensitive charges as a potential model for the future. Charging a fixed price per volume or connection time could be combined with a cost ceiling (such as a maximum fee per month) to avoid an incalculable price during peak periods.
In general, we believe service differentiation is imperative for service providers. The necessity of responding to competitive forces requires the ability to rapidly develop and implement new services, including the proper infrastructure to bill for them.