PDF of this tutorialA carrier SLA is a necessary piece within an overall process for managing WAN–service levels. As Figure 4 shows, there are three parts to the WAN–service-level management process.
Figure 4. The WAN–Service-Level-Management Process
Continuously determine what WAN service levels are needed. To do this, users (subscribers and providers) need to profile usage and baseline normal operation so they can plan WAN–bandwidth capacity. Users also need to base these service levels on actual traffic profiles (i.e., per port/circuit, host, and application). Because traffic can be burst-intensive, they must measure these usage profiles with fine granularity.
Once service levels are established, they must be verified. To verify service levels, users need to monitor performance in real time, report historical performance, and assess quality-affecting trends. These trends can only be tracked through an historical database of performance information, including the ability to report on a WAN SLA.
If service levels are not being met, then users must quickly determine why they are not being met to rapidly resolve the problem. To do this, users need to analyze, both historically and in real time, the physical and logical access to the cloud. Finally, the analysis of upper-layer–protocol traffic on the WAN is important as the configuration of how these protocols are to be run directly impacts their performance over the WAN service in use.
Each step is critical in implementing and monitoring frame relay–service levels. Accordingly, it is recommended that companies follow these guidelines:
Determine frame relay–network requirements and traffic patterns. Baseline the network. Understand applications, peak times, and areas of concentration. Are there some applications that require large frame sizes? Are there critical sites that require redundancy? Use this baseline to design the network—with the carrier. Show the carriers how network requirements were determined.
Determine network configuration. Involve the carrier from the start. Understand their backbone technology and how their configurations will impact the performance of applications over the WAN.
Negotiate service-level agreements. Analyze and compare. Read the fine print and do the calculations. If the negotiated network-availability guarantee is 99.5 percent, how many hours of outage does that mean for the network per month? If the round-trip network delay is 140 milliseconds for a voice packet, will the conversation between the chairman of the board and the vice president of marketing sound like a call from one of the first airline telephones? Will my SNA transaction timeout during end of month accounting close? If my throughput guarantee is 99 percent, how often is the application retransmitting data? How does that impact server performance? Will the president's e-mail download freeze up while waiting for the server to be available? How are the agreements to be reported? Web? What timeframes? What are the methods of alleviation and penalties? Is there a maximum penalty monthly, yearly? When are the service levels implemented—thirty, sixty days after installation? What is included and excluded? Does this meet company needs?
Determine a plan for monitoring the carrier. Will the carrier be monitored through a trouble-ticket system? Through user complaints? An implementation of automated planning, reporting, troubleshooting, and problem resolution tools is recommended. A company should let the carrier know the plan for monitoring their services.
Baseline the network during the frame-relay implementation. Utilize a system to determine baseline areas such as the following:
- PVC usage by protocol type
- PVC congestion performance (throughput)
- access channel usage
- access line performance
- availability
Discuss the results of the baseline process with the carrier. Make adjustments to network configuration if necessary. Many companies can actually reduce CIR on some PVCs.
Analyze network performance and reliability weekly. Companies should understand the nature of application and user traffic and be sensitive to change. Look for trends. Is performance degrading on specific links or PVCs? Has a new application been added? What is the impact to the network and other applications?
Meet with the carrier monthly and compare statistics. Determine courses of action if there are trouble spots. Seek compensation for service-level guarantees that are unmet. Review possible optimization plans and set the process in motion for change.
The partnership with a service provider through these steps will help ensure a successful network implementation. The added functionality of an end-to-end monitoring system assures that the complete business benefits of a WAN will be realized.
