PING! Is anyone there?

Frame relay has become quite popular, but how about making sure you're getting what you pay for?

Judith Allen

Frame relay can connect your data communication applications over cost-effective public lines, but how long will it take you to verify the service level agreement (SLA)? Frame relay WAN services have quickly evolved into outsourcing alternatives for service providers. In addition to revenue generation, they also provide a welcome solution to the shortfall in IT departments facing expanded network services.

Initially positioned as an interim solution until ATM was available, frame relay has become an effective long-term solution for low-speed connections. Its projected growth rates for 1998 are in excess of 100% over 1997, with no decline in sight.

While outsourcing may relieve many headaches and keep costs down, how can system administrators confirm base-level monitoring? The network service provider (NSP) specifies a warranted level of service. However, to monitor these SLAs and ensure they are being honored, enterprise system administrators need an easy method to see what is occurring in the network. The best frame relay monitoring solution will have the following characteristics:

It should be able to ping an IP address to see if it is working.

It should consist of flexible, cost-effective devices with easy-to-access information and provide an SNMP (simple network management protocol) browser to see captured data.

It should provide Web access to service provider reports to reduce the need for yet another management system.

It should have application probes to enable the end user and service provider to obtain the best performance.

PINGING 

The quickest basic method to determine if the link is working properly is to use the Internet control message protocol (ICMP) ping. Depending where you start, the ping can make troubleshooting easier.

Is the LAN-to-WAN router causing the problem or is it the network? Notice that a ping command from the DSU/CSU is sent to the actual application server (Figure 1). If the network exhibits idiosyncrasies, you can send a packet with a specific test pattern.

Base-level ping testing is not necessarily always an accurate measure, but it will give a round-trip response time. The success of the transmission indicates that the network path is available. Even daily comparisons of ping response time can help determine if there are network-related service issues.

COST-EFFECTIVE DEVICES

Frame relay monitoring in the DSU/CSU can offer the most effective way to maintain control and get the information necessary to run the operation efficiently. Several DSU/CSU vendors offer products with varying levels of frame relay monitoring, ranging in cost from $100 to $3,600.

The enterprise system administrator must regularly monitor several critical points: (1) Is the service provider living up to committed levels of service? (2) Is the uptime promised sufficient to support mission-critical traffic? (3) When a network slowdown occurs, how quickly can it be traced to service provider problems or local router issues?

The best monitoring solutions should work non-intrusively within the network environment to provide round-the-clock monitoring and analysis (Figure 2).

Some of the critical components to evaluate in choosing a low-cost DSU/CSU solution are:

Does it have configurable traps and alarms that enable the user to separate network performance issues from engineering problems?

Does it have real-time alarms that notify administrators when an event occurs?

Does it have the ability to scale by supporting data speeds from 56K through T3/E3 rates?

Is it fully SNMP?

Does it have configuration flexibility through download of software modules to meet customer needs, such as in-band management or performance management?

WEB ACCESS

The quality of frame relay monitoring has improved dramatically due to customer demand. As the popularity of frame relay solutions has expanded, NSPs have improved their reports. However, NSPs still focus on mountains of detailed historic reports that do not provide easy or immediate access to summarized information and exception reporting. They do not provide information on LAN activity. The only way customers can compare actual service with committed service levels, using NSP reports, is to pay for expensive custom reports. Even then, it is difficult to get information as events are occurring.

NSPs typically consider their responsibility for service to be the first two layers of the OSI (International Standards Organization) model - the physical and data link layers. A DSU/CSU provides link layer information. Coupled with probes at strategic locations, it can see traffic-related data and frame relay items.

Customers can add their own more comprehensive probes to provide visibility into all seven OSI layers. Some of the most critical functions a monitoring system can perform are: (1) Identifying network engineering problems that can be resolved in-house to reduce slowdowns and down time. (2) Determining trends that will impact the committed information rate (CIR) in the SLA and planning capacity for them. (3) Ensuring that SLAs are being honored.

Customers must determine their frame relay monitoring requirements at the time they first contract for services. Some service providers now provide probes to characterize the network and its applications before connecting and providing an SLA.

Full-function probes let you see the information to size and estimate the frame relay service SLA. You can install the nonintrusive probes when establishing the network or at a later time. The recommendation is to have at least one application probe at the central site. When shopping for monitoring equipment, it is necessary to evaluate the available options to make sure the vendor has full application monitoring. You must ask the question, 'Can I see who is playing 'DOOM' or using the new custom application?'

A wide range of monitoring capabilities is necessary to see all seven layers of the OSI model, including physical layer, data link layer, virtual circuits, and application detail monitoring. Where once simply monitoring the physical and data link layers might have been sufficient, today's pervasive use of the Internet, with its inherent lack of information, makes application-level monitoring critical.

When response time declines, it is not enough to know that the pipe is full-to-overflowing. The network manager needs to know what is affecting the business' bottom line. Is the problem Web-based access, file transfers, Lotus Notes, or a combination of all of them?

As costs drop for services and SLAs improve, more companies will move from private to public network solutions to grow their enterprises. It is incumbent on the industry to ensure that cost-effective, easy-to-use and easy-to-understand monitoring systems are available to allow enterprises to maintain control of their connectivity solutions, monitor their SLAs, and keep their WAN networks functioning smoothly.

Allen is marketing manager at Digital Link Corp., Sunnyvale, Calif.