The Management Implications of
Public Frame Relay Services
Preface
Frame Relay has flourished as many organizations have embraced this new technology to reduce the WAN circuit and CPE costs of LAN/WAN networks. But cost savings aside, the changes brought on by Frame Relay demand new methods for operational support. With traditional tools obsoleted, the lack of effective new tools is hampering the network manager's performance of the key network support functions.
This technical backgrounder reviews the operational support issues that arise when LAN/WANs are migrated from leased lines to Frame Relay. It then defines the ideal operational support model which forms the management foundation for these networks. The components and implementation of the Visual UpTime Frame Relay Network Support System are also reviewed.
| Fault Isolation |
Topology Optimization |
Performance Management |
Usage Accounting |
|---|---|---|---|
|
|
|
|
The New Horizon in Networking: Frame Relay
The network fabric for corporate computing applications has rapidly evolved to a complex topology of workstations and servers locally connected by LANs, then interconnected across geographical boundaries by WANs.
Given the high cost of WAN bandwidth, network managers have probably just deploying or are considering fast-packet technologies such as Frame Relay as a means of gaining bandwidth efficiencies for bursty LAN traffic. They may also see Frame Relay as a vehicle for a nearly seamless integration of legacy traffic (e.g. SNA) with the LAN traffic.
Clearly, Frame Relay technology is poised to quickly displace traditional WAN technologies based on leased lines and time-division-multiplexing (TDM).
But Network Managers are Not Equipped to Support It
Knowing that the case for Frame Relay is justified, most network operational infrastructures are not prepared to support, manage, and test the network. Frame Relay WANs operate very differently from TDM WANs, so traditional tools and techniques aren't adequate for managing these new networks.
And to add to their stress, network managers are faced with the imposing dilemma of how to deal with the increasing demand for support in the midst of decreasing budgets. There just isn't enough resource to provide the expected level of network service. Despite this, the number of networks and users continues to grow without bound and the network applications continue to grow in strategic importance.
Some organizations have adopted "open" management platforms based on SNMP as a means of consolidating the numerous support systems and staff traditionally dedicated to separate management activities. Unfortunately, they are still searching for the solutions that completely integrate the WAN management functions onto these platforms.
Many organizations running mission-critical applications have been putting off the wide scale deployment of Frame Relay for these reasons.
Frame Relay is Markedly Different than TDM
The numerous benefits of Frame Relay are achieved by radically changing the process by which data is transmitted over the WAN.
| Leased Lines | Characteristics | Frame Relay |
|---|---|---|
Circuit structure is now based on virtual (logical) connections rather than physical connections. User applications are allocated bandwidth statistically rather than through assignments of guaranteed timeslots. This creates the potential for congestion scenarios when many users contend for the same finite WAN bandwidth.
Delay becomes variable, not constant. Routing decisions are performed at the customer premises and at the carrier switches rather than just at the customer premises.
With a lack of inherent error-recovery, Frame Relay circuits are also vulnerable to transmission impairments. In the case of congestion or errors, network nodes can deliberately discard cells or frames. Loss, congestion, and delay variability impacts the Quality of Service (QoS) from a user's perspective.
Flexibility Brings Management Complexity
The virtual circuit and statistical orientation of Frame Relay promise substantial economic benefits for building complex enterprise-wide networks. However, these very attributes are the source of management complexities far greater than those associated with TDM technologies -- and much less understood.
In a Frame Relay environment, the network characteristics are extremely dynamic. Traffic patterns vary constantly, congestion scenarios are intermittent, and virtual paths between communicating end-points can change. Variance creates unpredictability, so that what works today may not work as well tomorrow.
New Methods are Required for Operational Support
Clearly, since Frame Relay WANs do not operate like traditional TDM WANs, traditional management methods are no longer applicable.
The virtues of Frame Relay will be exploited only when the need for operational support has been addressed.
Moving Up the Stack
The days of ignoring protocol issues are over. Frame Relay-based
LAN/WANs comprising multi-protocol routers, virtual circuits,
and network switches are very protocol intensive. Whether it's
the network service protocol such as Frame Relay, routing protocols
such as RIP and OSPF, or application protocols such as FTP over
TCP/IP, for today's WAN environment, network managers must be
fluent and capable of supporting the network at the protocol level.
Troubleshooting - Because problems with Frame Relay WANs can have many different causes, troubleshooting tools must include detailed information regarding the access lines, virtual circuits, and virtual paths. Network managers have got to be able to quickly identify the source of problems -- be it CPE, the public network, or user applications.
Historical statistics are required to isolate intermittent problems and to investigate trend-related symptoms. Virtual circuit utilization and end-to-end delay measurements are mandatory to assess the impact of network queuing. Packet capture and decodes verify the integrity of application traffic. Statistics, decodes, and traces of layer 3 protocols are required for complete isolation of virtual path problems.
Performance Optimization - Continuous performance assessment is the key to getting ahead of nasty network problems and taking action which prevents impact to network users.
Network managers need to establish operating thresholds and monitor performance trends. For example, they might want to be alerted when IP traffic exceeds 50% of a virtual circuit's bandwidth.
Network managers also need to know when the router-to-router protocols are chewing up the bulk of the WAN bandwidth. Or when congestion is caused by a bandwidth-hungry bulk file transfer that could be batched for a more opportune time.
Topology Optimization - Capacity planning requires the evaluation of historical utilization and point-to-point delay information. This allows network managers to implement topology changes which optimize the WAN for lowest cost and lowest path delay.
Again, network managers need insight into utilization by application. Aggregate frame counts aren't enough.
Usage Accounting - How does the cost of Frame Relay bandwidth get allocated properly to the users or departments who are consuming it?
Network managers need accurate information as to network usage. This information is only available by monitoring each virtual circuit for the network layer addresses and the amount of traffic generated from each user. It's just not enough to know that a virtual circuit is highly utilized -- there could be hundreds of different users from multiple cost centers within the organization using the bandwidth.
This Detailed Information Won't be Available from the Routers and Switches
Despite the hype about all the great management capabilities that are supposed to be integrated into the routers and switches, it just won't happen.
For starters, network managers demand an objective view on network conditions, particularly for troubleshooting and performance optimization. They require tools that can eliminate frustrating finger-pointing scenarios, not contribute to them. Furthermore, total reliance on router or switch diagnostics also means that test functionality is lost when the router or switch fails.
Routing and network analysis are both processor intensive functions. Expecting any device to perform both functions well is unrealistic. The more analysis functionality a router or switch is burdened with, the more it limits its core performance for routing or switching.
Network Operations: A Dual Element Model
The best model for management of Frame Relay-based LAN/WANs is a combination of equipment management capability and independent network analysis capability. Both elements
include applications which reside on "open" SNMP platforms. While equipment-management applications address the functions of configuration, security, and self-diagnostics, they fall far short in other important areas.
Embedded network analysis supplements the equipment-management applications by effectively addressing the remaining operations activities -- detailed troubleshooting, performance baselining and monitoring, topology optimization, and usage accounting.
Visual UpTime is the complete support system for managing a Frame
Relay-based LAN/WAN. As an integrated set of embedded analysis
tools, Visual UpTime allows both end-users and carriers to achieve
significant performance and productivity improvements from Frame
Relay network resources while dramatically reducing recurring
operational support costs. Management goals again can be met.
The heart of Visual UpTime is a network analysis database. The database comprises multiple Analysis Service Elements (ASEs) distributed out to each access point of a Frame Relay network collecting and periodically forwading data back to a centralized SQL database for long term archieval. This design ensures that the database is populated with real-time as well as historical information regarding the network operation. Furthermore, by locally storing the operational data at its collection point, network bandwidth used by Visual UpTime is minimized.
Software applications for network operations support are deployed at one or more network management centers and access the virtual database using the SNMP management protocol.
These client applications retrieve specific operational data collected by the ASEs and intelligently analyze it to provide meaningful management information on a network-wide basis. Client application "toolsets" encompass each network operations function and are tightly integrated into a graphical user interface.
For example, the Expert Event Processor monitors the network continuously and acts as an early warning system for impending performance degradation.
The Troubleshooter toolset allows detailed and rapid fault isolation from a central management console -- the equivalent of having powerful protocol analyzers on every access point in the network. By centralizing this activity, the need for remote support staff and field dispatches is greatly minimized. Perplexing Frame Relay problems are now rapidly resolved centrally with a network-wide scope. Furthermore, budget-draining purchases of expensive high-speed portable analyzers can be essentially eliminated.
The Network Planner toolset facilitates optimal network design resulting in least cost topologies. It includes the extensive Report Generator that draws on the long term database of network usage. This provides unparalleled capability, giving you powerful insight into creeping network trends and into the impact of network changes such as corporate application roll-outs, user migrations, and network software updates.
Dedicated Processors for Data Collection and Analysis
The ASEs employ dedicated high-speed processors optimized for the function of collecting and storing network operational data.
The core functionality of the ASEs is to collect extensive real-time operational data at a given point in the network and store the information in a structured database for historical records. The data is categorized as follows:
Open Management Integration
The client applications run on an SNMP management platform at a peer level to router and switch management applications. This provides the foundation for smart linkages between management applications. For example, the Troubleshooter tool kit could be launched from within a help-desk and trouble-ticketing environment.
Objective View
Visual UpTime provides a much needed objective approach for gathering and analyzing problem and performance data from the Frame Relay-based LAN/WAN. Most importantly, it provides a way to test the network when the active network components degrade or fail.
This independent view toward network analysis is paramount when evaluating the performance claims of network equipment vendors and when resolving the finger-pointing battles that are certain to occur between end-users, CPE vendors, and the Frame Relay carrier.
Security & Network Integrity
The ASEs monitor the network by passively bridging across the physical circuit at the network or data interface. Along with extensive password protection capabilities, unauthorized use is minimized and Visual UpTime does not constitute another potential network failure point.
The Economics
Visual UpTime not only enables significant improvements in network performance but also reduces recurring network support costs.
For a typical T1 Frame Relay network, consider the expenses that can be eliminated by deploying Visual UpTime: network managers can reduce (or at least stop growing) their field support staff; they can eliminate a large number of costly field dispatches for problem isolation; they don't need to buy expensive Frame Relay protocol analyzers; and they don't need to buy additional "SNMP-managed" DSU/CSUs. Depending on their current network support practices, organizations should probably realize financial payback in less than a year.
On top of that, network managers will sleep better at night knowing that they have a solid system working to meet their aggressive availability requirements.
And perhaps most importantly...go for it...deploy Frame Relay and know that they can get terrific WAN bandwidth cost savings without compromising the quality of service that their users have come to expect.
Home | Corp | News | Visual UpTime | Visual OnRamp | Tech Support | White Papers | Employment
Copyright © 1996 Visual Networks, Inc. All rights reserved.