What Is Network Optimization?

What has happened to your simple network? Do you struggle daily to make your network perform like it still has youth, stability, and endurance? Do users call with complaints: "The network is running slow"; or "I've been waiting ten minutes for my document to print"; or "Our file server just went down."? Is this the same network that used to have one file server, 25 workstations, and only one network protocol?

What you have now is an internetwork. Your network runs TCP/IP, SNA, NetWare, and NetBEUI protocols. Your simple network has grown and now comprises five separate token rings with over 400 workstations and 30 file servers.

The easy days are over. A network that used to be easy to manage now requires a complicated, intricate set of steps involving calculated technical moves to manage this monster you created--this internetwork.

Techniques are available to help you get a handle on your network--to understand how it is running and performing, and how it is utilizing the innovative technologies you implemented yesterday. You can also use certain techniques based on calculated measurements to implement changes so that your network will run and perform better. This is known as network optimization.

Network optimization is the process of measuring to a defined level a network's workload characteristics and then making modifications to the network's layout, design, and configuration to improve its overall performance.

Most often network optimization involves using a protocol analyzer to evaluate the operation of the complete network, including all its hardware and software components. After evaluating the operational state of the network, the next step is to tune all the components. The goal is to make the network components work together so that your internetwork can perform the mission-critical operations for which it was intended--at a higher performance level.

Protocol analyzers enable you to optimize and view data on your network to help you understand how that data is performing. Protocol analyzers are mostly independent of protocols, the network operating system you are running, or the type of applications on your network. Protocol analyzers need to be configured, and you need a certain skill set to use them effectively; nonetheless, these tools are extremely valuable.

Chapter 4, "Optimization and Troubleshooting Tools," discusses in detail the internal operations and workings of a protocol analyzer. It also reviews some well-known, field-applied protocol analyzers on the market today.

It is important to understand that a protocol analyzer actually enables you to examine the protocols on a network. The protocols are the actual transmission vehicles used to get data from one point to another within the internetwork, regardless of the entity doing the transmitting on the network. The only way to actually view and troubleshoot any problems in the protocols themselves is with a protocol analyzer.

Some tools available today enable you to view just the statistics from the protocols and how they affect the network. These tools could be appropriately labeled monitoring tools, but they are not true protocol analyzers. A protocol analyzer is more in-depth than a monitoring tool, and enables you to view the internals of the protocol and its operations.

Understanding When To Perform a Network Optimization Project

Good network optimization requires an understanding of when to use a protocol analyzer or network monitoring tool to obtain the statistics for the particular phase of network optimization that applies at a given time. Some problems identified while troubleshooting a large internetwork immediately trigger the need for a protocol analyzer or a network monitoring tool. (Later in this book, a detailed discussion illustrates the differences between an internal operation of network monitoring tools and protocol analyzers.)

Before you go any farther, the following terms should be defined:

When a network is newly installed or implemented at a single or multilocation site, it is important to evaluate the overall health of the network. But when a network has grown over a long period of time and has undergone a number of modifications, this network is considered a mature, technically migrated network, and will most likely exhibit a range of behaviors that require a network monitoring tool or a protocol analyzer. Usually these networks are prime candidates for network optimization to attain their best possible performance.

When networks are not performing well and are having problems that require troubleshooting, a better approach might be to attempt the optimization of their performance and operation, rather than just troubleshooting to resolve the current problem. In other words, a company might need to take a fresh look at its approach to network problems: instead of simply isolating the problem, it might be better to examine the entire network using a protocol analyzer or monitoring tool to optimize the network's performance.

A focal point of network optimization today is that many networks now are connected to other networks through wide area network connections. A variety of wide area connection methodologies and approaches are being used, from direct connect circuits to frame relay implementations. All these types of configurations need to be optimized. A protocol analyzer enables an analyst to perform communication link sizing and optimization of throughput from link to link.

For networks currently in place that have undergone extensive technical migration over a period of time, the focus should be to clearly examine the network from its baseline. Chapters 3 and 5 discuss network baselining: what it is, and how to use it as a study methodology for networks with problems.

Networks that have undergone extensive technical migrations may have a range of technology components implemented throughout their infrastructure. For example, networks may have multiple file servers in place; multiple applications placed at different points; a range of different user groups using services from multiple points; a group of network communication connections, including incoming and outgoing lines for a variety of reasons; and an extensive range of general network usage hardware and software points that are key and are utilized throughout the infrastructure.

A network that has undergone extensive migration usually is a network that has frequent changes made. In today's networking environment, the staff who maintains this type of network often does not have time to measure the effects of each change; the tools, however, are available to do so. Ways are available to measure these changes relatively quickly, and to gain an understanding of how the network performs from phase to phase and change to change.

Each network operates in a characteristic manner. As new hardware and software technologies are implemented upon the network, certain operational statistics can show how the network has changed in response to that particular modification. A protocol analyzer can be used to examine the network after any major modification has been implemented on any network design. The protocol analyzer can be used to examine the traffic flow and a range of statistics on the network, after the changes are implemented. This enables the network manager or analyst to look at the efficiency and performance of the network, and how the change actually affected its performance.

Through proper documentation and organization skill sets, the analyst or network manager can measure the change with a protocol analyzer or network monitoring tool, and then document that change. This enables the network management team to take a very close look at whether a particular modification was warranted and if the change actually improved or affected the operation the way it was intended.

The goal here is to start utilizing an analyzer to optimize the network's performance as changes occur. The products should be picked not only for their functionality, but also for their utility in optimization. The future holds many new requirements for networks; the network bandwidth, along with all the hardware and technology investments' can only provide so much capacity. The tools available today enable an analyst to make maximum use of each network and optimize it, in the face of daily changes.

New Network Implementations
The installation of a new network usually involves a variety of personnel and a group of products to be implemented, all in a fairly short period of time. Usually the approach is to have some sort of project plan, implement the products, and turn the network live into a production phase. A range of tests can be run with a protocol analyzer or network monitoring tool to ensure that the network is performing at its optimum level from the day it is turned back to production mode. This book discusses various baselining and testing statistics that can be examined with network monitoring tools and protocol analyzers to verify the health of a newly installed network.

The key is to make sure that a network operates at its optimum level on its original production installation date. Because a new network installation can involve new protocols, new applications, new workstation and file server shells, and new network operating systems, it is critical to utilize a protocol analyzer to focus on the performance of the applied protocols immediately upon the network's initial implementation.

In addition, many shell and file-server configuration parameters need to be set on the initial installation of a new network that will affect network communication processes. A protocol analyzer enables the user to examine the network performance upon installation, and to make changes to configurations on both workstations and file servers to implement the best network communication performance possible.

Industry-standard throughput considerations exist for how fast network data is transferred across networks and how the network performs. This book discusses in detail those throughput levels and standards. A modern protocol analyzer is equipped for network analysis in all the main network performance areas.

Reactive Problem Analysis
Internetwork support means daily troubleshooting of a range of issues and problems. A large number of complaints about internetworks today concern performance: network runs slow, network file server hangs, interruptions of service during certain file server operations, and so on. Clearly, a high number of these connection problems are related to network communication failures due to hardware or software configuration and layout problems. A protocol analyzer specifically allows for examining the network communication that occurs between workstations and file servers in all key internetwork points of communication.

A protocol analyzer can be used to examine from beginning to end a conversation that may occur between a workstation and a file server. The specific time of a conversation indicates whether the conversation occurs in what is considered a normal range. Later in this book, those standards and ranges are discussed along with how to examine those timing intervals. The point is that a protocol analyzer can be used to examine the communication on networks that are having problems, and you can troubleshoot a large number of the network failures and performance problems from the network communication point of view. Many times, a simple configuration parameter or a small modification to the hardware or software layout is all it takes to make the network perform at its optimum level.

This book discusses several techniques that focus on the performance and timing operation for network communications. Be aware that when certain types of failures occur (for example, a file server cannot be located or a certain operation is aborted), a protocol analyzer can be used to examine the timing between the particular processes of the workstation and file server, and may clearly identify the source of the problem for the network timing problem that is occurring.

Some statistics obtained from network monitoring tools and protocol analyzers can identify whether other contributing factors are on the network, such as a high traffic level, that might cause a symptom to occur. Chapters 3 and 5 discuss all the techniques and operations for optimizing a network for its best communication timing to eliminate impediments to network performance.

Wide Area Network (WAN) and Internetwork Analysis
As mentioned earlier, in today's corporate environment many networks are being connected to other networks to create a wide area network (WAN) configuration. This happens frequently in larger corporations, but many small companies also are finding the need to interconnect to remote sites for network communications. This is a good time to give you some definitions.

An internetwork is a group of networks that connect together across common or multiple topologies to communicate with common or multiple protocols to achieve full interoperability. The term open networking also is used sometimes. A wide area network (WAN) is a network or internetwork configured over different geographical locations. Note that the focal point of networks is to have communication between a range of different entities without limitations and with the most productive computing possible. Protocol analyzers enable the user to examine communication between multiple networks to determine the effectiveness of operation along with the measure of true interoperability across multiple communication links.

When multiple networks communicate, extensive routes usually are traveled between the various network points by the different processes involved. Point-to-point timing issues are therefore critical. A protocol analyzer or network monitoring tool, if properly implemented, enables an analyst to examine transaction times between different network points within an internetwork.

Communication links are required from point to point within the networks, and these need to be configured for optimum communication. If communication links are not properly configured, bottlenecks can occur and limit point-to-point data speed. The communication links chosen need to be sized to handle the applied/proper data flow for the different network concerns. A protocol analyzer enables the user to examine the overall communication link sizing between different points, and examine whether the throughput actually is performing at its optimum level. Overall, the user can examine these internetwork layouts from point to point, to verify that the wide area network is configured the way it should be and is performing at an optimum level.

A range of devices are involved in a wide area network configuration, including bridges, routers, hubs, and communication link configurations. All these devices need to be examined, and protocol analyzers and network monitoring tools are the way to do it.

Network Modification Analysis
After any major network modification, a critical final step is to retest the network with a protocol analyzer to get a benchmark of the network's performance level. Only one change should be made to a network at a time, followed by retesting the network with a protocol analyzer to see the effect on performance.

Testing the network before and after the change enables you to evaluate and document the performance effect of the network modification. For example, if a protocol analyzer captured a particular workstation-to-file-server read at one minute and there now has been a 16-megabytes upgrade in file server memory, the network read should be remeasured; this is called post analysis. After the analyzer test is complete, the results of the post analysis shows a workstation-to-file-server read of 20 seconds--this is a 40-second increase in performance for the particular task.

Now that you better understand what network optimization is and why it is required to keep an internetwork functioning at its optimum performance level, the next chapter introduces the two main topologies utilized most frequently for internetwork configurations today: Ethernet and Token Ring. Chapter 3 is a discussion of network optimization theory; Chapter 4 focuses on optimization and troubleshooting tools and their applied techniques.




(255)