![[previous]](../../../graphics/l_arrow.gif){kind=icon}
![[next]](../../../graphics/r_arrow.gif){kind=icon}
Use these sections to identify and correct problems that are indicated by changes in bandwidth utilization:
To determine how your network is operating on a day-to-day basis, examine its bandwidth utilization. Changes in utilization can alert you to actual or potential problems.
Utilization varies depending on the media and on how your network is configured and used. Become aware of your network's normal behavior so that you know when to examine utilization levels more closely. See "Identifying Your Network's Normal Behavior" for more information.
Determine the current utilization of all media on your network (Ethernet, Fiber Distributed Data Interface, token ring, and Asynchronous Transfer Mode) to determine whether utilization rates are exceeding thresholds that you have set in the management software.
On most networks, utilization gradually increases as users begin using more network resources, such as electronic mail, network printing, and file sharing. Be concerned with utilization peaks that do not follow this pattern of use.
The process of identifying immediate utilization levels is discussed in "Identifying Utilization Problems".
Examine your network's historical trends (its typical utilization over time) and note whether your network has experienced a gradual or sudden increase in utilization. Here are ways to assess trends:
The process of identifying historical trends is discussed in "Generating Historical Utilization Reports".
A high rate of utilization can lead to high rates of packet fragments. As utilization exceeds the alarm threshold, packet fragments become common. See "Ethernet Packet Loss" for information about identifying when packet fragments are occurring.
Narrow the utilization problem to the ports that have excessively high or low utilization. If necessary, redistribute network traffic accordingly by segmenting your LAN with a bridge, router, or switch.
Sometimes, a hardware problem can cause abnormal utilization rates. In this case, see "Ethernet Packet Loss" and "FDDI Ring Errors" for troubleshooting information.
First, determine utilization levels on your current network. Try to locate the segments that are experiencing high or low utilization levels.
Use Status Watch, which collects MIB-II data using Simple Network Management Protocol polling, to determine bandwidth utilization.
The Status Watch utilization tools monitor the amount of traffic on network segments and show how the bandwidth is being allocated. These tools provide a real-time report of utilization data on the selected device or group of devices.
Table 14 describes the Status Watch tools that monitor your network's utilization.
Follow these steps:
1 . Select the group that you suspect has a performance problem.
2 . Select the utilization tool icon that indicates a problem.
If an increase in utilization causes an increase in Error rates (other than collisions), look for MAC and physical layer problems (for example, faulty network cards, illegal repeater hops, and cables that are too long). Additionally, monitor Collision rates as utilization rises, looking for large increases that are out of the ordinary. In particular, search devices on the segment for "Excessive Collisions". While Collisions are normal, Excessive Collisions means network delays.
Use real-time utilization data to see how your network is operating at the moment. To gauge whether utilization is at a critical point for your network, look at historical data. Use Web Reporter to generate a historical report that shows the utilization trends for a specific set of devices on your network.
Using Web Reporter, you can save days and weeks of network data, save a baseline week of "normal" data, and determine when utilization is constantly high.
Follow these steps:
1 . Access Web Reporter. Use as the uniform resource locator (URL) the directory where you installed Transcend® NCS on the Web.
2 . Generate a weekly Historical report to see utilization rates for the whole week.
3 . Compare your weekly Historical report to a baseline of historical utilization data.
This section explains terms that are relevant to bandwidth utilization and provides additional conceptual and problem analysis detail.
Over time, if a port has experienced increased, sustained utilization levels, then you need to balance the load of your ATM segments.
Status Watch calculates ATM utilization in this way:
greater of (in_util, out_util)
where:
in_util =
( ((rate of ifInOctets)*8) / ((linespeed)*0.9875) )*100
out_util =
( ((rate of ifOutOctets)*8) / ((linespeed)*0.9875) )*100
The 8 factor converts octets to bits.
The 0.9875 factor offsets the interframe gap.
Over time, if a port has experienced increased utilization levels (often a sustained level of over 40 percent), then you need to rebalance the load of your Ethernet segments.
Typically, the larger the frame size, the more utilization your network can accommodate.
You may recognize utilization problems with certain protocols before other protocols because some protocols have less tolerance for high rates of traffic. When utilization becomes a problem also depends on users. For example, you may allow higher utilization rates on an engineering network, yet you want greater bandwidth availability on a financial network where data delivery is critical.
As general guidelines, your network is healthy in these conditions:
If the 30 percent utilization peaks start occurring very close together, your network starts showing signs of degraded performance.
Status Watch calculates Ethernet utilization in this way:
in_util + out_util
where:
in_util =
( ((rate of ifInOctets)*8) / ((linespeed)*0.9875) )*100
out_util =
( ((rate of ifOutOctets)*8) / ((linespeed)*0.9875) )*100
The 8 factor converts octets to bits.
The 0.9875 factor offsets the interframe gap.
FDDI accepts utilization levels that are equivalent to its rated speed. Unlike Ethernet, FDDI does not have delays and problems that cause collisions.
The best way to determine high FDDI utilization is to know the normal capacity of your FDDI network. Generally, if your FDDI network is consistently reporting 90 percent or more utilization, plan to balance the load on your network.
Status Watch calculates FDDI utilization in this way:
(1 - (delta(token_count)*latency) / delta(time) )*100
Token ring media accepts utilization levels equivalent to its rated speed. Unlike Ethernet, token ring does not have delays and problems caused by collisions.
The best way to determine high token ring utilization is to know the normal capacity of your token ring network. Generally, if your token ring network is consistently reporting 90 percent or more utilization, plan to balance the load on your network.
Status Watch calculates token ring utilization in this way:
( ( rate*8) / (speed) )*100
where:
rate = ifInOctets / delta(time)
speed = line speed of 4 or 16
The 8 factor converts octets to bits.