Packet loss, Network topology – Lenovo Secure Managed Client User Manual

Page 16

Advertising
background image

If you notice MAC flapping or duplicate packet error message in your switch log,
you can set up an Etherchannel across the 6 switch ports that you assign to the
storage array.

Note:

If you need to reprovision the storage array, you will have to remove this

Etherchannel for the reprovisioning to be successful. See “Provisioning the storage
array” on page 16

Packet loss

If Secure Managed Client has poor performance when first deployed, the network
should be checked for packet loss. Packet loss occurs when one or more data
packets fail to reach their destination over a computer network. This can happen
for various reasons: signal degradation, network problems, overloaded network
links, or corrupted packets. If the packets are lost using network based technology
there can be a significant performance impact. It is important to remember that
some packet loss is expected. Modern infrastructures tend to have an acceptable
packet loss rate of one percent. Packet loss in the Secure Managed Client
environment can impact client performance since the clients are virtualized over a
network. It is important to consider this issue when selecting hardware for the
Secure Managed Client infrastructure.

If the Secure Managed Client environment contains managed (smart) switches, it is
possible to measure packet loss through the individual ports. This is the preferred
method as it provides data at the switch level. However, if the packet loss occurs
at the switch, it might not detect the dropped packets. Dropped packets in the
switch’s statistics are usually drops that it detects due to issues such as framing
errors or bit errors. These are not the same as packets being dropped due to errors
in distributing the packets across multiple physical links. Another way to check for
dropped packets is to use a network performance tool that floods the network with
packets and then checks to see that they are being delivered correctly. An example
is Issuing the Linux ping –f command.

Another way to measure packet loss is to place a node in between the hardware.
For example, a Windows or Linux machine with two network cards can be placed
between a single client and its respective switch.

Network topology

A typical topology in a Secure Managed Client solution would consist of:

v

A central switch with the 6 1 GB links aggregated to the storage array

v

4 to 5 satellite switches, each with 1 GB links to the central switch

v

0 to 25 clients per satellite switch with 1 GB or 100 Mb links

8

Secure Managed Client Version 2.0 Deployment Guide

Advertising