Routing/Domain Name Service/Network Monitoring/Address Resolution Protocol
Overview
The TCP/IP Administration consists of three related exercises, it will
take less time overall if you can complete them as a unit.
There are several topics covered in the assignment. Static routing
will be examined in the context of using a W2K server to route IP packets
between two IP networks. Domain name service (DNS) will be implemented
for the hosts on the two IP networks. Network monitoring of traffic will
be performed using the Network Monitor. Address Resolution Protocol
used to translate an IP address into a NIC address on a LAN will be examined
using the monitor results.
Equipment/Supplies/Time
- One PC located in LF111 with two NICs labeled B438. It should the middle
machine on a table with three PCs.
- Two PCs on one table in LF111 on either side of the B438 PC.
- Two 10base-T Ethernet cables with crossed wiring to eliminate the
need for a connecting hub. These are labeled B438 Crossed Wires. These
should be in the B438 cabinet or attached to the two adjacent
computers.
- Your removable hard disk with W2K Server installed.
- Time
- Routing - About 45 minutes.
- DNS - About 10 minutes.
- Network Monitoring - About 60 minutes.
Routing
The general idea is to connect independent IP subnetworks into a single
network. The connection point is a router. We will use static routing
to construct a routing table by hand on the router (W2K server) that instructs
the router where to send IP packets. Note that a router normally has two NICs in
order to connect two IP subnets.
The following diagram illustrates how the IP subnets 149.160.24.xxx and
149.160.25.xxx can be joined by an IP router. The router has two NICs, one with
IP 149.160.24.1 and 149.160.25.1 on the appropriate nets. The IP client NICs
have been assigned the IP numbers of 149.160.24.2 and 149.160.25.2.

For IP routing, both the client and router need information about the
network. The clients need to know the gateway router to send off net
packets. The router needs to know where to send packets, for example that a
packet arriving on NIC
149.160.24.1 for net 149.160.25.xxx should be routed through NIC
149.160.25.1. Note that this is an obvious routing example, where both the
router and destination client are on 149.160.25.xxx. A more realistic example is
at right where several routers must be crossed to reach a host. The A
router must know that a packet arriving for new 149.160.26.xxx (it could only
arrive on NIC 149.160.24.1) must be forwarded on NIC 149.160.25.2 to reach
149.160.26.2 host. Router B must also know that a packet for
149.160.26.xxx must be sent through NIC 149.160.26.1.
The assignment is to construct two networks and join them through a router.
The networks are simple, a single client directly connected to your server using
a cable with the wires crossed (similar to a null modem) to eliminate the need
for multiple hubs. The server will be configured for routing IP packets, the
clients must be configured to send packets to the router gateway.
Routing Instructions
The following provide certain details, though not all instructions, needed to
construct two IP subnets and install the router. The first step is to configure
the server NICs with static IP numbers and the Router services.
Server NIC
- On server machine:
- Disconnect the LAN cable from the back of the machine.
- Connect server machine to two crossed wire cables.
- Login on the server as Administrator.
- Open Settings | Network and Dialup Connections and click on either
connection icon as in the figure:

- Select Properties | TCP/IP from list as at right | Properties button:

- Change the properties for the NIC to the IP numbers you wish to use. The
figure at right is host IP address 149.160.29.3 and uses itself as the Domain
Name Server. The subnet mask should match that used by other hosts on the
network.

- Click back through the OK buttons for the changes to take effect.
- Do the same for the other NIC but use a different IP address, the
DNS server can be the same.
Server Routing
- Open Programs | Administrative Tools | Routing and Remote Access
- Router Configuration - By default the server has IP routing should be
enabled, you must define the route that packets must take to be delivered to
another IP network.
- Right click on the computer name.
- Select Configure and Enable Routing and Remote Access
- Select Network Router and follow the wizard.
- Select IP Routing | Static Routes
- Right click, select New Static Route...
- Interface: Local Area Connection
- Destination: IP of destination LAN, for any IP starting with
169.254.98 use 169.254.98.0 but can use any IP when not connected to IUS
LAN.
- Network mask: AND filter, to allow any 169.254.98.xxx IP to be
routed use 255.255.255.0 as a mask. The mask is ANDed with the IP to
define the valid IP numbers on the net. A net with 256 IP numbers could
assign IPs from 149.160.25.0 through 149.160.25.255 using a mask of
255.255.255.0 while a net of 149.160.25.0 through 149.160.25.7 has a mask
of 255.255.255.248 where the mask has 1's in all but the last three bits.
ANDing the address 149.160.25.3 with the mask produces 149.160.25.0 which
is part of the eight host net. You must choose a mask that matches the
network IP.

- Default gateway: IP of NIC to serve as gateway for
routing. The default gateway on the router can be the same as the NIC IP.
On a client, the default gateway should be the router IP to which the
client is connected. The 169.254.98.0 subnet must be
physically connected by the crossed wire cable to the gateway.
- Metric: 3
- Repeat the above steps for the other NIC adapter. Use a different
destination subnet.
- Action Refresh to have the changes take effect. The static routes
should appear similar to below:

- The routes and masks defined construct a routing table that will route a
packet from host 169.254.47.56 to host 169.25.98.4 through the router. The
router receives the packet with destination 169.25.98.4, AND's with the MASK
255.255.255.0 to determine the destination subnet of 169.254.98.0, the
router sends the packet out gateway 149.160.29.2 to reach subnet
169.254.98.0 and host 169.25.98.4.
- 1) Print - At a Command Prompt, capture the screen for a display
of the current routing table: route print
Client NIC - The two client machines adjacent to the server must have
static IPs installed and instructed where to send off net packets as was done
for the server.
- On each client machine:
- Disconnect the LAN cable from the back of the client machine.
- Connect each machine to the server using one of the crossed wire
cables.
- Login to the client machine by its name, not using ADS.
- The machine name will be SE....
- User: admin
- Password: cscib438
- IP - Follow the instructions for the Server NIC
but use an IP number in the network given in the Destination for the
Routing and Remote Access. You will need to assign the default gateway.
Router Testing - The following is the basic test of an IP network,
determining whether a path exists from one point to another using ping. The
examples are using the IPs from the simple network diagram above, you'll need to
use your network IPs.
- At the Command prompt on server:
- 2) Print - Use ping to verify that the router (server) can reach
its own NIC IP addresses.
- 3) Print - Use ping to verify that router can reach each host.
- At the Command prompt on each host:
- 4) Print - Use ping to verify that each host can reach the router
NIC.
- 5) Print - Use ping to verify that one host can reach the other.
- Problems
- Nothing works - Probably using the wrong IP in the ping.
- Router can ping itself but nothing else - Switch the LAN wires on the
back of the server, the gateway and hosts should be on the same channel.
- Can't ping an IP across the router - Router configuration may not have
refreshed. Click on Static Routing, View the Routing table. Each of the
routes entered should appear in the table, if not try to refresh again. The
refresh occurs by default every 60 seconds but the results are somewhat
unpredictable.
- Manual route entry - The routing table can be examined and manually
entered at the command prompt. route will display the help page. To
manually enter the route for
:
- route add 169.254.98.0 mask 255.255.255.0 149.160.29.2 metric 3
- route add 169.254.47.0 mask 255.255.255.0 149.160.29.3 metric 3
- 6) Print - If the router and hosts can ping each other, the browser on
either host should be able to reach the server homepage. Connect from either
host by entering the IP location of either server NIC. You should have a
test.htm file that can be accessed via a browser. Assuming the server has
an NIC with IP address 149.160.25.1 the following should display the
page:
- http://149.160.25.1/test.htm
Finish - If you are done for the day, restore the cabling and TCP/IP
configuration back by: 
- Reconnect NIC of the two machines to IUS LAN. Server connects via the
built in NIC.
- TCP/IP - Follow the instructions for the Server NIC
setting to automatic configuration.
- Put the crossed wires back in the B438 cabinet or leave
attached.
Domain Name Service
The general idea of DNS is that it is easier to remember names than IP
addresses and it is simpler to have name administration done centrally than each
host maintain its own name database. DNS is a service that translates the
name into an IP that is used for routing, etc. over the network. The
names are used as indexes in a database of IP numbers. Hosts needing DNS must
know the IP of the machine with the DNS database, usually the DNS machine is
part of a heirarchy of name servers. For example, DNS requests for
ius.indiana.edu would be forwarded to the DNS machine for edu which
would forward to indiana.edu which would forward to the
ius.indiana.edu DNS machine. Our test network is not connected to a DNS
heirarchy so only those names defined by us will be translated into IP
addresses.
DNS Instructions
DNS Server Instructions
- Reconnect the machines using the crossed wire and reconfigure the
client hosts to the same IP addresses and gateways consistent with the
router(server)
configuration.
- Login on the W2K Server as Administrator.
- Programs | Administration Tools | DNS
- Click on the server icon | Action | New Zone | Accept the Wizard
defaults
- Enter zone using your last name following the pattern: wisman.com
- Double click on Forward Lookup Zones | Click on zone name
| New host
- Enter name and the assigned IP addresses for server and each of the
two hosts (it may be helpful to name them www, left and
right).
- Action | Update Server Data File
Testing - With the W2K Server running routing and DNS, and the clients
using the server as a gateway and DNS server, you should be able to test using
either raw IP or names.
- 7) Print - At the Command prompt:
- On the left host ping the router.
- On the right host ping the router.
- On the right host ping the left host.
- 8) Print - If the pings worked, try the browser on either host to
reach the W2K server page test.htm using the server name.
Network
Monitoring and Address Resolution Protocol
The general idea of a packet monitor is to selectively examine traffic on a
network. W2K Server comes with a Network Monitor package that monitors
the traffic
to and
from the server, much more powerful monitors are commerically available that
monitor all traffic on a network. The network monitor can selectively monitor
traffic for selected NIC address numbers, protocols (IP, IPX, etc.),
computer names, etc. combined using Boolean operators. The philosphy of
Network Monitor is to capture traffic and then view the
captured data.
The exercise is intended to provide some experience with traffic monitoring
and specifically to examine Address Resolution Protocol (ARP). ARP is
used by LAN stations to translate the destination IP number into the NIC number.
The figure at right lists the ARP table generated when a ping of an IP
address is made on a LAN. Since LAN traffic uses only NIC numbers for addressing
with the IP packet carried in the frame data load, ARP is needed for IP address
resolution to an NIC address before two LAN connected machines can communicate
using the higher level IP protocol numbers of the ping command. The
command arp -a lists the current table of IP to NIC addresses. Using the
Network Monitor the ARP interaction can be fully examined.
Network Monitor Instructions
Configuration
- Reconnect the machines using the crossed wire and reconfigure the
client hosts to the same IP addresses and gateways consistent with the
router(server) configuration.
Monitoring - With the W2K Server running routing and DNS, and the
clients using the server as a gateway and DNS server, you should be able to test
using either raw IP or names. To monitor traffic some traffic will need to be
generated. You'll notice that the server generates some traffic by broadcasting
service advertising packets, you will generate traffic using TCP/IP applications
such as ping and a Web browser.
Server
- At a Command Prompt, generate ping and ARP traffic by pinging
each client host.
- 9) Print - List the ARP table by:
- arp -a.
- Remember which IP address corresponds with an NIC address.
- At a Command prompt clear the ARP table.
- At the server command enter: arp -a
- Remove each table entry, for example Internet Address of
149.160.29.92 by: arp -d 149.160.29.92
- Start Network Monitor - Programs | Administrative Tools |
Network Monitor
- Select either one of the two router NICs to monitor, listed by the
corresponding NIC address.
- Capture | Start
- At the server Command prompt, ping the client IP address on the
NIC being monitored to generate ARP traffic.
- View the captured data - Capture | Stop and View
- To examine the frame or packet in detail double click the frame in the
Summary window. Two additional windows should open that provide
detail down to the individual bytes of the frame.
- 10) Print - Windows similar to that below should be displayed. Select
the ARP: Request.
- In the selected frame below the ARP: Request, Target IP: 149.160.29.92
is being broadcast onto the LAN to discover the NIC number that
corresponds to the IP. This can be seen in the bottom HEX window where the
destination NIC address is FF FF FF FF FF FF followed by the
source NIC address of 00 0C F0 40 73 2E. The ARP request broadcasts
the target IP, if the station with that IP is on the LAN it responds back
with its NIC number. The requesting station stores the IP/NIC
correspondence in its ARP table for use whenever an IP number to NIC
translation is needed. Note that in the bottom pane is the Ethernet frame
contents, the FF FF FF FF FF FF is a NIC broadcast destination address,
the 00 C0 F0 40 73 2E is the NIC source of the ARP requestor.

- Select the ARP: Reply frame. The previous ARP broadcast by IP
149.160.24.1 of the IP 149.160.29.92 address has received a response from
NIC 000CF040732E as seen in the Hdwr Addr: and in the HEX window.

- At the Command Prompt display the ARP table to verify that the table
has been updated with the NIC address of a client.
Client - The client browser will be used to access the router Web
server. This will generate communication of an ARP, DNS, TCP, and HTTP
protocol interchange between the client and the server so that each protocol
frame or packet can be examined in detail.
- Client - Verify that a client browser can access the test.htm
page on the server using the server DNS entry (e.g. http://www.wisman.com/test.htm).
- Client - Restart the client machine to force it to forget the ARP and
DNS lookup since we want to monitor the process.
- Server - Start the Network Monitor and capture from the router
NIC address connected to the client machine.
- Client - Access the test.htm page on the server.
- Server - Capture | Stop and View | Window | uncheck HEX | Window |
Check HEX
- 11) Print - You should see a window similar to that below. Click on
the frame to examine in the Summary window. The GET request from
the browser on IP 149.160.29.92 is displayed, in the Hex window the data
sent from the browser to the server can be viewed as:

Performance Monitoring
W2K provides a tool for monitoring a rich set of factors affecting network
performance. These include CPU, Web services, raw network byte count and many
others. The following explores the use of the performance monitor tool of
the Network Monitor to determine the approximate maximum number of bits that can
be sent over the network between two machines.
Programs - Two programs are used to exercise the network: PERFserver
and PERFclient. The server receives and displays bytes to the screen, the client
sends bytes to the server over the network as TCP segments. The client has three
parameters:
- IP - the IP number of the PERFserver.
- bytes - the number of data bytes to send in each TCP segment. Range is 0
to 231.
- packets - the number of packets (well really segments) to send. Range is 0
to 231.
The server has one parameter: the number of bytes to allocate for the receive
buffer. Generally the larger the faster it can process bytes received.
Performance - Performance is affected by many variables. In this
exercise only a few variables can be controlled: the client and server
parameters, whether the server displays received bytes to the screen or writes
to a file (which is faster), and whether one client and or two clients are
transmitting. As discussed in the text, utilization of a resource is higher for
larger data transmissions so one would expect the maximum to be reached when the
number of bytes transmitted at one time was large.
Monitoring - A counter monitors some performance factor such as the
network card total byte count. Several counters can be defined to monitor and
display in graph form several factors simultaneously. The following details
running the Performance Monitor and placing a load on the network using the
client and server programs.
- Download to your server and both connected hosts the
PERFserver and
PERFclient programs. You'll need to run
on both machines.
- In the Network Monitor click Tools | Performance Monitor
- Right click in right window | click on Add Counters
- Select Performance Object: Network Interface. See figure at right.
- Select Bytes Total/sec and click Add button. Do the same for
Bytes Sent/sec and Bytes Received/sec.
- Select Performance Object: Processor.
- Select %Processor Time and click Add button.
- Close.
- The display should be updated each second. The figure at right is a graph
of a network under load.

- Load the network with data.
- Determine the IP of the machines using ipconfig /all.
- On one host machine run the Command prompt run the PERFserver program
using defaults by:
PERFserver
- On the other host machine run the PERFclient using defaults by:
PERFclient <IP of PERFserver>
- The graph should display similarly to that at right and give the maximum
number of bytes for a counter. You can click on each counter to read
the performance measures.
- 12) Print - Determining maximum bits transferred. Try to increase the load
on the network and processor to determine the maximum under different loads by
the following:
- Run both PERFserver and PERFclient simultaneously on each machine with
PERFserver output to the screen.
- Redirect the PERFserver output to a file instead of the display by:
PERFserver > x
- Send a larger number of bytes from the client to the server. To send 50
segments of 10000 bytes in each segment to 149.160.29.92:
PERFclient 140.160.29.92 10000 50
Turn In
- Routing
- DNS
- Network Monitoring
- Performance Monitoring
- Screen shots showing maximum and graph for Bytes total/sec. from
tests. Discuss your observations regarding interaction of
execution of PERFserver, PERFclient, network load and processor utilization.
Document last modified: