Mitel Voice Processing Solutions Instructions Manual

							Overview of TCP/IP 
Subnet Addressing 
Organizations are often assigned one Class B network number by the Internet 
Advisory Board. If the organization needs multiple network numbers to distinguish 
among its servers, the network administrator uses subnet addresses to refer to the 
multiple networks. Subnet addressing is a mechanism whereby address bits that 
would otherwise be part of the host number are designated as part of the network 
number. 
In a class B address, its format is nnnnnnnn.nnnnnnnn.hhhhhhhh.hhhhhhhh, 
where n stands for a bit in the network number and h stands for a bit in the host 
number. To increase the amount of available network numbers, host bits from the 
third octet are used as network bits, which, for each bit borrowed, doubles the 
number of usable network numbers and halves the number of host numbers. The 
format becomes, for example, nnnnnnnn.nnnnnnnn.nnhhhhhh.hhhhhhhh. 
Borrowing bits from the host portion is accomplished using a subnet mask. 
Subnet Mask 
To create a subnet, you apply a bit mask, known as the subnet mask, to the II’ 
address. If a bit is on (1) in the mask, the equivalent bit in the address is treated as a 
network address bit, whether or not it was a network bit originally. If the bit is off 
(0), then the equivalent address bit belongs to the host address. 
The subnet mask is a 32-bit number with a similar format as the II’ address. Default 
subnet masks are 255.0.0.0 for class A, 255.255.0.0 for class B, and 255.255.255.0 
for class C. Expressed in binary form, the network mask for class B is 
11111111.11111111.00000000.00000000. Th 
eoretically, in order to provide for 
two subnetworks with 32,768 hosts per subnetwork, you could use the network mask 
255.255.128.0. However, because of restrictions on the use of some subnet values, 
you nked to use the subnet mask 255.255.192.0 (the third octet is 11000000) to 
obtain two subnetworks, 01 and 10. e remaining six bits from the third octet, 
plus the entire fourth octet, are used for up to 16,384 host addresses. (Network 00 
identifies the network and 11 is for broadcast.) 
Subnet masks can be expressed in bit or decimal format. While decimal format is 
easier to read, remember that network address portions can cross the octet 
boundaries, and using the decimal masks might be confusing in such cases. 
Ask your network administrator if any subnet masks are used on your network. 
tin 
Routing is the process of sending a message to either a destination host (if it is on the 
same network) or a router or a gateway (if the host is on another network). Routing 
is based on the network portion of the II’ address. If the destination network is on 
the local network, the subnet mask is applied. 
2-G  
						

							Overview ofTCP/IP 
ARP, Address Resolution Protocol, translates IP addresses to Ethernet addresses. It 
maintains a table of both sets of addresses, which is built dynamically. 
As data moves up and down the TCP/IP layers, each layer must be able to handle it. 
Data is combined as it moves from applications to transport protocols, and then to 
the Internet Protocol. Many sources of data are combined, and then must be 
separated when they arrive at the destination network or host. II’ uses protocol 
numbers to identify transport protocols, and the transport protocols in turn use port 
numbers to identify applications. 
Some protocols are reserved for specific services, such as standard network protocols 
(for example, fip and telnet). They are defined in the file /etc/protocols. Figure 
2-2 shows a sample protocols file. 
ote: This chapter has several examples of files used with TCP/IP, which 
are provided for your understanding. You cannot access these files 
on the Series 6 server. Contact your Centigram representative 
should you need to change any TCP/IP files on the server. 
# /etc/protocols 
# 
# format is: 
# 
# protocol number aliases 
# 
ip 0 IP # internet protocol, 
pseudo protocol 
icmp 4 ICMP 
# internet control message protocol 
tcp 6 TCP 
# transmission control protocol 
udp 16 UDP 
# user datagram protocol 
Figure 2-2 Sample /etc/protoeols File 
port Numbers 
Port numbers below 256 are reserved for specific services (for example, fip and 
telnet). Port numbers from 256 to 1024 are reserved for UNIX-specific services, 
(for example, rlogin). These numbers must be unique within a specific transport 
protocol. The combination of protocol and port number identifies a process to send 
the data to. Port numbers are defined in the file /etc/services. Figure 2-3 shows a 
sample services file. 
2-7  
						

							Overview ofTCP/IP 
# /etc/services 
# format is: 
# service 
port/protocol aliases 
# 
# Network services 
# 
ftpdata 
20/tcp ftpd 
ftP 21/tcp 
telnet 
23/tcp telnetd -d 
smtp 
25/tcp 
bootps 
67/udp bootpd 
bootpc 68/udp bootp 
tftp 69/udp tftpd 
snmp lGl/udp snmpd 
echo 7/tcp 
echo 7/udp 
discard 9/tcp sink null 
discard 9/udp sink null 
daytime 13/tcp 
daytime 13/udp 
chargen 19/tcp ttytstsource 
chargen 19/udp ttytstsource 
time 37/tcp timserver 
time 37/udp timserver 
finger 79/tcp 
domain 53/tcp nameserver #name-domain server 
domain 53/&p nameserver 
nb-nn-srv 137/udp netbios nameserver 
nb-sssrv 139/tcp netbios session server 
nb-dgsrv 138/udp netbios datagram server 
# 
# UNIX specific services 
# i 
exec 512/tcp -.h 
login 513/tcp rlogin rlogind 
shell 514/tcp rsh rshd cmd 
who 513/udp rwho rwhod who 
ntalk 518/udp 
route 52Q/udp 
router routed 
lOOO/tcp qnxserver 
Figure 2-3 Sample /etc/services File 
2-8  
						

							, - 
Overview of TCP/IP 
If two users are accessing the TELNET service, for example, the system would give 
each of them a different dynamically assigned port number for the source port, and 
the default port of 23 for the destination. The pair of port numbers, both source 
and destination, identifies a network connection. The combination of an II’ address 
and a port number is called a socket. A socket identifies a single network process 
within the Internet. A pair of sockets uniquely identifies a network connection 
between a receiving and a sending host. 
For example, if host 128.66.12.2 connects to host 192.178.16.2 using TELNET, 
and is assigned port 3382, the socket for the source side of the connect-ion is 
128.66.12.2.3382. The destination side socket is 192.178.16.2.23 (defaulting to the 
TELNET port number). These two sockets are a socket pair, and this identifies the 
connection. If another user from the same host connects to the same destination 
with TELNET, the second user would get a different source port number, for 
example 3610. The socket pair for this user would be 128.66.12.2.3610 and 
192.178.16.2.23. 
It is easier to use names for hosts than II? addresses. Names are easier to remember, 
and it is easier to notice a mistake in a name. Here are two services that allow you to 
use alphabetic names instead of numerical II? addresses. 
The host table associates the addresses and names. You can also designate aliases for 
hosts in this table. The table is in the file /etc/hosts. There is also a file called 
/etc/networks, which can translate between nenvork names and network IP 
addressFs. Figure 2-4 shows a sample /etc/hosts file. 
-z. 
# 
# Table of IP addresses and host names 
# 
128.66.12.2 mast.sail.com yacht 
127.0.0.1 localhost 
128.66.12.1 main.sail.com loghost 
128.66.12.3 jib.sail.com jib 
128.66.12.4 spinnaker.sail.com spinnaker 
128.66.5.2 genoa.sail.com genoa 
128.66.7.3 
boom.sail.com boom 
128.66.7.2 
halyard.sail.com halyard 
Figure 2-4 Sample /etc/hosts File 
Most large networks use DNS instead of the host table, but it is used for small 
networks or for all hosts on the local network, in case DNS is not running. 
2-9  
						

							Overview of TCP/IP 
2-10 
The Domain Name Service is used in large interconnected networks because it 
avoids the problems of a large host table. It also eliminates the need to keep 
updating your table as new hosts are added or changed. DNS is assigned port 
number 53, and its service is called domain. Do not confuse it with name service, 
port 42 (this is an older name service). 
DNS is a set of distributed name servers which pass information to each other as 
needed. The naming conventions are hierarchical. At the top level is the root 
domain, served by a group of name servers called root servers. 
There are two top-level domains, geographic and organizational. Geographic 
domains are assigned by country; each country has a two letter code. Other 
assignments are made underneath the country code. In the United states, the next 
level is the two-letter state code. Thus, a valid domain name could be 
myhost.sanjose.caus. The highest level of the domain is the last part of the name. 
The second type of top-level domain is organizational; assignment is made based on 
what kind of group is using the host. The top-level domains are listed in Table 2-6. 
Table 2-Q Top Level Domains 
Domain 
Name Used By 
COM Commercial organizations (businesses) 
EDU Educational sites (schools, colleges, universities) 
GOV Government agencies and groups, not including 
military y 
MIL Military organizations 
NET Network support organizations, such as sites that 
run root servers 
ORG Any organization that is not one of the above (for 
example, non-profit groups) 
Again, the highest level of the domain is the last part of the name. A valid domain 
name is 
whitehouse.gov; whitehouse is a site on the government domain. In both 
cases, domain names are written from most specific (for example, host name) to least 
specific (top-level domain).  
						

							This chapter describes the TCP/IP protocol suite and explains a number of specific 
concepts. If you are very comfortable working with TCP/IP, you can probably skip 
this chapter. 
TCP/IP protocols were developed for the ARPANET network of computers, 
administered by the Defense Communications Agency. This network began as an 
experimental packet switching network and the organizations involved with the 
project found it useful for their data communications. The TCP/IP protocol suite 
was adopred as a standard for ARPANET in 1983 and implemented in BSD UNIX, 
which led to its association with UNIX systems. TCP/IP is now used by most UNIX 
systems, and many other host computers and peripheral devices can be included in a 
TCP/IP network. 
This chapter is not meant to provide complete coverage of TCP/IP. There are many 
good books you can refer to for detailed information. One is TCP/IP Network 
Administration by Craig Hunt, published by O’Reilly & Associates. 
er 
There are four layers in the TCP/IP protocol architecture. Here is a brief 
explanation of each layer. Figure 2-l shows the relationship between TCP/IP and 
the OS1 model. This relationship is not a perfect fit, but is helpful if you are already 
familiar with the OS1 model. 
This is the lowest layer of the architecture, and corresponds to the lowest two layers 
of the OS1 model (Data Link and Physical). The Network Access Layer protocols 
allow the system to deliver data to orher devices in a directly attached network. 
Functions performed include encapsulation of IP datagrams into transmitted frames, 
and mapping of IP addresses to physical addresses used by the network. Protocols 
implemented in this layer include device drivers. 
Internet Protocol Layer 
The most important protocol in this layer is the Internet Protocol, the IP in TCP/IP. 
The IP is used to deliver data, as this layer provides routing functions (transferring 
information from one device to another). IP defines the datagram, the basic unit of 
transmission. It also defines the Internet addressing scheme. IP is connectionless, 
which means it does not verify that the receiving system is ready to receive data. This 
layer corresponds to the Network Layer in the OS1 model. 
2-1  
						

							Overview of TCP/IP 
TCPAP 
Application 
Host-to-Host 
Transport (TCP) 
Internet Protocol 
w 
Network 
Access OSI 
Application 
Presentation 
Session 
Transport 
Network 
Data Link 
Physical 
Figure 2-1 TCP/IP and the OSI Model 
Transport Layer 
The full name for this layer is the Host-to-Host Transport Layer. It uses two 
protocols: Transmission Control Protocol (TCP) and User Datagram Protocol 
(UDP). TCP p rovides reliable (connection is made) data delivery service with error 
detection. UDP provides connectionless datagram delivery service. Applications can 
use either protocol. 
Both protocols deliver data between the Application Layer (see below) and the 
Internet Layer (see above). TCP corresponds with the Transport Layer in the OS1 
model., 
-4” 
Application layer 
This layer includes any process using the Transport Layer (see above) protocols for 
data delivery. There are many application protocols, most of which provide user 
services. The most widely known are: 
0 telnet, the Network Terminal Protocol, which provides remote login 
over the network 
0 ftp, the File Transfer Protocol, which provides interactive file transfer 
0 
SMTP, the Simple Mail Transfer Protocol, which provides electronic 
mail 
e DNS, the Domain Name Service, which maps IP addresses to names 
assigned to network devices 
2-2  
						

							Overview of TCP/IP 
rip, the Routing Information Protocol, which is used by network devices 
to exchange routing information 
e 
NFS, the Network File System, which allows files to be shared by 
different hosts on the network. This protocol is not supported by the 
QNX 2 version of TCP/IP. 
The Application Layer corresponds to the top three layers in the OS1 model 
(Application, Presentation, and Session). 
You should be familiar with IP addresses, because they are used by the Internet 
Protocol to send datagrams from one node to another. A datagram includes a 
destination address, which is a 32-bit IP address. 
TCP/IP has 5 address classes - A, B, C, D, and E. Classes A, B, and C are available 
for government and organizational use. Classes D and E are reserved for special use. 
This section discusses classes A, B, and C. 
etwork and 
Each address class uses one part of the ?&bit address to identify the nerwork number 
and another part to identify the host number. The demarcation for host and 
network numbers for classes A, B, and C is one octet. An octet is eight bits and is 
usually represented as a decimal number between 0 and 255. A group of four octets 
comprises an IP address. An example of an IP address is 129.52.111. GO. 129, 52, 
111, and 60 are the four octets, each one representing a series of eight bits. 
Class A addresses use the first octet for network numbers and the next three for host 
numbers. Class B addresses use the first two octets for network numbers and the 
next, two for host numbers. Class C addresses use the first three octets for network 
numbers and the last octet for host numbers. Table 2-I summarizes this 
relationship. 
Table 2-1 Network and Host Numbers 
Network Numbers 
I Host Numbers 
A First octet 
Last three octets 
B First and second octets Last two octets 
C First, second, and third octets Last octet 
2-3  
						

							f i 
To install, configure and use the Unified TCP/IP option, you mat have 
VoiceMemo software version 6.OA or later installed on your server. You also 
must have the 6.OA hardware, including the 6.OA Ethernet card. Ethernet 
cards used in 5.x servers are not usable. 
This chapter contains reference and background material for the hardware 
installation and configuration. Refer to the Task List for step-by-step instructions on 
Unified TCP/IP installation and configuration. 
ardware 
In order to connect your Series 6 server to an Ethernet network, you need an 
Ethernet card. The Unified TCP/IP option includes an Ethernet card that you 
install in a server module. You need at least one Ethernet card per server, and you 
can install up to two cards per module. 
There are three kinds of cabling that can be used in Ethernet networks: thick, thin, 
and twisted pair. Thick Ethernet uses 0.4 inch diameter, 50-Ohm, double shielded 
coaxial cable, based on IEEE standard 802.3 1 OBase 5. Thin Ethernet uses 0.2 inch 
diameter, 93-Ohm coaxial cable, based on IEEE standard 802.3 1OBase 2. Twisted 
pair Ethernet uses unshielded twisted pair cable (AT&T D-inside wire or IBM Type 
3), based on IEEE standard 802.3 lOB=e-T. The cable is two sets of twisted wire 
pairs, with a gauge of 22, 24, or 26. All three standards can handle a data rate of at 
least 10 megabits per second. 
The’Ethernet card provided with the Unified TCP/IP option has a twisted-pair 
connector (an RJ-45 connector) for lOBase-T wiring. If you have thin or thick 
Ethernet cabling, you must purchase a transceiver to interface between your cabling 
and the Ethernet card’s RJ-45 connector. These transceivers are available through 
many vendors, such as MiLAN and Cabletron. 
For complete information on the Ethernet card , refer to the Technical Reference in 
the Series G Technical Reference Manual. 
3-1