ES 101. Module 3Domain Name System (DNS)

Last Lecture

• Routing and IP addressing

This Lecture

• Domain Name System

Domain Name System History

• DNS provides host IP address to mnemonic name mapping

• Originally this was accomplished by a set of configuration files on each client and server– /etc/hosts– /etc/networks

• These files at first were manually configured by the System Administrator using a text editor– Highly subject to typographical errors

DNS History (cont’d)

• In the early days of the Internet, the Network Information Center (NIC) maintained a single etc/hosts file, which could be downloaded for host configuration– HOSTS.TXT

– ftp’d by all hosts (RFC 952 and 953)

• The bandwidth consumed by this process is proportional to the square of the number of hosts– Outgoing ftp load on the NIC host was very high

– Made worse by the astronomical number of hosts on the Internet

DNS History (cont’d)

• This situation caused the development of a network-wide, distributed database system which did this host name to IP address mapping– Originally known as Berkeley Internet Domain Server,

or BIND– Later renamed to Domain Name System (DNS)

• We will study the host configuration files to gain insight into the operation of the DNS– Host files are still used for small private networks on

both Unix and Microsoft networks

Internet Hosts vs. Time

Date No. Hosts1984 1,0001989 100,0001992 > 1.0 Million1994 3.2 MillionToday ????

Internet Names

• With millions of Hosts on the Internet, how do you easily specify the host that you want to interact with?– IP addresses are very hard to memorize for humans

– DNS was developed to provide an easy to remember addressing scheme

• Originally specified in RFCs 1032, 1033, 1034, and 1035

Checkout the following web site for access to all RFCs:http://www.cis.ohio-state.edu/cs/Services/rfc

/etc/hosts File

• Every computer on a TCP/IP network has an IP address, canonical host name, and zero or more host name aliases

• The /etc/hosts file is the classical method for mapping IP addresses to host names

• Format of the hosts file consists of one IP address per line– Note that the 127.0.0.0 address is the local loopback IP

address

Host File Example

/etc/networks File

• Networks and subnets are also named• The handling of network address to network name

mapping is accomplished by the /etc/networks file

Network File Example

Introduction to DNS

• DNS is essentially a hierarchical, distributed local hosts database

• The DNS name provides an easily recognized structure

• DNS can be conceptually divided into three parts– Domain name space

– Name servers

– Resolvers

• Each of these will be discussed in the following charts

Domain Name Space

• DNS is partitioned into different “Domains”

• It is organized hierarchically

• Top level domain names are globally specified

• Second-level domain names are unique to each organization, and must be registered with via a “registrar”– Domain names are not associated with a location

– You can search for a unique name using the “whois” utility of the registrar

– When you find no match, the name is available

– You can then register the name with a registrar to reserve it

Top Level Domain Names

Fully Qualified Domain Names

• The highest level of the hierarchical structure (root) is “dot”

• The proper syntax for a fully qualified domain name includes the trailing period– Standard practice is not to include it, since it is

understood

Domain Name Hierarchy

Second Level Domain Names

• These names are registered by qualified registrars– www.networksolutions.com originally

– Many other registrars are now available, for example:• Domain Direct (http://www.domaindirect.com/)

• Each domain is given its own authority to create subdomains– Example: http://eecs.vuse.vanderbilt.edu

• A (sub)domain may contain an entire network, or only parts of a network

Domain Physical Representation

Domain Name Tree

• The easiest representation of the DNS name is as a “tree”

• Each domain represents a “branch” of the tree• The “leafs” of the tree are individual host

machines

DNS Servers

• The DNS database is distributed among a number of servers– These servers perform the DNS name to IP address mapping

– They are referred to as name servers

• Each domain is responsible for maintaining a minimum of one primary and secondary DNS servers– This function can be subcontracted to others

– No single DNS server could hold the database for the entire Internet

DNS Servers (cont’d)

• DNS servers that are in different locations operate cooperatively to resolve the name to IP address mapping request

• Distributed databases have the advantage that they are efficient

• If the names that are accessed most frequently are in a local cache (DNS server), the DNS lookup traffic across the Internet is greatly reduced

DNS Servers (cont’d)

• Distributed databases are also more reliable than a single machine– One failure cannot “take down” the network

DNS Forwarding

• If you do not wish to host your own DNS server, you can pay for a service known as “DNS Forwarding”– Typically provided by the registrar

• What happens is that the registrar maintains the redundant DNS servers

• The URL of interest is locally controlled by the registrar• A table entry is made into the registrar’s DNS database to go

to the IP address of a server that is specified by you– This is done by a “control panel”

• Name resolution requests can then be referred to any IP address on the Internet– Even your dorm PC, assuming that it has a static IP address

Static IP Addresses

• You may request a “static” IP address from your ISP

• However, this is becoming more difficult, since ISPs like to manage their own IP address space– This is because IP addresses are in short supply

• When this is done, your PC requests an IP address from a “Dynamic Host Configuration Protocol” (DHCP) server– Your machine is then assigned an IP address

• This address “ages” and is changed periodically– However, Vanderbilt does not change this often

Dynamic Address Determination

• How would you determine your DHCP assigned IP address?– “ipconfig” command– “What’s my IP” web site

• http://www.whatismyip.com/

• If your IP address does not change often, you can DNS forward to it– If the IP changes, you simply log onto the registrar’s

control panel and repoint the URL to the new IP address

Assignment

• Those of you willing to spend ~$30/year should register a domain name of your choosing

Web Servers

• Note that the DNS only points to an IP address• A server that hosts many web sites, must still

resolve the URL to the proper “index.htm” file for the web site of interest

• This is done by a local DNS server process that only points to the internal folders– No updates from the Internet– This information is not distributed out to the Internet

• An example is the Helios server here at Vanderbilt

DNS Name Resolution

• Each client executes a DNS client software process known as a “(DNS) name resolver”

• This software communicates with the DNS Server established in the client by the network administrator

• Name servers are arranged in a tree structure that corresponds to the naming hierarchy

• Each lower-level server knows what the higher-level servers are by their IP address

DNS Name Server Registration

• An organization registers for a second-level domain name

• It is required to provide and maintain a two name servers for that sub-domain– These servers are known as the “Authority Name

Servers” for the domain

• The name server must be linked into the DNS server tree

If you use an ISP, you can use their DNS servers.

Name Server Modes

• Depending on the client request, the name server operates in one of two modes

• When the resolver queries a name server, the message contains the following information– Name to be resolved

– Class code of the name• Protocol group to be used, ie. IP, IPX, etc.

• For IP, the class code is IN

– Type of response desired, ie. IP address

– Action code that specifies whether the name server should translate the name completely

Object Types

• DNS servers can be used for multiple functions– Translation of a host name to an IP address

– Look up a mail server address

• This information is differentiated by Object Types– Also known as resource record types

Common DNS Resource Record Types

DNS Query Processing

• When a domain name server receives a query from a client, it checks to see if the name is within its subdomain– If so, it sends back the information to the resolver

• If the name server cannot resolve the name completely, it checks the “action code” to see what the client specified– Recursive resolution (complete lookup)

– Iterative resolution (non-recursive lookup)

Recursive Resolution Schematic

Name server asks for assistance from other higher level name servers

Iterative Resolution

• Name server generates an error if it cannot resolve the name

• Client is informed of the name server that it should query next

Non-Recursive Resolution Schematic

Name Caching

• Name resolution is performed from left to right– Local authority up to higher levels

– Most efficient path for name resolution

• Each name server also keeps a cache of recently resolved names and the mapping information– This information “ages off” the name server

• It is up to the clientsoftware to decide if the returned information is good enough

Name Resolution Sequence

• Client queries the name server

• Name server first checks for local resolution

• Next checks its name cache to see if the name was recently resolved

• If found, the name server reports back to the resolver

• If not, the name server follows the “action code” provided– Recursive resolution

– Non-recursive resolution

Next Lecture

• Basic Network Troubleshooting and Utilities