The IP part of TCP/IP is the low-level switched packet protocol standard that allows the internet to exist. At it's essence, IP is a standard for how a group of bytes is formatted so that it can be delivered across a network to a destination. It is this network addressing that you must deal with when you setup a TCP/IP network.
A computer system on a TCP/IP network has one or more IP addresses associated with it. For workstations, there are TCP/IP solutions that will assign these automatically using protocols like DHCP (dynamic host configuration protocol) and BOOTP. These auto configuration protocols require a configuration server to "rent" IP addresses to workstations. Server systems usually need to have IP addresses assigned to them manually. When you install PicLan-IP (and most other TCP/IP products), you will need to assign one or more IP addresses to the server computer system.
If you are connecting your network to the public internet, you cannot make up your own addresses but must instead use addresses that are assigned to you by the internet's governing body. You get these addresses assigned to you from the company that you get internet connectivity from (your ISP or Internet Service Provider).
The problem with routers is that they must decide which network packets need to move between the networks. IP uses a set of conventions to make this job easier for routers called sub-nets. A sub-net is a grouping of IP addresses that are all located on a single part of a connected network. The first example with IP addresses from 22.214.171.124 to 126.96.36.199 demonstrates one of the most common classes of sub-nets called a "class-C" sub-net. The term class-C is somewhat old-fashioned and dates back to earlier, less conjested days for the internet. There used to be three type of sub-nets:
The internet itself has become very crowded and is literally running out of addresses. For this reason, it has been years since a class-A network has been assigned, and several early class-A assignments have been taken back from users. One fix for this is to use network sizes that are between the various classes. For example, a class-C network is also sometimes called a /24 network because the network part of the address uses 24 bits. You can get a /26 network which has enough room for 64 IP address (62 of which are usable). The smallest possible network is a /30 with 4 IP addresses (2 of which are usable).
Sub-net masks for class-A, class-B, and class-C networks are quite simple. The sub-net mask for a /26 network would be 255.255.255.192 (the 192 is 128+64 or the two highest bits in the low byte). Figuring out network numbers with non-traditional sub-net masks generally requires that you convert the numbers to binary first as decimal bit-wise AND operations are not generally obvious to carbon-based lifeforms (although the silicon type seem to have not problem with this).
For a simple single-segment network, the router table is basically empty. All packets are sent directly to their destination address. When a network has routers, the table gets larger with an entry that describes a set of destination addresses and the local IP address that the packet should be sent so that it can reach the destination. A routing table is typically expressed as a list. If an IP packet is to be transmitted, the table is searched starting at the top until an entry matches the destination IP address of where the packet is to be sent. If your example network was connected with a router to another network, a routing table might look like:
Address Mask Gateway Address 188.8.131.52 255.255.255.0 0.0.0.0 0.0.0.0 0.0.0.0 184.108.40.206What this table tells the IP network driver is that any destination on the current 220.127.116.11 network should be sent directly to the destination and everything else should be sent to the router at 18.104.22.168. The use of the concept "send everything else to an address" is called a "default gateway". Most TCP/IP networks only have a single "way out" so the default gateway takes care of addressing.
If you have a TCP/IP network with only a single interconnection to the outside world (whether the public internet or just more of your company's local network), then you use the default gateway to specify the IP address of the router that will send the packets on to the rest of the network.
So what if the environment is more complicated. For example, a router also needs a routing table to decide where packets will be sent. If this router is a CISCO 7500 at Mea-West (the Internet's busiest west-coast interconnection hub), the routing table literally describes all of the fragmented sub-nets for the western half of the U.S. Suffice it to say that the routing table has tens of thousands of entries and a human did not type them in. Also suffice it to say that this is beyond the scope of this primer and is hopefully someone else's job.
http://22.214.171.124/work.htmand this will connect you with the web server at address 126.96.36.199. If you are building a purely local or test network, using IP addresses directly is quite workable. If you are connecting to the public internet, you will probably want a public internet style name such as:
http://cnn.comThe "cnn.com" is a host name for a computer system (well in this case seven different computer systems that have duplicated content and load-balance). Internet names are based on a hierarchial naming system. The highest part of the name is "com" which stands for "US commercial". Common top-level domains are:
When you have a DNS setup, you can create names for computer systems with IP addresses within the domain. This is how www.microsoft.com is setup. The name www.microsoft.com is the name for an IP address (actually 14 addresses) at Microsoft that runs their web servers. You can create as many names within your domain as you wish to point to computer systems on your network plus you can let the domain itself point to a computer system. You can also create domains within domains with addresses such as:
xyx.apweb.piclan.comOne of the more interesting things about domain names is that you can create names for computer systems that you do not own. If a system has an IP address, you can assign it a domain name (within your domain) without even asking the system's owner. Again, a domain name is just a lookup proceedure that converts a name into an IP address.
So how do you setup DNS systems. For many sites the answer is that you
don't. Setting up your own primary and secondary DNS systems requires two
computers running Unix or NT and a bit of setup. Many sites will let their
ISP (Internet Service Provider) run the DNS for them