At its simplest a router provides a link or route between 2 or more network segments.
When you setup a dial connection for internet access vi an ISP you are in effect setting up a very simple router. It recognises to network segments, your local network with one computer your PC, and the ISP's network with everything else.
When setting up a multi-site corporate network however each site will need to have a route configured for every other site. While there is only a few sites it is practical to work out the list of routes and enter it on each router manually.
But as the number of sites increases this rapidly grows into a time consuming and error prone task. To simplify this task you want routers that speak a common routing protocol and which will exchange information on which networks they know automatically.
The most commonly used routing protocol in small to medium sized networks, say up to 100 sites is RIP.
For larger networks there are other protocols that may also be appropriate. These include OSPF, EGP, and BGP.
This is usually an issue that comes up with pay as you use connections. In 199x this would be PSTN and a modem, in 200x, most likely ISDN, and currently most likely to come up with Mobile Data SIMs.
Most users have switched to flat rate, always on, broadband connections. Misbehaving applications, just slow thing up a bit, rather than resulting in a very large bill.
If installing a router that is to use an ISDN link, it is worth asking the Telecom supplier for weekly billing at leased for the first couple of months.
This may be usefully supplemented by monitoring the activity of the router.
A router makes an ISDN call when it receives a packet that it thinks is for a remote router that it can best reach via an ISDN call. Note the call will in most cases be cleared after a short idle period. typically around 20 seconds.
Note in finantial terms the worst case is when a program has a short conversation with a remote server, waits a fraction longer than the idle timeout and then starts another conversation.
Frequent causes.
Many mail clients can be set to poll for new mail as frequently as every 30 seconds.
Especially if there was any data pending on the connection. Server will keep retrying to deliver the data, tipicaly ever 30 minutes for around 24 hours.
If some of the clients on the local network have private addresses, or hostnames that are not resolvable on the internet. Make sure that your local DNS server is athorative for the, private network address ranges. {10.x.x.x, 172,[16-31].x.x, 192.168.x.x}
According to the standards for Email, if delivery fails a server should wait, and retry after a reasonable interval typically 30 minutes. It delivery still fails, the retry interval should be progressively increased to around once per day. With the sender being notified that there is a problem. Most UNIX based server will finally give up after a week.
I have in the past run in to a case where Exchange may continually retry delivery to non-existent addresses every minute:-(
NFS in common with many programs derived from the Berkley Distribution makes use of the statd process.
This logs information on remote hosts based on the host name they provided. If the remote client is a PC this is derived from the Computer Name field {Right click on network neighbourhood, select properties, click on the identification tab}.
The snag is that this field is not validated in any way:-(
On the NFS server, run
ls -lb /etc/sm /etc/sm.bak
If this shows any names not resolvable using the host command, you
need to find the offending PC and fix.
Outlook has the ability to send an automatic reply back to the sender of an Email saying I am out of the office until next....
The snag is that unlike the UNIX vacation program which tracks senders and only notifies them once, Outlook sends it for every message including server bounce and delivery notifications.
So if you configure delivery notification, then configure auto reply, and then send a message just be for you leave....................... notification request,
There are lots of companies that make routers, and to make life fun they keep taking each other over.
A useful range of small to medium sized routers.
E2PROM is used to store multiple configs and software updates.
Configs may be saved across the network using TFTP.
Based in Germany, Bintec make a range of routers from a small single ISDN-2 unit, through to a large unit aimed at the major corporate market.
Cisco were an early player in the router game. They concentrated on mid to top range routers used by ISP for inter ISP links.
A large company formed by a lot of mergers.
| Feature | Nautica Marlin | Cisco | Bianca |
|---|---|---|---|
| Model | 3200 | 2503 | brick X21 |
| SNMP | Y | Y | Y |
| DHCP | N | ? | Y |
| NAT | N | ? | Y |
| TFTP config | N | Y | Y |
| Throughput ping | N/A | 91.5% | 82.9% |
| with compression | 735.5% | N/A | 82.9% |
| Interfaces Available | |||
| ISDN-2 | Y | Y | Y |
| ISDN-30 | ? | ? | ? |
| X21 | Y | Y | Y |
| Link protocols supported | |||
| Proprietary | Y | Y | Y |
| PPP | Y | Y | Y |
| Frame relay | Y | ? | Y |
| Compression modes | |||
| Propriatory | Y | Y | N |
| MPPC | N | ? | Y |
| V42bis | N | ? | Y |
| STAC | N | ? | Y |
| MS-STAC | N | ? | Y |
| Encryption | |||
| MPPE | N | ? | Y |
| Telephony services | |||
| TAPI server | N | ? | Y |
| CAPI server | N | ? | Y |
| Diagnostics | |||
| HTML server | Y | ||
| Telnet server | Y | ||
| Telnet client | Y | Y | Y |
| Ping | Y | Y | Y |
| ipxping | Y | ||
| traceroute | Y | ||
| netstat | Y | ||
| Throughput | |||
packet size x number of packets returned x 8 x 100
-------------------------------------------------- %
time taken x nominal link bandwidth
For Example A Marlin running over 64 kbps link.
Packet size :- 1024 bytes
Packet returned :- 5831
Time taken :- 101.48s
1024 x 5831 x 8 x 100
--------------------- = 735.5%
101.48 x 64000
Speed usually in the range 64kbps to 2Mbps.
This is an on-demand network. Each member of the network has a number {phone number} which uniquely Identifies there connection. When setting up a connection, the calling party specifies, the remote parties number, the type of data to be transferred, and the type of channel {bandwidth} required.
The network then either grants or refuses the request. Where the request is granted the subscriber has the use of the requested bandwidth until the call is closed.
| Chanel | Description |
|---|---|
| B | 64K |
| D | Control data |
| H | 384K {6xB} |
| J | 1920K {5xH} |
Not all telco's provide this service