Q&A

Paul Fontaine, networking technical manager from 3Com, attempts to answer any questions you might have about networking.

QUESTION

e, networking technical manager from 3Com, attempts to answer any questions you might have about networking. I am experimenting with networks. I have successfully set up using Microsoft Family as my primary logon. I have Windows 95, Windows 3.11 and NT on a peer-to-peer bus configuration, and my Windows 95 is the client-server.

What I now want is to compare, using the same setup, with Personal Netware.

I have a client netware installed and have loaded the software, but when Netware starts to run, all hell breaks loose and my log shows VLM's/IPXs not loaded. It beeps and then boots into Windows 95 with the message box 'network not loaded'. I then get my Microsoft logon box for network.

For the time being I have remmed out Net start in autoexec. Is what I am trying to do OK if I change the primary logon to Netware and presumably make sure a shell program for Netware can load?

What would be the effect on changes to file shares if I then reverted to Microsoft networking? You will understand that I am a complete novice in networking.

Andrew Clark

ANSWER

Personal Netware is an old Dos-based product from Novell, which has been obsolete for the past two to three years. The functions provided by Personal Netware (file and printer sharing) are included in Windows NT and Windows 95 as well as in Windows 3.11. As far as we know, only MS-Dos and Windows 3.x support Personal Netware.

You can use the Windows 95 built-in Netware client if you want to connect to a Novell server, or you can use the IPX, instead of Netbios, for a peer-to-peer network, which will bring you the functionality you are looking for with Personal Netware. The second advantage of this system is that it uses the 32-bit functionality of Windows 95.

If you still want to use Personal Netware, you should look into the configuration with Windows 95 and use the ODI driver for the Windows 95 network setup.

I hope this helps.

QUESTION

What does the term spoofing mean?

Dave Birkett

ANSWER

Spoofing is a unit capability which locally acknowledges the periodic control information exchanged during active sessions between client-server, server-client or from server to server. Using spoofing, a dial-up ISDN line can be physically closed down when no real user data needs to be transferred. The logical session is maintained without there being a physical ISDN connection. As soon as real user data needs to travel over the link, the ISDN connection is automatically re-established, the user hopefully unaware that the ISDN line was temporarily closed down. This feature is a must over dial-up ISDN lines because it keeps the ISDN line free from being permanently connected and open, therefore incurring call charges.

Care must be taken in the definition of real user data. In this instance, there may be a number of forms of data, specific to applications, which are viewed and treated as real user data by the units, when in fact it is control or management information used by the application. An example of this is SNMP traffic. This is not spoofed, so SNMP polling may keep ISDN calls up if the polling rates are set to a short time period. Spoofing can only be achieved for specific applications/sessions like NetWare, IPX sessions and IP connections.

QUESTION

Why should I use a firewall?

John Aldrige

ANSWER

Many medium to large sites can have dozens, even hundreds of server resources.

These can be physical services such as file servers and printer servers as well as soft services, providing access to specific applications. All these services are advertised using Novell's service advertising protocol (SAP).

SAP messages are recognised by each network router (and file server which acts as a router) and propagated throughout the network, so everyone knows what services are available and where. In addition, all routers (and file servers which act as routers) tell each other where they are on the network in relation to each other.

These messages are updated and re-issued every minute to ensure all components are kept abreast of network changes, with SAP updates offset by 30 seconds from RIP updates.

On a high-speed Lan backbone, information about a few servers is of no consequence. Information on hundreds of services sent across relatively low speed, wide-area network connections can be highly undesirable.

The impact for a router is that it has to understand and propagate this RIP/SAP information. A bridge does not look at the information in quite the same way - it just passes it. However, this information is used on both 3Com bridges and routers to derive AutoCall information.

On the edge of a network, small offices or even single users, especially in a dialled ISDN environment, may only want to access a small number of central site services. They probably do not want to see any information for any other devices, such as print servers and all other non-relevant file servers, for three main reasons.

First, a communications bandwidth is wasted in transmitting information of no value. Second, hiding such services from people who do not need to access them provides extra security. And third, memory and processor resources saved at the remote site on the bridge/router will aid performance in transmitting real and useful data.

In large IP networks, particularly those with a connection to the internet, it is likely that you will want to restrict access to your network or, at the very minimum, to specific hosts storing sensitive data on your network.

Using the IP firewalling option you can filter data by source and/or destination address to limit access to your network from remote sites.

IP networks rarely use a single IP protocol and it is possible that you do not wish all protocols to be passed across your ISDN or Wan link. The IP firewalling option allows you to accept or deny different types of IP frame.

QUESTION

Do I have to configure my PCI adaptor for my computer?

Fred Hardy

ANSWER

PCI is a self-configuring bus architecture. Most of the time you will only need to install the board in your system - PCI then does the rest.

However, on some PCI computers, mostly with combination PCI/ISA or PCI/EISA buses in the same machine, you may be required to configure the computer's Bios manually after installing your PCI adaptor.

If you need to configure your system manually, refer to the computer's owner's guide. In addition, see the hints at the end of this document.

3Com testing has shown, for example, that the AST Premmia GX P90 seems to require you to run the Eisa configuration utility once you install the board to make sure that the I/O base address of the device does not conflict with other devices.

QUESTION

Which PCI slot is best for my PCI adaptor?

Lindsey Collins

ANSWER

3Com PCI adaptors are designed to work in any PCI slot, including slave-only slots. The adaptors perform best in those slots that support bus mastering data transfers.

Refer to your owner's manual for information on which slots support bus mastering data transfers.

QUESTION

Which PCI slot is best for my PCI adaptor?

Which PCI slot(s) are bus mastering in my PCI machine?

John Gotted

ANSWER

Generally, if you have three PCI slots in a system, one slot will be designated as a slave-only slot (that is, it does not support bus mastering data transfers).

Slots are not always marked clearly to distinguish between slave-only and bus mastering slots. It is best to refer to your owner's manual or contact your computer manufacturer for this information.

Also, make sure that you have the latest version of your system's Bios.

QUESTION

Why is my 3Com 3C589 PCMCIA not seen by my laptop?

Peter Swift

ANSWER

Try using the PCMCIA card with card and socket services (CSS) enabled in your config.sys file. Many problems with installation of PCMCIA cards are resolved with the latest release of CSS from your notebook manufacturer.

After obtaining the latest CSS drivers, load them in your config.sys file and execute the diagnostics program provided on your EtherDisk to verify the functionality of the adaptor.

QUESTION

How does remote node differ from remote control?

Sean Caldwell

ANSWER

Remote control systems allow a remote user to take over a processor (CPU) on the Lan, with a software package that minimises information sent over the communications line. The processor could be the CPU of a single PC, or a remote control server which has multiple processors to support multiple simultaneous sessions.

The remote PC acts like a dumb graphics terminal, transmitting keystrokes and mouse clicks, while the dedicated processor transfers screen painting information. The actual applications - the .exe files - are being executed on the Lan PC.

Remote node products, however, provide true peer Lan access and take advantage of the remote PCs power to run the executables. Using this method of remote access, remote users interact with the Lan in the same way as they would if directly Lan-connected. With remote node, the remote user logs into the network, gets the files he/she needs, checks email, updates a data base, and logs off.

QUESTION

What are the possible bottlenecks when running remote access?

Almar Sobell

ANSWER

Assuming there are no significant Lan degradation issues, possible bottlenecks that could affect throughput occur between the remote access server and the workstation/PC. The most common possible bottlenecks could be:

remote access server (RAS)

communication port of the RAS

modem/DCE attached to the RAS

telco/cloud,

modem/DCE attached to the workstation/PC

communication port of the workstation/PC

workstation/PC processor

workstation/PC Ram (random access memory)

Lan protocol (IP, IPX, DEC)

the application

If you are a small business and think you may need a network or have problems with the one you've got, please email us with your questions to be answered on this page. Write to us at [email protected]