GB2340699A - Automatic network protocol configuration in a data processing system - Google Patents

Abstract

A data processing network comprises a plurality of client systems and a controlling system which transmits "broadcast" data packets which may be received by any of the client systems attached to the network. The client system, on initial connection to the network, analyses the transmitted "broadcast" data packets and determines the network protocols in use on the network. The client system then configures software located at the client system so as to use one of the network protocols in use on the network. Furthermore, the client system customises the operating parameters it uses to communicate with the network so as to optimise the performance of communication with the network. Additionally, the client systems may, optionally, analyse transmitted non-broadcast data packets.

Description

2340699 AUTOMATIC NETWORK PROTOCOL CONFIGURATION IN A DATA PROCESSING

NETWORK

Technical Field of the Invention

The present invention relates to automatic network protocol configuration for data processing systems, in particular for client computer systems on a network where the client computer is required to use a variety of different network protocols when contacting different servers.

Background of the Invention

In a typical data processing network environment, multiple client computer systems (clients) are connected to one or more server computer systems (servers). In one common arrangement, each client system includes an operating system, and optionally other software, stored on a mass storage device such as a hard disk drive within the client. On power-up or reboot, the client boots from the hard disk drive without reference to the server computer. Once operational and connected to the network, the client may choose to access and use programs or data resident on mass storage in the server.

The growth of client/server computing has led to an increase in the number of different application programmes executing on a client and a corresponding increase in the number of different types of server systems with which the client must communicate. Although the client will typically only be physically connected to just one network (for example, an Ethernet or Token-Ring LAN), the client will be required to use a variety of different network protocols when contacting the different servers (for example, NetBEUI, IPX, TCP/IP, SNA and the like).

In general, it is the user's responsibility to install the correct network protocols for the required applications, the application installation program does not do this. Additionally, since some applications support a variety of protocols, it is impossible for the application installation program or documentation to know which protocol is correct for the user's environment - it depends solely on how the servers in the user's environment are configured. This makes network setup error-prone.

2 If a client is configured without a required network protocol, it will be unable to communicate with certain servers even though they are connected to the same LAN. On the other hand, if a client has an unnecessary protocol installed, the client will suffer degraded performance and unnecessary network traffic will be generated. This is because application programs running on the client (for example, Microsoft windows Workstation Network Client, Lotus Notes and the like) will attempt to use all the available network protocols when trying to contact the server even though the server will only respond to one.

windows is a trademark of Microsoft Corp. and Lotus Notes is a trademark of Lotus Development Corporation.

It would be desirable to have a technique for installing the correct network protocols for applications in order to ensure that clients are able to communicate with the servers with which they desired to communicate and make network setup less error-prone. It would also be desirable to have a technique which does not install unnecessary network protocols which may degrade client performance and generate unnecessary network traffic.

Disclosure of the Invention

Accordingly, the invention provides a data processing network comprising a plurality of client systems and a controlling system operable to transmit "broadcast" data packets which may be received by any of the plurality of client systems, wherein each of the client systems, on initial connection to the network, is operable to analyse transmitted "broadcast" data packets and to determine the network protocols in use on the network therefrom. Through the analysis of broadcast packets, a client system which is newly connected to a data processing network is able to receive a huge amount of information even though no other client computer or controlling system on the network is aware of its presence. Analysis of this information provides the advantage that the correct network protocols for applications are installed in order to ensure that clients are able to communicate with the servers with which they desired to communicate. Network setup is also made less error-prone. End users do not like having to do network setup and find it difficult to do. Additionally, unnecessary network protocols which may degrade client performance and generate unnecessary network traffic are not installed. The response time of applications is speeded up because it is ensured that the wrong protocols are not tried 3 when contacting a controlling system. All of this is achieved without any modifications to the server or the network itself, merely requiring an application program to be installed and run on the client system.

Preferably, the controlling system is a server system. Since a server is the first system to be installed on a network and is constantly active once it is installed, then it is advantageously used as a controlling system. One or more servers may be operable in the data processing network and each of them may perform the role of a controlling system. A console system, if optionally used, will not generally perform the role of a controlling system, since it is less likely to be constantly active once installed.

In a preferred embodiment, the transmitted "broadcast" data packets are analysed for a predetermined time. It is likely that all or most of the protocols in use in the data processing network will have been found in a short, predetermined period of time and so limiting the analysis step to a predetermined time has the advantage that the installation of the client computer on the network is completed more quickly.

In a preferred embodiment, each of the client systems configures software located at the client system so as to use one of the network protocols in use on the network. The software will then be correctly configured to use a protocol which is available on the network with user intervention.

In a further preferred embodiment, the client systems customises the operating parameters it uses to communicate with the network so as to optimise the performance of communication with the network.

In an variation of the preferred embodiment each of the client systems, on connection to the network, is operable to analyse transmitted non-broadcast data packets. By analysing non-broadcast data packets, the can find such useful information as the default router address, the subnet mask, the DNS (Domain Name System) address, the web proxy address, the address of Lotus Notes servers and the like.

The invention further provides a method for use in a data processing network, the data processing network comprising a plurality of client systems and a controlling system, the method comprising the steps of: transmission, by the controlling system, of "broadcastn data packets 4 which may be received by any of the plurality of client systems; reception, by each of the client systems, on initial connection to the network, of said "broadcastn data packets; analysis, by each of said client systems, on initial connection to the network, of the transmitted "broadcast" data packets; and determination, by each of the client systems, on initial connection to the network, of the network protocols in use on the network from the analysis of the transmitted "broadcast,, data packets.

The invention yet further provides a computer program product for use in a data processing network, the data processing network comprising a plurality of client systems and a controlling system, the computer program product comprising: a computer usable medium having computer readable program code means embodied in said medium for determining the network protocols in use on the network from analysis of transmitted "broadcast" data packets, said computer program product having: computer readable program code means for transmission, by the controlling system, of "broadcast" data packets which may be received by any of the plurality of client systems; computer readable program code means for reception, by each of the client systems, on initial connection to the network, of said "broadcast" data packets; computer readable program code means for analysis, by each of said client systems, on initial connection to the network, of the transmitted "broadcastn data packets; and computer readable program code means for determination, by each of the client systems, on initial connection to the network, of the network protocols in use on the network from the analysis of the transmitted,broadcast" data packets.

The invention yet further provides a client system for use in a data processing network having a controlling system operable to transmit "broadcast" data packets which may be received by the client system, wherein the client system, on initial connection to the network, is operable to analyse transmitted "broadcast" data packets and to determine the network protocols in use on the network therefrom.

Brief Description of the Drawings

A preferred embodiment of the present invention will now be described, by way of example only, with reference to the accompanying drawings in which:

Figure 1 is a schematic representation of a computer network in which the preferred embodiment of the present invention is practised; Figure 2 is a block diagram representation of a client computer system connected to a server computer system in the network of Figure 1; Figure 3 is a block diagram representation of the basic functions of a typical network adapter card, such as that shown in figure 2; Figure 4 is a block diagram representation of the software interfaces present in a typical network adapter card, such as that of figure 3; Figure 5 is a flow chart showing an embodiment of the present invention; Figure 6 is a representation of protocol layers in the OSI model; and Figure 7 is a representation of headers being added as data moves down the layers of the OSI model.

Detailed Description of the-Invention

Referring firstly to Figure 1, there is shown, in schematic form, a local area network (LAN) 10 in which a preferred embodiment of the invention is implemented. The network of Figure I may be configured as an Ethernet or Token Ring local area network or other arrangement. The network comprises a server computer system 20 (which in the present embodiment may be an IBM PC Server 330 computer system) connected for communication by a network 26, in a loop configuration, with a plurality of client computer systems 30, 32, 34, 36. The client computer systems may be personal computers based on the Intel X86 family of microprocessors or other forms of computer system. Each client system includes a LAN adapter card or network interface card (NIC) 38, 40, 42t 44 to provide communication with the server computer over network 26.

Optionally, the network includes one or more further server systems 22 and a console computer system 24 through which the network administrator may control the network. The console system is generally similar to a client system except that it includes mass storage and 6 further is arranged to provide the network administrator with certain privileges, not available to a client user, which allows him/her to control the network. The use of a console computer system is optional.

Control of resources on the network, including communication between server and clients, is effected by means of a network operating system (NOS), for example, OS/2 LAN Server from IBM Corporation, having a server' component which executes on the main processor (s) of the server computer system and a corresponding 'requester' component which executes on the main processor of each client computer system. other suitable network operating systems include Netware from Novell Inc and OS/2 WARP server from IBM.

Figure 2 is a simplified block diagram showing the connection of server computer system 20 to client system 30 over network 26. As is normal, the client system, constituted in the present embodiment by a personal computer, includes a keyboard 131 and a display 132 operating under the control of control logic in the form of main CPU 133 which is connected by a system bus to system memory (RAM) 134 and non-volatile memory (ROM) 135, in which is stored system BIOS including POST code.

The client system further includes a network adapter card 38 which, in the present embodiment, may be either an ethernet or token-ring adapter card. This adapter card includes non-volatile memory in the form of ROM in which is stored code employed in providing communication between the client and server. The client system further includes mass storage 138 in the form of a magnetic disk drive (hard file) for storing operating system and other software. The client may further include (not shown) a diskette drive, CD-ROM drive or similar.

The server computer system of Figure 2 includes a keyboard 121 attached to a system unit 122 including a main CPU 123, system RAM 124, system ROM 125 and mass storage capability 126, typically in the form of multiple magnetic disk drives constituted in a RAID (redundant array of independent disks) arrangement. Stored on the disk drives are software images comprising operating system software and optionally application software for downloading to the client systems. The server system may optionally include a display 127 (if the network administrator requires direct interaction with the server system) and other storage devices such as a diskette drive (not shown) and CD-ROM drive 129. Communication over the network 26 is provided by input/output logic 128 which may take the 7 f orm. of an adapter card (which may be the same type adapter card employed in the client system).

The present embodiment of the invention provides a means by which the client system 30 can automatically determine which network protocols are required to enable it to communicate with servers on the network and disable any unnecessary ones. This prevents unnecessary traffic, reduces the CPU utilisation required to monitor the network traffic and speeds up the response time of applications by ensuring that the wrong protocols are not tried when contacting a server system 20. It also prevents calls to the company's IS help desk (or, in the case of a dial-up user, a call to the ISPIs help desk) when a required protocol has been omitted causing an application to fail.

In the present embodiment, a client computer 30 connected to a network 26 not only receives those data packets which are specifically addressed to it but also any packets which are broadcast, on the network 26.

Data packets addresses are typically 12 hex characters (6 bytes) in length. Data packets may be lunicast, packets which contain the address of one computer in the network and are only received by that computer.

The first bit of a packet address being set to 1011 indicates that the packet is a unicast packet. The address 08:00:5A:12:34:56 is a unicast address, the packet only being addressed to the computer having this unique address. Data packets may be Imulticast, packets which contain a group address and are received by all computers in the network which are part of that group. The first bit of a packet address being set to "111 indicates that the packet is a multicast packet, the packet being received by all computers in the group addressed by the multicast packet.

The address 18:00:88:88:88:88 is a multicast address, the packet being addressed to any computer in the network which is part of the group 88:88:88:88. Data packets may be broadcast, packets which are received by all computers in the network. The first bit of the packet address being set to a I'll, and a predetermined "special" group, being a broadcast group, indicates that the packet is a broadcast packet and is to be received by all computers. The special group typically has an address of FF:FF:FF:FF, so an address of FF:FF:FF:FF:FF:FF corresponds to a broadcast packet.

8 Broadcasts are very common on a network because they are used by high level protocols to discover the low level network address of a target computer in order to communicate with that computer on a network.

For instance, the TCP/IP network protocol broadcasts an ARP (address resolution protocol) request in order to discover the MAC (medium access control) address of a computer and the NetBIOS network protocol multicasts a FIND NAME request to discover a MAC address. A MAC address is a 12 digit address which is burnt into a ROM on the Network Interface Card (NIC) by the NIC manufacturer. Each MAC address is unique. Address Resolution Protocol is responsible for obtaining hardware addresses of TCP/IP hosts on broadcast-based networks.

A typical Address Resolution Protocol, used with TCP/IP, consists of the steps of:

(i) Any time a source computer wishes to communicate with a destination computer, an ARP request is initiated. This ARP request will contain the IP address of the destination computer, but NOT the MAC address.

(ii) The source computer will first check its own ARP cache to see if it contains the MAC address of the destination computer.

(iii) if the ARP cache does not contain the MAC address of the destination computer, the source computer will send a broadcast on to the network (either local or remote), each computer will then receive the broadcast packet (identified because the destination address is set as all 1,s, as described above for a broadcast packet), and compare its own IP address with that of the IP address contained in the broadcast packet.

(iv) The computer whose IP address matches the one contained in the broadcast packet will respond to the source computer with the details of its MAC address. The source computer stores this information in its ARP cache. in future communication the source computer uses the destination computer's address, rather than a broadcast address. Computers whose IP address does not match the one contained in the broadcast packet discard the packet and do not reply.

Because of this use of broadcast packets, a client computer 30 which is newly connected to a network 26 is able to receive a huge amount of information even though no other computer on the network 26 is aware of its presence.

9 Figure 3 shows a block diagram of a typical adapter card 38. The adapter card includes a bus interface 302 for connection to a bus to which the CPU 133, RAM 134, ROM 135 and MSS 138 are also connected. The adapter card also includes a memory 304 in which is stored the MAC address. In the typical card shown in figure 3, there are three types of connection to the network 26 possible using twisted pair cable, Coaxial cable or an Attachment Unit Interface connection.

The program of the present invention interfaces directly with the network adapter card device driver (using the NDIS3 or equivalent API) and receives these broadcast packets. Figure 4 shows the software interfaces present in a typical network adapter card. The NDIS3 (Network Driver Interface Standard 3.0) software driver 402 communicates with the Network Interface Card 38 in order to transfer data to and from the network 26. The drivers for the various protocols, such as DLC, NetBEUI, TCP/IP and IPX/SPX, shown as 404 to 410 in figure 4, communicate with the NDIS3 driver 402 so that they can transfer data to and from the network 26. user mode applications communicate with file systems, which in turn communicate with the protocol drivers in order to transfer data to other devices on the network. As mentioned above, the program of the present invention interfaces directly with the network adapter device driver to receive broadcast packets and so does not operate through the protocol drivers. It receives packets which are destined for any of the protocol drivers, thereby enabling it to determine what protocols are in use on the network and what protocol drivers need to be installed.

Figure 5 shows a flow chart of an embodiment of the present invention. The flow chart is entered at step 500. At step 502, a broadcast packet transmitted on the network is read by the client computer 30. At step 504, the packet is analysed to determine which protocol is contained within it (since each successive protocol layer is encapsulated within a lower layer protocol).

Referring to figure 6, the various layers of a typical protocol stack implementing the OSI model are shown. The network interface card 38 provides layer 1, the NDIS3 diver 402 forms part of layer 2 and the protocols such as DLC, NetBEUT, TCP/IP and IPX/SPX, shown as 404 to 410 in figure 4, form layer 3 as well as parts of layers 2 and 4. Figure 7 shows how each of the layers adds a header to the data it receives as it passes that data down the protocol stack. At the physical layer the Data has a Data link Header, a Network Header, a Transport Header, a Session Header, a Presentation Header, an Application Header and a Data Trailer, added by the Data Link layer to aid in synchronisation.

Packets which are transmitted in a communications network conform to a well defined set of frame and structure rules. It is therefore possible to read the bits within a packet and break them up into their constituent parts. For example, a packet is typically made up of an address field, control field, data field and error control field. The beginning and ending bit sequences of the packet are also well defined.

Once the packet has been received by a client it can break the packet up and compare the bit sequences with a database or a Look Up Table of known or defined bit sequences, allowing it to determine the byte (s) which will identify the protocol in use.

For example, a Token Ring packet consists of 3 bytes preamble, 6 byte destination address, 6 byte source address, 18 bytes routing information, 1 byte DSAP, 1 byte SSAP, 1 byte control then N bytes of data. The DSAP field (Destination Service Access Point) tells the token ring driver which protocol stack to send the data to - FO means NETBEUI, AA means TCP/IP, EO means IPX and so on. By using this well known format, the present invention allows the determination of which protocol is contained in the packet. Having decided that a Token Ring packet contains, for example, TCP/IP, a known offset within the data portion of the Token Ring packet can be examined to determine which higher level TCP/IP protocol (HTTP, FTPf TELNET) is contained in the data portion.

Referring again to figure 5, at step 506 it is determined whether the protocol detected at step 504 has been detected before or whether the protocol detected at step 504 is a previously undetected protocol. If the protocol has been detected before, then the program proceeds to step 510. otherwise, the newly detected protocol is entered into a list of detected protocols at step 508. A check is made at step 510 as to how long the client computer 30 has been listening to broadcast requests in order to identify protocols. If this time does not exceed a pre determined value, then the program returns to step 502. If this time exceeds a pre-determined value, typically one minute, then the operating system network protocols are modified at step 512, the operating system is configured to enable the protocols at step 514 and the client computer is re-booted at step 516. The program finishes at step 518.

If the program is run for the pre-determined time (say one minute) when the client computer 30 is first installed, it is likely to discover all of the protocols supported by the servers 20 on the network 26 (since the servers 30 are installed first and are constantly active) These steps all take place without requiring user intervention.

This analysis of broadcast packets allows the program to determine which protocols are in use on the network. As an example, if the program detects a packet intended for protocol FO then it would know the NetBEUI protocol was in use on the network. The program then automatically modifies the operating system's network configuration to support the protocols which it has detected and removes any protocols which are not required.

once the correct protocols have been selected, techniques such as DHCP (on TCP/IP), Browse Server broadcasts (on NetBEUI/Lan Server) and Server Advertising Protocol (on Novell) are used to further customise the network setup, again without requiring user intervention. Servers use broadcast packets to advertise data about themselves and any other relevant information, such as routing information between other computers on the network. For example, if a Novell Netware server is broadcasting using SAP (Service Advertising Protocol), it is possible to determine important configuration information, such as Server type (file server, job server, gateway, print server, bridge server and the like), server Name, Network address, Socket address and the like.

In a variation of the present embodiment, the network adapter 38 uses a technique called 'promiscuous copy,. This is a function used by network sniffers, which allows them to receive packets addressed to other computers and is supported on most Ethernet adapters and newer Token-Ring adapters (although the user is usually restricted from using it for security reasons). Using this technique, the program receives non-broadcast packets and analyses them to find such useful information as the default router address, the subnet mask, the DNS (Domain Name system) address, the web proxy address, the address of Lotus Notes servers and the like.

I I 12

Claims (23)

1. A data processing network comprising a plurality of client systems and a controlling system operable to transmit "broadcast" data packets which may be received by any of the plurality of client systems, wherein at least one of the client systems, on initial connection to the network, is operable to analyse transmitted "broadcast" data packets and to determine the network protocols in use on the network therefrom.

2. A data processing network as claimed in claim 1 wherein the controlling system is a server system.

3. A data processing network as claimed in claim 1 wherein at least one of the client systems, on initial connection to the network, is operable to analyse the transmitted "broadcast" data packets for a predetermined time.

4. A data processing network as claimed in claim 1 wherein at least one of the client systems configures software located at the client system so as to use one of the network protocols in use on the network.

5. A data processing network as claimed in claim 4 wherein at least one of the client systems customises the operating parameters it uses to communicate with the network so as to optimise the performance of communication with the network.

6. A data processing system as claimed in claim 1 wherein at least one of the client systems, on connection to the network, is operable to analyse transmitted non-broadcast data packets.

7. A method for use in a data processing network, the data processing network comprising a plurality of client systems and a controlling system, the method comprising the steps of:

transmission, by the controlling system, of "broadcast" data packets which may be received by any of the plurality of client systems; reception, by at least one of the client systems, on initial connection to the network, of said "broadcastn data packets; 13 analysis, by at least one of said client systems, on initial connection to the network, of the transmitted "broadcas t n data packets; and determination, by at least one of the client systems, on initial connection to the network, of the network protocols in use on the network from the analysis of the transmitted "broadcastn data packets.

8. A method as claimed in claim 7 wherein the controlling system is a server system.

9. A method as claimed in claim 7 wherein the analysis step is performed for a predetermined time.

10. A method as claimed in claim 7 further comprising the step of the client system configuring software located at the client system so as to use one of the network protocols in use on the network.

11. A method as claimed in claim 10 further comprising the step of the client system customising the operating parameters it uses to communicate with the network so as to optimise the performance of communication with the network.

12. A method as claimed in claim 7 wherein at least one of the client systems, on connection to the network, analyses transmitted non-broadcast data packets.

13. A computer program product for use in a data processing network, the data processing network comprising a plurality of client systems and a controlling system, the computer program product comprising:

a computer usable medium having computer readable program code means embodied in said medium for determining the network protocols in use on the network from analysis of transmitted "broadcastn data packets, said computer program product having:

computer readable program code means for transmission, by the controlling system, of "broadcast" data packets which may be received by any of the plurality of client systems; 14 computer readable program code means for reception, by at least one of the client systems, on initial connection to the network, of said,, broadcast" data packets; computer readable program code means for analysis, by at least one of said client systems, on initial connection to the network, of the transmitted "broadcast" data packets; and computer readable program code means for determination, by at least one of the client systems, on initial connection to the network, of the network protocols in use on the network from the analysis of the transmitted "broadcast,, data packets.

14. A computer program product as claimed in claim 13 wherein the controlling system is a server system.

15. A computer program product as claimed in claim 13 wherein the means for analysis is performed for a predetermined time.

16. A computer program product as claimed in claim 13 further comprising computer readable program code means for configuring software located at the client system so as to use one of the network protocols in use on the network. 25

17. A computer program product as claimed in claim 16 further comprising computer readable program code means for customising the operating parameters it uses to communicate with the network so as to optimise the performance of communication with the network. 30

18. A computer program product as claimed in claim 13 wherein the means for analysis, analyses transmitted non-broadcast data packets.

19. A client system for use in a data processing network having a controlling system operable to transmit "broadcast" data packets which 35 may be received by the client system, wherein the client system, an initial connection to the network, is operable to analyse transmitted "broadcast" data packets and to determine the network protocols in use on the network therefrom.

20. A client system as claimed in claim 19 wherein the client system, on initial connection to the network, is operable to analyse the transmitted "broadcast" data packets for a predetermined time.

21. A client system as claimed in claim 19 wherein the client system configures software located at the client system so as to use one of the network protocols in use on the network.

22. A client system as claimed in claim 21 wherein the client system 10 customises the operating parameters it uses to communicate with the network so as to optimise the performance of communication with the network.

23. A client system as claimed in claim 19 wherein the client system, 15 on connection to the network, is operable to analyse transmitted nonbroadcast data packets.