CA2216530A1 - Process for operating an exchange and scaleable digital switch system for carrying out the process - Google Patents

Abstract

A method of operating a switching facility is based on a data base consisting of individual data sets which can be addressed via a key term. The data base is held on a mass store with relatively high access time. To reduce the data access time, a RAM is provided in which the data sets from the data base can be stored. Also provided is a second store area in which a field of pointers to such data sets can be stored, and access occurs in such a way that a field index is derived from the key term. By requesting a pointer to a data set in the first store area, the data set is taken from either the RAM or the mass store. In the latter case, the data set once read out is duplicated in the RAM and an appropriate pointer is entered. For that purpose, a preferably digital and expandable switching system is provided for switching the PCM signals; the said switching system comprises at least one host computer (60), at least one telephone server (70) and a plurality (50) of elementary telephone switching contacts (100, 101...) which are interconnected via a high-speed communications network (80).

Description

- CA 02216~30 1997-09-26 PROCESS FOR OPERATING AN EXCHANGE AND SC~ RT~ DIGITAL
SWITCH SYSTEM FOR CARRYING OUT THE PROCESS

The invention relates to processes according to precharacterizing clause of Claim 1; in particular, it relates to an exchange consisting of a switch and a host and also to a process for operating said exchange.
An exchange has to perform various tasks. These include call setup and call routing, which are also referred to as call management; furthermore, the so-called call services, which include call charging among other things. For this purpose, a data record in which the start and end times of the call, date information, the telephone numbers involved and the costs of the call are entered and are permanently stored is created, for example, for each newly setup call, that is to say when a connected subscriber himself calls or is called.
Finally, it is also necessary to mention the operator services, which account for the more organizational part of the system, ~or example setup and clearance of customers, configuration of circuits or the charge services based on the call service data.
Until a few years ago, the telecommunication switches were provided with "intelligence" so that the total handling of a call could be performed in the switch itself. Progr~mming was carried out essentially in a so-called assembler with non-standardized operating systems which had to be specially developed for each switch. A great deal of time was required for the development and modifications were very expensive.
Outgoing services over and above the elementary call management could of course scarcely be implemented, if at all.
An important improvement arises from the use of a standardized operating system, such as, for example, interactive UNIX of a higher programming language, such CA 02216~30 1997-09-26 = as, for example, "C", and of a standardized, very rugged database system, such as, for example, "Informix". These software modules are available for all hardware platforms from the simplest PC to so-called mainframes, so that the software adaptatlon to hosts o~ different efficiencies can be carried out in a simple manner.
Since, however, the tasks to be performed by the host are very varied and furthermore must be executed for the most part in real time, creation of the software can be extremely complex. Moreover, where the sizes o~ the switch differ, the host has to meet very different performance requirements. In particular, hard disk accesses during the database operations impose a very considerable burden on the host.
Furthermore, in general different operating systems and database programs are required for different performance ratings, with the result that software creation and adaptatlon to different numbers of subscribers continues to be very expenslve. Several hosts which execute the various tasks essentially parallel are therefore frequently networked with one another and moreover the operations executed in real tlme are reduced to a minlmum.
The problem of hard disk access for extractlon of the data from the typically indexed databases r~mAin~. Therefore a flrst object of the invention is to provide a simpllfied and/or faster access to the data, and this is achieved by the characterlzlng features of Clalm 1.
Relevant advantageous embodlments are described ln Claims 2 to 4.
As mentioned above, conventional exchanges comprise one or more hosts. In addition there are the actual swltch matrix and one or more telephone servers.
Because of the virtual absence of modularity, thls architecture can be expanded only at great financial CA 022l6~30 l997-09-26 cost and with a great deal of work. Modular architectures typically have small "sub-exchanges", formed by elementary telephone switches which are connected, by way of a bus or by way of a ring structure, to each other and to the hosts and the telephone servers. Thls architecture is modular and therefore can also be more easily expanded or adapted to varylng requirements. The primary problem lies on the one hand ln the lack of reliability in bus-based systems and generally the capacity limit, which is quickly rea,ched when the system is expanded, in respect of the transmission medium, whether it is a bus or a ring.
Therefore the second object of the invention is to provide an architecture for an exchange, which permits virtually any expandability with an inexpensive modular structure, with enhanced reliability of the overall system.
This is achieved by the features of Claim 5.
The use of the elementary telephone switches (ETS) affords the possibility of building very small installations up to installations of almost unlimited size, with a suitable high speed communication network.
The advantage of a solution according to Claim 6, in which the communication network is formed by at least one ATM-switch lies in the enormously high processible data rate (about 150 Mbps) at each port of the ATM-switch, whereby it is possible to achieve an enormous gain in speed in terms of the throughput of data through the entire installation, and the installation can be expanded virtually without limit.
The problem when using an ATM-switch in exchanges is that the PCM (pulse code modulated) data rates are not sufficiently high with small elementary telephone switches (ETS) and therefore, without further measures, it is only meaningful for large elementary telephone swltches (ETS) to be connected together by CA 022l6~30 l997-09-26 way of an ATM-switch. This results in a rather confused design configuration in which expansion is possible only ln very large steps in terms of capacity, which lS contrary to the desired modular structure.
Data concentrators (which are known to be used for reducing channels in data transmission) in conjunction with ATM-switches were not used hitherto, probably for reasons of the level o~ complexity which is difficult to control As mentioned above, these disadvantages are ellminated ,by the features of Claim 6 In accordance with the lnventlon, a plurality of elementary telephone switches (ETS: small, simple, modular) are not directly connected to ATM-ports but by way of concentrators; in this way the ATM-switch is no longer used as a switch in the true sense but is only used as a link between a plurality of partial switches formed by elementary telephone switches (ETS) and down-circuit concentrators In this way an ATM-switch, as a connecting element, can advantageously display its enormously high data rate By using an ATM-switch as a link and not as a switch, it is possible to achieve a substantially higher data transmisslon rate than hitherto, Gbytes per second being thought to be achievable This combines the two above-mentioned aspects of the invention, in that not only is an expandable circuit obtained but also the deslred faster operation is permitted To this extent, the embodlment protected in Claim 6 is an independent invention, even separately from the other features of Claim 5, although the modular structure according to Claim 5 is thus also obtained in a particularly advantageous manner. In any case, however, it is also possible to offer fundamentally novel transmission services which were possible with the conventional circuits only to a very limited extent, without involving a change in the architecture.

CA 02216~30 1997-09-26 A further advantageous embodiment of the system according to the invention is described in Claim 7.
Further details of the inventlon are evident from the following description of embodiments shown schematically in the drawing.
Fig. 1 shows an exemplary arrangement for illustrating the process according to the invention;
Fig. 2 shows the procedure for call handling by the host;
Fig. 3 and 4 show the access to a first type of data;
Fig. 5 and 6 show the access to a second type of data;
Fig. 7 is a block diagram of a conventional exchange having a bus structure;
Fig. 8 is a block dlagram of a conventional exchange having a ring structure;
Fiq. 9 is a block diagram of a scaleable digital switch system (SDSS) according to the invention;
Fig 10 is a block diagram of an expanded SDSS;
Fig. 11 shows a first example of the structure of an elementary telephone switch (ETS); and Fig 12shows an alternative structure of an ETS
In Fig. 1, a switch (1) is connected to other exchanges of a telecommunication network via groups of incoming lines (2) and outgoing lines (3).
Furthermore, subscribers (8) are connected via lines (7) A host (4) having at least one mass storage unit (6) controls the switch (1) via a data link (5).
The switch (1) - controlled by the host (4) executes the actual switching of the incoming and outgoing calls. For example, a commercial unit "Excel Switch PCX512" may be used as switch (1), its internal structure and the internal function being known per se and therefore not being discussed in detail here.
On the basis of a program contained in a memory segment of the host or of an external program, the host (4) performs - under program control - essentially the CA 022l6~30 l997-09-26 above-mentioned tasks of call management, call services and operator services.
Call management relates malnly to call setup and call routing. A trunk group for the outgoing half of the call and, within the trunk group, a channel are selected in accordance with various routing tables.
The call services include call setup and call routing, which - as explained above - are re~erred to as call management, as well as the so-called call services which include mainly call charging. For this purpose, fo~ example, a data record in which the start and end times of the call, date information, the telephone numbers involved and the costs of the call are entered and are permanently stored is created for each newly set up call, that is to say when a connected subscriber himself calls or is called The data required for calculating the costs are taken from a customer database as data record "ANI" which identifies the customer and also contains a reference "DNIS" in time-specific and weekday-specific rate tables.
The operator services include, inter alia, customer management, which supports the setting up and clearance of customers, switch management for configuration of the switch, call monitor for real-time representation of the internal state of the switch and a billing servlce The call data records stored in the billing period are evaluated for each customer, the call charges incurred are summed and finally a statement is prepared.
Fiq. 2 shows, in simplified form, the procedure for handling a call by the host. In Fig. 2, the handling of a call is shown schematically in the form of a block diagram;
2a denotes entry of customer data;
2b denotes entry of the area code, exchange of data;
and 2c denotes further handling of the calls.

CA 022l6~30 l997-09-26 , 70910P3073 7 The data assigned to the customers are taken from the ANI database (20), the subscriber number typically serving as a key. This necessltates several hard disk accesses, each of which requires several milliseconds in a typical PC system. The database itself may comprise, for example, 150000 entries.
Furthermore, the routing information required for controlling the switch (1) (cf. Fig. 1) and billing the calls must be taken from a further database, the NPANXX database (21). Several hard disk accesses are once again, required for this purpose. The NPANXX
database typically comprises 60000 entries.
Since at least 3 or 4 hard disk accesses, each with an access time of about 10 milliseconds, are usually requlred per database access, PCs as hosts reach their per~ormance limit already at a low level of traffic; disk caching reduces the access time for the individual data record to mllliseconds, but this does not solve the problem.
A remedy for the time-consuming hard disk access is the holding o~ the two databases in the memory of the host. Access to a data record ln the memory is thus faster by a factor of about 1000 to 10000 than access to a ~ata record on the hard disk.
As a result of this improvement in the access time, in particular even higher-value service features, such as the servlces call-back and ready-rlng explalned below, are permltted wlth a single host instead of a network of hosts connected to one another. In the case of the call-back service, a subscriber calls the swltchboard, l.e. practlcally, the swltch (1) with a specified number, allows ringing to take place and hangs up the receiver again; the exchange determines who has called and calls the relevant subscriber back and gives him a dial tone. The subscriber dials the number o~ hls desired call partner and the switchboard routes the call.

CA 02216~30 1997-09-26 The service ready-ring, that is to say periodic polling of a subscriber line in order to determine whether the subscriber has lifted the receiver in order then to send him a dial tone, also requires particularly frequent accesses to the database.
In the so-called booting process, a first storage area for the NPANXX data records used in the swltchboard and a second storage area for the customer-speclfic ANI data records are set up in the memory of the host. At the beginning of operation, these two storage areas are still unfilled but have already been created in a size sufficient to be able to hold the total database in each case.
Alternatively or in a further embodiment of the invention - particularly if the databases are so large that they cannot be completely held in the memory - the memory areas may be created in such a way that they can hold only the most frequently used data records, it being possible to use a special LRU algorithm to create space in the event of overfilling or for selecting external memories When, after the beginning of operation, the calls arrive from the various subscrlbers, the informatlon relevant to the users and the lnformation on the selected numbers are searched for in the corresponding databases on the disk, read out and stored in the pre-reserved memory areas The strategy for access to the data involves first searching for the data in the memory and reloading the required data from the disk only if said data are not yet present in the memory. In this way, the disk accesses are minimized because, after a certain time, most data are already in the memory and can be read out from there. New data records must be reloaded from the disk only when new users call or new destinations are called.
Access to the NPANXX data is described more exactly below, with reference to Fig. 3 and 4.

CA 02216~30 1997-09-26 The database NPANXX contains location information, coordinate information and distance information of a telephone call. For this purpose, the possible destinations of a call are NPANXX-encoded in a 6-digit number. Out of the 1 million possible data records, however, only about 61000 are in fact occupied. It would therefore not be reasonable to reserve 1 million data records; instead, it is sufficient in the booting process to reserve space for the 61000 possible data records (30) in the memory. At the same time, a field (31) for 1 million pointers is set up, said polnters being capable of pointing to the data records A further pointer freeptr (32) points in each case to the first free pre-reserved data record (33) in the memory.
If an NPANXX data record havinq a certain NPANXX
number is now requested (41), in a first step (42) a search is carried out in the pointer field VNHPTR[NPANXX] under the index NPANXX to determine whether there is already an entry there which points to a data record in the memory. If so, the data record can be read out from the memory under the pointer address (43) The access to the data record in this case is permitted by only two memory accesses without a single access to the hard disk. This process is then complete (4a) If no entry exists in the pointer field VNHPTR
under the index NPANXX, the relevant data record must be read out from the hard disk (44) and is entered in a free memory area reserved for data records in the memory (45) The address of this newly entered data record is en~ered in the index field VNHPTR under the index NPANXX (46), with the result that the next access to the relevant data record can likewise take place in the memory. This process is then complete (4b).
This type of memory organization may be denoted by hashing, this hashing function, which generates from CA 022l6~30 l997-09-26 the automatic number ldentification an index under which the data record can be ~ound in a manner known per se, being the identificatlon function. The key value itself gives the index under which a pointer to the data record can be found.
This access process is not possible for the user data from the ANI database since the automatic number identification ANI is formed by a 10-digit number. The access to the ANI data is therefore described more exactly below with reference to Fig. 5 and 6.
Here, according to the invention, the hash value is formed by the last four digits of the automatic number identification (= ANI number). Accordingly, a pointer field hptr (51) havlng 9999 elements is set up, and each element may polnt to one of the pre-reserved ANI data records (50). A further pointer afreeptr (52) points to the first free pre-reserved ANI data record (53) in the memory.
As a result of the stated method of formation of the hash value, multiple assignments of one and the same hash value by several automatic number identifications ANI do of course occur According to the invention, the resulting conflicts are solved by virtue of the fact that every data record in the memory is extended by a concatenation field (54) which - unless empty - points to further data records which have the same hash value but di~ferent automatic number identifications If any ANI data record having a certain ANI
number is now requested, the hash value hval is first formed (60) and a test is carried out to determine whether an entry exists under the index hval in the pointer field hptr (61) If the entry is bLank, the data record must in any case be read from the hard disk (62) and is entered in the first free reserved storage area (63).

~ CA 02216~30 1997-09-26 If the pointer fiela for the relevant index was free (64), the address of the newly stored data record can be entered directly in the pointer field hptr under the index hval (65); otherwise, the chain of all data records concatenated with the entry hptr[hval] as the root must be rolled up (66) so that the address of the newly stored data record can be finally entered in the link field of the last data record of the chain, which in fact must be empty (67).
If testing of the pointer field hptr [hval] (61) reveals tha~ an address entry already existed, this is temporarily stored. This address entry addresses a data record whose ANI field is tested for conEormity in the next step. If the ANI entry of the data record agrees with the required ~NI number, the correct data record is ~ound and can be read out (72) I:~ not, a data record havlng the correct hash value but a false ANI number has been found If the link fleld of this data record is blank (70), the required data record must be read from the hard disk, which process has already been described (62-67).
If the link field of the data record investigated is not blank, the new data record addressed by the link field is investigated (71), which process has also already been described (69-72).
6a, 6b and 6c represent the termination of the particular process.
Figure 7 is a view showing the principle of an exchange with elementary telephone switches (ETS): a plurality of ETS (11, 12 ...) which each represent small, relatively independent exchanges and which have PCM multiplex lines (for example T1 trunks) (2) connected to their inputs are connected by way o~ a central bus (3). Also connected to the bus (3) are a plurality of telephone servers (21, 22, ...) for carrying out the switching activity which goes beyond CA 02216~30 1997-09-26 an ETS, and a host (4). The host (4) not only permits operation and maintenance of the installation but also billing of the telephone calls An exchange having the architecture shown in Figure 7 is modular in the sense that the elementary telephone swltches (ETS) and the telephone servers represent individual modules, and therefore it can only be easily expanded. It does however suffer from two major disadvantages: the central data exchange element of the installation is formed by the bus (30). If the bus should fail due to a fault, the entire installation becomes inoperative.
I~ the installation is to be expanded by ha~ing additional ETS and telephone server modules added thereto, finally the bus will represent a hard limit in terms of expansion, both mechanically (because of the restricted number of plug-in locations), and electrically (because of the restricted driver outputs) and also from the point of view of data transmLssion speed.
The indicated disadvantages are alleviated by an architecture as shown in Figure 8. Here, the elementary telephone switches (ETS) (31, 32, ...), the telephone servers (41, 42, ...) and the host (40) are connected together by way of a data ring (300). As the ring is preferably in the form of a double ring, corresponding to the prior art, failure of a ring sector does not immediateIy result in total failure of the entire installation. In this case also however expandability is limited by the ring, by virtue of the limited throughput of data.
Figure 9 shows a block diagram of a scaleable digital switch system (SDSS) according to the invention. The SDSS is formed from a quantity (50) of elementary telephone switches (ETS) (100, 101, . . . ), a host (60) and a quantity (70) of communication systems (80) which comprise concentrators (140, 141) and at least one ATM-switch (160) CA 022l6~30 l997-09-26 Each o~ the elementary telephone switches (ETS) (100, 101, ...) is connected to a respective group (110, 111 . .) of PCM MPX lines (120, 121, ...).
Typically each elementary telephone switch (ETS) is connected to 20 T1-MPX lines corresponding to a capacity in each case o~ 20 x 24 = 480 PCM connections The multlplex connections (130, 131, ...) of optionally five such elementary telephone swltches (ETS) ln each case are connected to a respective concentrator (140, 141) which is connected by its multiplex connection (150, 151) ~o a data connection of an ATM-switch (160).
Connected to further connections of the ATM-switch are one or more hosts (60) and a plurality of telephone servers (70, 71, . ).
Each of the elementary telephone switches (ETS) can however support a limited number of ports and has an ATM interface which can be integrated with other ETS
by using the concentrator and the ATM-switch. The concentrator (140, 141) is responsible for demultiplexing the incoming bandwidth from each ETS
into the ATM-switch. The ATM-switch routes all the incoming data streams into outgoing streams. The total numbers of ports to be provided can be determined by the matrix size of the ATM-switch, which can be easily upgraded without any hardware changes in the rest of the system.
Provided in the basic extension stage shown in Figure 9 are 2 concentrators (140, 141) with a total of 10 elementary telephone switches (ETS) and a 2x2 ATM-switch, which gives a switching capacity of 4800 ports.Division of the switching assembly into elementary telephone switches (ETS) on the one hand and the high-speed communication system with concentrator and ATM-switch on the other hand makes it possible for the ETS
to be kept relatively small and thus flexible, while nonetheless affording the ATM the required high data rate (about 150 Mbit/s) at the data connections In CA 02216~30 1997-09-26 , 70910P3073 14 this respect the ATM-switch itself is not used as a swltch in the true sense but as a llnk for high-speed connection of the data streams which are produced in the ETS and which are assembled in the concentrators.
The partlcularly hlgh level o~ flexibility and scalability of the architecture shown in Figure 9, for very different requirements, ls described below Fig.
10 shows the expansion of the architecture described in Fig. 9 to switch twice the number of connectlons:
Connected to a 4x4 ATM-switch (200) are a total of ~our concentrators (201 .204) each wlth five elementary telephone switches (ETS) 210, 211, .. 229) The further modules connected to the ATM-switch, such as host and telephone servers (cf. Fig 1) have been omitted from this figure ~or the purposes of simplification thereof. The scalability of the architecture is clearly evident. Similar ETS, only doubled in terms of their number, are connected to an ATM switch of double the capacity, by way of also twice as many similar concentrators, and afford an exchange with a capacity which is doubled overall ~in this case for a total of 9600 ports).
The scalability o~ the arrangement is made further clear by the fact that, to achieve a still substantially higher number of ports, a plurality of ATM-switches can be connected to form an ATM-network ln a manner known per se It is thought that ln that way installations with up to 1 million ports are possible The elementary telephone switch (ETS) is a non-blocking switch with only a small to minimum capacity,which can handle approximately 480 ports (20 T1 lines).
The basic approach in the design of the ETS is open modular architecture, such that the users can easily change the configuration by installing more m~dules in the switch. As a result, the switch can be upgraded to include telephone services other than only the basic CA 02216~30 1997-09-26 services, for example by adding ISDN interfaces or data communication services etc.
Figures 11 and 12 show the internal configuration of an ETS. Flgure 11 shows an embodiment given by way of example with VME-bus modules, while Fig 12 shows a PC-based design with corresponding plug-in boards.
The elementary telephone switch (ETS) in Figure 11 comprises ~i~e module types each module being preferably formed by one or more printed circuit boards), na~ely a main control module (MCM) (301), a switching matrix module (SMM) (302), a T1 interface module (TIM) (303), a digital signal processor interface module (DIM) (304) and an ATM concentrator interface module (ACIM) (305). In the ETS every unit except the main control module (MCM) is connected to two standard interface units: the standard VME-bus (307) and the telephony path unit (308). The standard VME-bus permits the use of so-called shelf technology, gives a certain flexibility to the system design, reduces the costs and increases the reliability since additional modules can easily be added to the basic system. The ETS exchanges information to control its modules and the telephone path unit (308) via the VME-bus. Data exchange itself occurs by means of a memorymat I/O (input/output) with dual port RAM, that is to say by means of shared memory. In comparison with alternative methods such as, for example, an HDLC
connection (a high-level data link control or an ISO-standardlzed blt-oriented data transmission protocol), this procedure affords substantial advantages in terms of speed.
The telephony path unit (308) consists of conventional PCM multiplex lines and is used for switching voice and data with a digitized PCM format.
The maximum bandwidth of the telephony path is CA 022l6~30 l997-09-26 determined by the maximum number o~ non-blocking ports of an ETS
The maln control module (MCM) (301) of the ETS
must be able to control the ETS and maintain its operation, such as routing, controlling the telephony path and generating statistics. Therefore the MCM has a powerful micro-controller, for example based on an MC68302 whlch lS specially designed for communication applications; MC68302-based VME boards are available off the shelf. To simplify control and data management in the ETS; the MCM can be equipped with a data interface (for example an SCSI interface) for the connection of mass storage devices such as, for example, a hard disk.
The SMM (302) provides a system-wide interconnection of digital telecommunication data (i e.
both volce and data) over the PCM multlplex llnes) (308). If the maxlmum switching slze of one S~ lS 512 x 512, an ETS can handle up to 512 non-blocking ports.
That means that an ETS is able to support the connection of up to 512 trunk lines at the same time.
The main component of the SMM is a switching matrix device such as, for example, Mitel MT8980 The T1 interface module (TIM) (303) interfaces T1 trunk digital data to the telecommunication processing system's digital data streams. For thls purpose, one or more standardized T1 PCM multiplex lines (309) with the data rate of, for example, 1.55 Mbps can be connected to the module (303). For example, there are four T1 lines per circuit board, while the functions of signalling, monitoring and supervlsion of the connected PCM lines are also carried out ln the T1 lnterface module (TIM), in addltion to signal interfacing. Each elementary telephone swltch (ETS) is provided with a number of T1 interface modules (TIM) (303), which corresponds to the number of T1 lines (309) to be supplied If an ETS is to be able to - CA 02216~30 1997-09-26 switch telephone lines with particular properties, in order to offer for example ISDN services, the T1 interface modules (TIM) of the ETS have to have only the corresponding features.
The DSP interface module (DIM) (304) preferably includes one or more digital slgnal processors DSP;
here on the one hand multi-frequency tones are produced and detected for dealing with certain classes of calls;
on the other hand, many different announcements can be stored in a read only memory (ROM) and fed into the system by way of the DSP. Finally, produced in the DIM
are all tone signals which are used within the system, (continuous tone, cadence tone and shot tone), whose parameters are also read out of a read only memory and evaluated by the DSP
Finally the ETS also has an ATM-concentrator interface module (ACIM) (305); it serves to pass on the data in the individual ETS, in order then to have the higher switchlng tasks performed by the ATM-switch which is used as a link. For this purpose the ACIM has a PCM-data connecting line (310) for connection to a concentrator.
Figure 12 shows an alternative configuration of an elementary telephone switch (ETS) as a PC-based system. The main control module (MCM) (351) is, for example, formed by a main board of a PC, provided with a 486 or Pentium processor, to which the other modules of the ETS are connected by one of the available PC
buses (356), such as ISA or EISA. Some commercial T1 interface boards (353) which can interface to a PC
system with ISA or EISA and MVIP bus already include a switching matrix device (354). The use of this type of board makes the SMM unnecessary. The DSP interface module (DIM) (355) is formed by a DSP insert board which ls available for PC applications. In thls design of the ETS there is also an ATM concentrator interface module (ACIM) (357). The ETS can be connected to a concentrator by way of a PCM data connection (360) extending from the ACIM. Transportatlon of the PCM
data between the individual modules of the ETS is effected by way of PCM multiplex lines (358) which are preferably formed by a MVIP bus.

Claims (7)

1. Process for operating an exchange with a database which consists of individual data records which are addressable via an automatic number identification and which is created on a mass storage having a relatively long access time, characterized in that a sufficiently large first, dynamic storage area for storing data records from the database is set up in an RAM, that furthermore a second storage area for storing a field of pointers to such data records is set up and that access to a data record comprises the following process steps:

1: derivation of a field index from the automatic number identification;

2: inquiry as to whether, in the pointer field of the second storage area under the index derived according to step 1, a pointer to a data record has been entered in the first storage area of the RAM;

3a: if an entry exists: read out of the required data record from the first storage area of the RAM with addressing by the pointer entered in the pointer field, with the result that the access procedure is successfully completed;

3b: if no entry exists in the pointer field: read out of the data record with the aid of the automatic number identification from the mass storage;

4: duplication of the data record read out from the mass storage in the first storage area in a free data record area, without overwriting or deleting a data record area which is already available;

5: entry of a pointer to the data record just duplicated in the pointer field at the index derived in step 1; consequently, any subsequent access to the same data record is possible only through RAM access.

2. Process according to Claim 1, characterized in that the automatic number identification is an integer and the field index and the automatic number identification are identical.

3. Process according to Claim 1, characterized in that the automatic number identification is an integer and that the field index is derived from the automatic number identification in such a way that different automatic number identifications may lead to the same field index.

4. Process according to Claim 3, characterized in that the data records stored in the first storage area are extended by a pointer field compared with those in the mass storage, and steps 1 and 2 are unchanged and steps 3 to 5 are changed as follows:

3.1a: if an entry exists: checking of the data record addressed by the pointer entered in the pointer field to determine whether it is actually the data record matching the automatic number identification;

4.1a: if the automatic number identification corresponds and hence the required data record has already been read out, the access procedure is successfully completed;

4.1b: if the automatic number identification does not correspond and an entry exists in the pointer field of the same data record: use of this entry as a new pointer and continuation of the process with step 2;

4.1c: if the automatic number identification does not correspond and no entry exists in the pointer field of the data record: read out of the data record with the aid of the automatic number identification from the mass storage and duplication in the first storage area in a free data record area; the pointer field of this data record remains without an entry, with the result that the data record is characterized as the last data record belonging to the relevant index;

5.1: entry of a pointer to the data record just duplicated in the pointer field of the data record processed in step 4.1; consequently, any subsequent access to the same data record is possible only through the RAM access;

3.1b: if no entry exists in the pointer field: read out of the data record with the aid of the automatic number identification from the mass storage;

whereupon the process is continued with steps 4 and 5 and the pointer field of the data record remains without an entry and hence the data record is characterized as the last data record belonging to the relevant index.

5. Scaleable digital switch system for the switching of PCM signals, in particular for carrying out the process according to any of the preceding Claims, characterized by at least one host (60), at least one telephone server (70) and a plurality (50) of elementary telephone switches (ETS) (100, 101, ..) which are connected together by way of a high-speed communication network (80).

6. Scaleable digital switch system according to Claim 5, characterized in that the high-speed communication network (80) includes one or more ATM
boardes (160) serving as a link, with at least two up-circuit concentrators (150, 151).

7. Scaleable digital switch system according to Claim 5 or 6, with elementary telephone switches (ETS) formed from individual modules, characterized in that at least two modules of an ETS are provided with a shared memory or a memory having at least two connections and data exchange occurs between these modules by way of the shared memory.