JP2007148531A - Pc cluster system, server, address assignment method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To avoid deterioration of performance caused by execution of parallel jobs on a PC extending between switches while reducing manual cost by permitting connection of an individual PC constituting a PC cluster to an optional switch independently from its physical position and saving the labor hour in establishment of the PC cluster. <P>SOLUTION: The system constitutes a sub-network containing only one switch to which PCs are connected, and forms a partial physical address/logic address list registering all PCs connected to the switch. Partial physical address/logic address lists for all switches are formed while changing the structure of the sub-network. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a technology for constructing a PC cluster that performs parallel computation, and in particular, a list in which physical addresses and logical addresses of PCs (Personal Computers) accommodated in one switch in an arbitrary hierarchy are associated with each other, as necessary. In response, a list in which a PC logical address and a host name accommodated in one switch of an arbitrary hierarchy are automatically associated and integrated to obtain a logical address assigned to the PC or The present invention relates to a PC cluster construction technique that makes it possible to uniquely identify one switch in which a PC is accommodated from a host name.

  In recent years, with the improvement of computer performance, computer simulation has come into practical use for automobile design and the like. In response to such demand for computers, many large-scale PC clusters have been constructed, and construction of large-scale PC clusters has become common. In the construction and operation of a large-scale PC cluster, manpower according to the scale is required at the time of construction, and manpower according to the operation period is necessary for dealing with failures, and the reduction of the human cost is a major issue.

  When a failure occurs on a specific PC, it is possible to identify which PC is operating strangely by a logical address or host name. For example, when communication with a specified host name does not function properly on a particular PC, it can be identified that a failure has occurred on that PC.

  Therefore, in the maintenance management of a PC cluster, it is generally one problem to know which PC a logical address or host name physically corresponds to. This is self-evident when considering the need to take measures such as physically replacing the failed node.

  Most simply, the physical arrangement of PCs is regarded as physical addresses, and logical addresses and host names are assigned corresponding to the physical addresses. As a result, when a failure occurs, the physical address is uniquely determined according to the logical address or host name where the failure occurred, and the failure occurs in the first PC in the physical array of the PC. In this way, the faulty PC can be physically identified. This PC cluster configuration method makes it easy to identify the PC where the failure occurred, but it requires the setting of logical addresses and host names according to the physical sequence, and requires a lot of manpower for the construction. In particular, the construction of a large-scale system is very expensive.

  On the other hand, means for sending a command by specifying a logical address or a physical address corresponding to a host name using a special device and lighting the LED of the device is also known as a known technique. A person who performs maintenance management can physically identify the PC in which the failure has occurred and use it to replace the PC. In the configuration method of the PC cluster using such a device, it is not necessary to consider the physical arrangement of the PCs. Therefore, it is possible to reduce the construction cost, which is a problem in the former method.

  Application data handled on a PC cluster has a certain finite parallelism, and it is known that it is difficult to shorten the calculation time according to the number of PCs even if trying to calculate in parallel using a certain number of PCs or more. It has been. For this reason, it is a common practice to divide a large-scale PC cluster into a plurality of partial PC clusters and simultaneously execute calculations (jobs) using different data for each. If such an operation is performed, parallel computations performed in each partial PC cluster can be executed in a state where the efficiency is not lowered so much, so a plurality of jobs to be executed simultaneously are sequentially and one by one. It is possible to execute in a short time in total, compared to using all PCs in parallel.

  In order to cope with such a divided operation, it is an important issue to achieve high performance to adopt an appropriate network hierarchy. If you try to connect many PCs, it will not be physically accommodated in one switch, so a certain number of PCs will be accommodated in one switch, and these switches will be connected in a hierarchical configuration. There are many cases. In the case of a hierarchical structure, it is common to make the throughput of the upper network higher than the throughput of the lower network, but from the cost perspective, the upper network throughput is lower than the lower network throughput. It is difficult to make it equal to the sum of

  In such a case, the performance of PCs that perform parallel computation in the same switch is higher than those that are accommodated across two or more switches and connected by a higher-level network. Can be demonstrated. This is because in the latter case, the network on the upper side becomes a bottleneck, waiting for communication may occur and performance may be reduced. Such a decrease in performance becomes significant when the total number of PCs accommodated in one switch in the throughput of data exchanged by individual PCs for performing parallel computation exceeds the throughput of the upper network. .

  In order to avoid performance degradation due to such inconvenience of allocation of PCs, the job scheduler for parallel calculation identifies the switch in which each PC is accommodated, and only the number of PCs accommodated in the same switch is necessary for parallel computation At the same stage, it has a function to execute the parallel calculation using only PCs accommodated in the same switch. In order to make effective use of this function, it is necessary to give an attribute as to which switch is accommodated to a logical PC. In order to make it compatible, it is necessary to perform manual work that has become a problem when configuring a PC cluster. For example, the physical arrangement of PCs is regarded as physical addresses, and logical addresses and host names are assigned corresponding to the physical addresses. If such a setting is made, it is possible to identify what number to what number PC is accommodated in the same switch in the physical array of PCs, and which switch accommodates a specific PC. Attribute can be given. However, as pointed out as a problem in PC cluster construction, such work is very complicated when there are many PCs that make up the PC cluster, and not only can settings be mistaken because of manual work. There was a problem that it took a long time to build a PC cluster.

  Therefore, the problem to be solved by the present invention is to allow the individual PCs constituting the PC cluster to be connected to an arbitrary switch regardless of its physical position, and to save time and labor when building the PC cluster. The object of the present invention is to provide a PC cluster construction technology capable of avoiding the performance degradation caused by executing parallel jobs on PCs across switches while reducing labor costs.

The first invention for solving the above-described problems is
A PC cluster system having a plurality of terminals,
Grouping means for grouping the plurality of terminals into at least one group;
Logical address providing means for collecting physical addresses of the terminals, and for each collected physical address, a logical address including identification information that can uniquely identify a group to which the terminal having the physical address belongs;
It has physical address / logical address list generation means for associating the physical address with a logical address assigned to the physical address to generate a physical address / logical address list.

According to a second invention for solving the above-mentioned problem, in the first invention,
It has a logical address / host name list generating means for generating a logical address / host name list by associating a logical address with a host name.

According to a third invention for solving the above-described problem, in the first or second invention,
The grouping means is configured to block communication between groups.

A fourth invention for solving the above-described problem is
A server,
A logical address that collects the physical addresses of the terminals that make up the PC cluster and gives each collected physical address a logical address that includes identification information that can uniquely identify the group to which the terminal having the physical address belongs. Granting means;
It has physical address / logical address list generation means for associating the physical address with a logical address assigned to the physical address to generate a physical address / logical address list.

A fifth invention for solving the above-described problem is
An address assignment method,
A step of grouping a plurality of terminals into at least one group;
A setting step for setting only terminals and servers belonging to one group to communicate with each other among the groups;
A logic that collects physical addresses of terminals that can communicate based on the setting, and assigns a logical address including identification information that can uniquely identify a group to which the terminal having the physical address belongs to each of the collected physical addresses. An address assignment step;
A physical address / logical address list generating step for generating a physical address / logical address list for each group by associating the physical address with a logical address given to the physical address;
An integration step of integrating the physical address / logical address list generated for each group into one.

A sixth invention for solving the above-described problem is the above-mentioned fifth invention,
It has a logical address / host name list generation step for generating a logical address / host name list by associating a logical address with a host name.

A seventh invention for solving the above-mentioned problems is the fifth or sixth invention,
The grouping step includes a blocking step for blocking communication between groups.

An eighth invention for solving the above-mentioned problems is
A PC cluster system program having a plurality of terminals, wherein the program
Grouping means for grouping the plurality of terminals into at least one group;
Logical address providing means for collecting physical addresses of the terminals, and for each collected physical address, a logical address including identification information that can uniquely identify a group to which the terminal having the physical address belongs;
The physical address and the logical address assigned to the physical address are associated with each other to function as a physical address / logical address list generating unit that generates a physical address / logical address list.

The ninth invention for solving the above-mentioned problems is
A server program, said program comprising:
A logical address that collects the physical addresses of the terminals that make up the PC cluster and gives each collected physical address a logical address that includes identification information that can uniquely identify the group to which the terminal having the physical address belongs. Granting means;
The physical address and the logical address assigned to the physical address are associated with each other to function as a physical address / logical address list generating unit that generates a physical address / logical address list.

  In the PC cluster system of the present invention, a PC (terminal) is arbitrarily accommodated (connected) in a switch without being constrained by its physical location.

  The subnetwork configuration unit 1 divides the network so that the server can communicate only with a PC accommodated in one switch in a certain hierarchy.

  The physical address collection unit 11 collects physical addresses of individual PCs accommodated in the switch.

  The server assigns to the physical address collected by the physical address collection unit 11 a logical address that can uniquely identify the switch in which the PC corresponding to the physical address is accommodated.

  The partial physical address / logical address list generation unit 2 generates a partial physical address / logical address list 21 in which pairs of physical addresses and logical addresses of all PCs accommodated in the switch are registered.

  The server assigns a host name to the logical address of the PC accommodated in the switch as necessary.

The partial logical address / host name list generation unit 3 generates a partial logical address / host name list 31 in which pairs of logical addresses and host names of all PCs accommodated in the switch are registered. Generally, OS (Operating
Since the PC is identified by the host name on the (System), it is preferable that the switch accommodating the PC to which the host name is assigned can be uniquely identified from the host name. These operations are performed simultaneously or sequentially on all divided sub-networks.

  The physical address / logical address list integration unit 4 and the logical address / host name list integration unit 5 integrate the partial physical address / logical address list 21 and the partial logical address / host name list 31 of all the sub-networks, respectively. Then, a physical address / logical address list 41 and a logical address / host name list 51 in which all PCs constituting the PC cluster are registered are generated.

  When a PC cluster is activated, each PC specifies a physical address and requests the server to assign a logical address. Once the logical address is obtained, the server may record the logical address and perform setting so as not to request the provision of the logical address from the next time. The server refers to the physical address / logical address list 41 and assigns a logical address to each PC.

  There is no problem even if this logical address is assigned and the logical address is set to the PC manually. Further, the server performs conversion between the logical address and the host name, and each PC can search the logical address by designating the host name, and can retrieve the host name by designating the logical address.

  The server refers to the logical address / host name list 51 and performs a search requested by each PC. There is no problem even if a list of logical addresses and host names is sent to each PC and a search is performed on the PC side. For example, when setting an attribute of a PC on the operation management system, a logical address that can uniquely identify which switch contains the PC to which the logical address is assigned is given. Since the host name that can uniquely identify which switch contains the PC to which the host name is assigned is assigned based on the contents, physical attributes relating to the switch in which the PC is accommodated can be easily set. Can be set. From the set attribute, the operation management system can determine a PC to be used for parallel computation.

  According to the present invention, it is possible to connect the individual PCs constituting the PC cluster to an arbitrary switch regardless of the physical position, thereby saving the labor for building the PC cluster and reducing the human cost. Reduce.

  Furthermore, an IP address and a host name that can be identified by the switch to which the PC is connected can be assigned to each PC constituting the PC cluster.

  In addition, by assigning attributes related to switches that contain PCs as attribute information of PCs on the operation management system, it is possible to prevent parallel jobs from being executed across switches. It is possible to avoid performance degradation caused by executing parallel jobs on PCs that straddle.

  Embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a schematic diagram of a PC cluster system according to the present embodiment.

  The system according to the present embodiment includes a server, a switch, and a plurality of terminals (PC: Personal Computer).

  The server provides various commands and data to the PC.

  The PC is arbitrarily connected to the switch to form a PC cluster.

  FIG. 2 is a block diagram of the PC cluster system in the present embodiment.

  The switch includes a subnetwork configuration unit 1.

  The subnetwork configuration unit 1 groups PCs and performs partitioning so as to configure one switch for each group. Accordingly, by accommodating (connecting) the grouped PCs in one switch, a plurality of terminals constituting the network are converted into a plurality of sub-networks. In the present invention, a switch configured for each sub-network is a lower switch, and a switch for grouping the lower switches is an upper switch.

  The server includes a physical address collection unit 11, a partial physical address / logical address list generation unit 2, a partial logical address / host name list generation unit 3, a partial physical address / logical address list integration unit 4, a partial logical address / And a host name list integration unit 5. In addition to the above, the server has a control unit that performs various controls and a function as a DHCP (Dynamic Host Configuration Protocol) server.

  The partial physical address / logical address list generation unit 2 registers, for each lower switch, a partial physical address / logical address list 21 in which the physical addresses and logical addresses of all PCs accommodated in the lower switch are registered in association with each other. Is generated. Further, the partial physical address / logical address list generation unit 2 repeats the same processing for all the lower switches constituting the network, thereby creating the same number of partial physical addresses / logical address lists 21 as the number of lower switches. Generate. In this embodiment, the description will be made using the MAC address of the PC as the physical address and the IP address assigned to the PC as the logical address, but this is not restrictive.

  The partial logical address / host name list generation unit 3 generates, for each lower switch, a partial logical address / host name list 31 in which the host names of all PCs accommodated in the lower switch are registered in association with the logical addresses. . Further, by repeating the same processing for all the lower switches constituting the network, the same number of partial logical address / host name lists 31 as the number of lower switches are generated. The partial logical address / host name list is generated as necessary.

  The partial physical address / logical address list integration unit 4 creates a physical address / logical address list 41 from the partial physical addresses / logical address list 21 generated in the same number as the number of lower switches. The physical address / logical address list 41 is a list obtained by integrating the physical addresses / logical address list 21.

  The partial logical address / host name list integration unit 5 creates a logical address / host name list 51 from the partial logical address / host name list 31 generated by the same number as the number of lower switches. The logical address / host name list 51 is a list obtained by integrating the logical address / host name lists 31. The logical address / host name list 51 is generated as necessary.

  Next, the operation of the present embodiment will be described in detail with reference to FIGS. In the present embodiment, a case will be described in which power control is used using an IPMI (Intelligent Platform Management Initiative) interface. When using the IPMI interface, you can turn on and off any PC by specifying the IP address assigned to the IPMI interface of each PC. In addition, it is possible to check whether the power is turned on and in an operating state. Note that control of the power supply of the PC and monitoring of the state using the IPMI interface are performed using a well-known technique.

First, it connects so that it can communicate in the whole network. In this state, the server starts up the power of all PCs in the network using, for example, Magic Packet, etc.
System) is started (step S401). The temporary setting OS requests the server to allocate a temporary IP address, sets a response from the server to this request as a temporary IP address for the IPMI interface (step S402), and then shuts down itself (step S403).

  The server confirms that all PCs having temporary IP addresses for the IPMI interface issued by the self have been turned off (step S404).

  After the server confirms that all the PCs are powered off, the sub-network configuration unit 1 generates a sub-network (step S405), and further, between the PC and the server accommodated in one lower switch. A sub-network that can communicate only with is configured (step S406). Further, communication between lower switches in which PCs are accommodated is blocked at the upper switches.

  In this state, the server uses the IPMI interface to specify the temporary IP addresses of all PCs and turn on the PCs (step S407). At this time, since only the sub-network that can communicate only between the PC accommodated in a certain lower switch and the server is configured, the power of only the PC accommodated in this lower switch is turned on. Power-on instructions for PCs that are housed in other lower switches and cannot communicate are ignored.

  When the PC is powered on, the PC specifies its own MAC address and requests the server to obtain an IP address. Since the server functions as a DHCP (Dynamic Host Configuration Protocol) server, an IP address is assigned to an IP address acquisition request from the PC (step S408), and this is recorded in a log. In order to be able to recognize whether the server is accommodated in a specific switch from the IP address, for example, a network with 192.168.0.0 and a net mask of 255.255.0.0 is set as the system network, and the server is different for each switch. An IP address, for example, 192.168.n.0 (where n is identification information that uniquely identifies the switch) is assigned.

  The partial physical address / logical address generation unit 2 creates a partial physical address / logical address list 21 by associating the MAC address of the PC with the IP address assigned to the MAC address from the DHCP log of the server. (Step S409). In this embodiment, the DHCP function unit of the server functions as a physical address collection unit 11 that collects a MAC address that is a physical address.

  Generally, a PC is identified by a host name on the OS. When a database for correspondence between IP addresses and host names is required on the OS, the partial logical address / host name generation unit 3 makes a list of combinations of PC IP addresses and corresponding host names as necessary. The partial logical address / host name list 31 is created (step S410). When naming a host name, the server uses the host name to identify the system, the information that uniquely identifies the switch, and the information in the switch to check whether the host name is contained in a specific switch. The host name is named according to a rule such as a combination with a number indicating whether it is the second PC.

  When the creation of the partial physical address / logical address list 21 for one lower switch and the partial logical address / host name list 31 created as necessary is completed, the server connects to the switch using the IPMI interface. All the PCs that have been turned off are turned off (step S411).

  After confirming that the power of all the PCs accommodated in the switch has been turned off by the IPMI interface, the server configures the subnetwork of the next lower switch by the subnetwork configuration unit 1 as described above. Similar processing is performed on the switch. A partial physical address / logical address list 21 for the switch and a partial logical address / host name list 31 are created as necessary. Such processing is repeated by the number of lower switches constituting the system (step S412).

  When the creation of the partial physical address / logical address list 21 and the partial logical address / host name list 31 for all the lower switches is completed, the partial physical address / logical address list integration unit 4 performs the partial physical address / logical address list 21. To generate a physical address / logical address list 41 in which all PCs constituting the PC cluster are registered (step S413). Further, the partial logical address / host name list integration unit 5 creates a logical address / host name list 51 in which all PCs constituting the PC cluster are registered from the partial logical address / host name list 31 (step S414).

  When processing corresponding to each switch is sequentially performed as in this embodiment, a partial physical address / logical address list 21 to be created and a partial logical address / host name list 31 to be created as needed are physically If the contents are added as the address / logical address list 41 and the logical address / host name list 42, creation of the necessary lists is completed when the processing for all the switches is completed. Therefore, in this embodiment, the functions of the partial physical address / logical address list integration unit 4 and the partial logical address / host name list integration unit 5 are respectively the partial physical address / logical address list generation unit 2 and the partial logical address / The host name list generation unit 3 collects them.

  The operation management system creates a DHCP configuration file from the contents of the physical address / logical address list 41. If necessary, the contents of the logical address / host name list 51 are registered in an NIS (Network Information Service) database so that a PC can be designated based on the host name.

  Further, as the PC attribute information of the operation management system, the switch in which the PC is accommodated is identified from the IP address or the host name, and an attribute indicating that the PC is accommodated is set. For example, in an LSF (Load Sharing Facility) often used as an operation management system for a large-scale PC cluster system, a color attribute can be defined as a processor attribute.

  In this case, a specific color is assigned to the switch, and the same color attribute is given to PCs accommodated in the same switch. With this setting, when executing a parallel job that requires multiple PCs, the LSF will start executing the parallel job when the required number of PCs with the same color attribute have gathered. can do.

  With this kind of management, jobs that are executed in parallel can be executed on PCs that are accommodated in the same switch, and as described in the above problem, when executed on PCs that span between switches. Performance degradation can be avoided.

  Note that it is possible to identify which switch contains each PC from the IP address or host name, and the location of the physical PC is undefined. This is self-evident because the purpose is to allow connection to any switch regardless of the physical location of the PC, reduce the time and effort required to build a cluster, and reduce human costs. That is.

  After the above settings are completed, when the OS is first started on the PC, the IP address uniquely determined corresponding to the IP address assigned to itself is reset as the IP address for the IPMI interface. For example, the IP address of 192.168.101.1 is assigned to the IPMI port of the PC with 192.168.1.1. As a result, if a failure occurs during operation, if the device has an LED display function, specify the IP address for the IPMI interface of the PC that corresponds to the IP address of the PC that is assumed to have failed. LED can be turned on. Those who perform operation and maintenance can physically identify the PC on which the failure has occurred by the display of the LED, and can perform operations such as replacement.

  In the above embodiment, the case where IPMI is used has been described. However, power can be shut down from the OS, and power can be turned on using WOL (Wakeup on LAN) or operating the power switch sequentially, so there is no problem even if there is no IPMI interface. However, when the power supply is sequentially operated, the selective power supply cannot be turned on. In this case, as many servers as the number of switches are prepared. In the above embodiment, the partial physical address / logical address list 21 sequentially executed for each switch. Can be created at the same time. For this purpose, an inexpensive notebook PC or the like may be prepared as a server, which is not a problem as compared with the effect of reducing human costs by eliminating manual work.

  Next, the effect of this embodiment will be described.

  In this embodiment, while controlling the power supply of the PC by the IPMI interface, the subnetwork configuration unit 1 changes the configuration of the subnetwork, and the partial physical address / logical address list 21 for the PC accommodated in the lower switch is provided. The partial logical address / host name list 31 is sequentially created as necessary.

  Finally, the DHCP server setting file is created from the physical address / logical address list 41 and the host name information to be registered in the NIS server is created from the logical address / host name list 51.

  Both the logical address and host name are determined based on a naming rule that can identify the switch where the PC with the logical address and host name is stored, and is set by an operation management system such as LSF. As the PC-specific attribute information, an attribute indicating which switch the PC is connected to can be given.

  This allows connection to an arbitrary switch regardless of the physical location of the PC, saving labor when building a cluster and reducing human costs. Furthermore, it is possible to avoid performance degradation caused by executing parallel jobs on PCs that span switches.

  In addition, when setting the attributes of the PC on the operation management system, a logical address that can uniquely identify which switch contains the PC to which the logical address is assigned is given, and Since the host name that can uniquely identify which switch contains the PC to which the host name is assigned is assigned based on the contents, physical attributes relating to the switch in which the PC is accommodated can be easily set. Can be set. From the set attribute, the operation management system can determine a PC to be used for parallel computation.

FIG. 1 is a schematic diagram of a PC cluster system according to the embodiment of this invention. FIG. 2 is a block diagram of the PC cluster system according to the embodiment of this invention. FIG. 3 is a conceptual diagram of the PC cluster system according to the embodiment of this invention. FIG. 4 is a diagram showing the operation of the embodiment of the present invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Subnetwork structure part 2 Partial physical address / logical address list generation part 3 Partial logical address / host name list generation part 4 Partial physical address / logical address list integration part 5 Partial logical address / host name list integration part 11 Physical address collection part 21 Partial Physical Address / Logical Address List 31 Partial Logical Address / Host Name List 41 Physical Address / Logical Address List 51 Logical Address / Host Name List

Claims (9)

A PC cluster system having a plurality of terminals,
Grouping means for grouping the plurality of terminals into at least one group;
Logical address providing means for collecting physical addresses of the terminals, and for each collected physical address, a logical address including identification information that can uniquely identify a group to which the terminal having the physical address belongs;
A PC cluster system comprising physical address / logical address list generation means for generating a physical address / logical address list by associating the physical address with a logical address assigned to the physical address. 2. The PC cluster system according to claim 1, further comprising logical address / host name list generation means for generating a logical address / host name list by associating a logical address with a host name.   The PC cluster system according to claim 1, wherein the grouping unit is configured to block communication between groups. A server,
A logical address that collects the physical addresses of the terminals that make up the PC cluster and gives each collected physical address a logical address that includes identification information that can uniquely identify the group to which the terminal having the physical address belongs. Granting means;
A server comprising: a physical address / logical address list generating means for generating a physical address / logical address list by associating the physical address with a logical address assigned to the physical address. An address assignment method,
A step of grouping a plurality of terminals into at least one group;
A setting step for setting only terminals and servers belonging to one group to communicate with each other among the groups;
A logic that collects physical addresses of terminals that can communicate based on the setting, and assigns a logical address including identification information that can uniquely identify a group to which the terminal having the physical address belongs to each of the collected physical addresses. An address assignment step;
A physical address / logical address list generating step for generating a physical address / logical address list for each group by associating the physical address with a logical address given to the physical address;
And an integrating step of integrating the physical address / logical address list generated for each group into one. 6. The address assigning method according to claim 5, further comprising a logical address / host name list generation step for generating a logical address / host name list by associating a logical address with a host name.   7. The address assigning method according to claim 5, wherein the grouping step includes a blocking step for blocking communication between groups. A PC cluster system program having a plurality of terminals, wherein the program
Grouping means for grouping the plurality of terminals into at least one group;
Logical address providing means for collecting physical addresses of the terminals, and for each collected physical address, a logical address including identification information that can uniquely identify a group to which the terminal having the physical address belongs;
A program that functions as a physical address / logical address list generation unit that generates a physical address / logical address list by associating the physical address with a logical address assigned to the physical address. A server program, said program comprising:
A logical address that collects the physical addresses of the terminals that make up the PC cluster and gives each collected physical address a logical address that includes identification information that can uniquely identify the group to which the terminal having the physical address belongs. Granting means;
A program that functions as a physical address / logical address list generation unit that generates a physical address / logical address list by associating the physical address with a logical address assigned to the physical address.

Images