JP6102347B2 - Information device, printing system, computer program, and data transfer method - Google Patents

Description

  The present invention relates to a technique for transferring data via a load balancer.

  In a client server system including a plurality of servers, a load balancer (load distribution device) that distributes a load to the plurality of servers is used. For example, Patent Document 1 discloses a client server system that monitors access from a client to a load balancer and increases or decreases the number of servers to be operated according to increase or decrease of access.

  There is prior art relating to job control in a printing system having a load balancer. In the distributed printing system described in Patent Document 2, a job that requests printing is assigned to one of a plurality of print servers by a load balancer. The print server to which the job is assigned submits the job to the printer designated by the job. At this time, the print server that has submitted the job is registered in the database. As a result, job control such as job execution cancellation or printer change can be instructed to a print server that has submitted a job among a plurality of print servers.

  One type of load balancer is a so-called L4 load balancer. The L4 load balancer performs load distribution based on information up to the transport layer which is the fourth layer (layer 4) in the OSI reference model (OSI: Open Systems Interconnection). The transport layer is a layer corresponding to TCP and UDP (User Datagram Protocol) in a network using TCP / IP (Transmission Control Protocol / Internet Protocol). That is, the L4 load balancer discriminates the received packet based on the transport layer protocol (TCP / UDP) and the IP address in the third network layer. Unlike the L7 load balancer, which is another widely known type of L4 load balancer, the L4 load balancer does not refer to information in the application layer, and therefore has an advantage of higher distribution performance than the L7 load balancer.

JP 2002-163241 A JP 2009-151467 A

  In a network using the above-described L4 load balancer, a packet is determined based on a unique IP address of the transmission source. Therefore, a load caused by a persistence function for fixing a transfer destination of a packet received from the same transmission source within the persistence duration time. Concentration can occur. For example, when a large number of requests are sequentially sent from a single information device to the server cluster via the L4 load balancer, all requests are sent to one server in the server cluster regardless of whether the clients are the same. Will be assigned. Usually, the processing speed in the server is greatly reduced when the load is excessively concentrated. A decrease in the processing speed of the server causes a delay in packet transfer to the server.

  In view of such circumstances, an object of the present invention is to realize transfer control useful for reducing delay in data transfer via an L4 load balancer.

  An apparatus that achieves the above object is an information device that communicates with a load balancer, wherein a transmission unit that transmits a request for a server from a client to the load balancer, and the transmission unit transmits the request to the load balancer. An acquisition unit that periodically acquires the data transfer rate in the transmission, and the data transfer rate acquired by the acquisition unit is equal to or less than a preset threshold rate that is smaller than the normal data transfer rate A suspension instruction unit that causes the transmission unit to defer transmission of a request that has not started transmission, and the request that was performed when the transmission unit was instructed to hold by the suspension instruction unit More than the persistence duration in the load balancer after transmission is completed After a lapse of the set time set in advance in the, and a resumption instruction section resuming the transmission of the by releasing the hold of the transmission to the transmitting unit to the load balancer.

  According to the present invention, it is possible to realize transfer control useful for reducing delay in data transfer in a load balancing environment using an L4 load balancer.

1 is a diagram illustrating a configuration of a printing system according to an embodiment of the present invention. FIG. 6 is a diagram illustrating job transfer control in a print server. FIG. 10 is a diagram illustrating an example of transfer destination switching by a load balancer caused by job transfer control in a print server. 2 is a diagram illustrating a hardware configuration of a print server. FIG. 2 is a diagram illustrating a functional configuration of a print server. FIG. 3 is a flowchart illustrating an outline of an operation of a print server. 6 is a flowchart of a transfer processing routine executed in the print server. 6 is a flowchart showing a hold instruction operation of the transfer monitor in the print server.

  FIG. 1 shows a printing system 1 including a plurality of devices on a network as an example of a load balancing environment. The printing system 1 is a so-called pull printing system, and a printing apparatus captures and executes a stored job (print job) from a storage destination. In the printing system 1, the job is transferred from the print server 10 as the first information device related to load distribution to one of the plurality of management servers 31, 32, and 33 as the second information device via the load balancer 20.

  In pull printing, first, users of client terminals 5, 6, 7, and 8, such as personal computers and smartphones, instruct printing of a document. Upon receiving the instruction, the client terminals 5, 6, 7, 8 generate a job in a predetermined data format requesting printing and send it to the print server 10. The job includes user information. The job may include document data (print data), or may include information specifying a document stored in advance in the document server as a print target instead of the document data.

  The print server 10 temporarily stores the jobs received from the client terminals 5, 6, 7, and 8 in the memory as the spool 110, converts them into packets destined for the server 30, and transmits the packets to the load balancer 20. The server 30 here is a virtual server having a plurality of management servers 31, 32, 33 as entities. When sending a job, it is possible to handle a plurality of jobs of a prescribed number or less collectively.

  The load balancer 20 transfers the packet received from the print server 10 to one selected from the plurality of management servers 31, 32, 33 according to the load distribution algorithm. The load distribution algorithm is a method of allocating requests to the server with the smallest load (for example, the least connection method). Depending on the selection of the transfer destination in the load balancer 20, the job is allocated to one of the plurality of management servers 31, 32, 33.

  The management servers 31, 32, and 33 perform authentication processing for the jobs assigned to them. Then, the management servers 31, 32, and 33 store the job determined to be appropriate in the job server 35. The management servers 31, 32, and 33 share information indicating the storage status in the job server 35.

  After the job is saved, the user who has instructed printing at the client terminals 5, 6, 7, 8 can cause the plurality of printing apparatuses 41, 42, 43 to execute printing. The user arbitrarily selects one of the printing apparatuses 41, 42, and 43, and operates the selected printing apparatus to instruct display of a list of jobs related to the user. Upon receiving the instruction, the printing apparatus inquires of the stored job to the server 30 connected via the load balancer 40. The inquiry reaches one of the management servers 31, 32, 33 selected by the load balancer 40. The inquiring printing apparatus displays job information provided from the management server that has received the inquiry. When the user designates a job to be executed and instructs execution of printing, the printing apparatus acquires a job from the job server 35 via the server 30 and prints a document specified by the acquired job. The printing apparatuses 41, 42, and 43 may be MFPs (Multi-functional Peripherals) or other image forming apparatuses capable of network printing.

  In such a printing system 1, an L4 load balancer is used as the load balancer 20. However, it is not necessary to be an L4 switch dedicated machine, and the load balancer 20 may be a device capable of switching between the L4 switch mode and the L7 switch mode.

  The load balancer 20 selects a packet transfer destination (any one of the management servers 31, 32, and 33) from the print server 10 based on information in the fourth layer and lower layers in the OSI reference model. The information below the fourth layer is information recognized by the protocol of the fourth layer (transport layer) (for example, port number) and information recognized by the protocol of the third layer (network layer) (for example, IP address). is there. This means that the client terminals 5, 6, 7, and 8 that generated the job transmitted from the print server 10 are not distinguished. In other words, when the job is viewed as a request to the server 30, the load balancer 20 recognizes that “the client for the server 30 is the print server 10” for all requests.

  The load balancer 20 has a persistence function. The persistence function here refers to the transfer destination selected for the previously input request when multiple requests with the same client are input in order within the specified persistence time. It is a function that selects even for a requested request. That is, the load balancer 20 fixes the transfer destination for requests from the same client within the persistence duration.

  The print server 10 that transmits a job to the load balancer 20 has a transfer monitor 120 that operates to characterize the transmission timing. The transfer monitor 120 monitors a data transfer rate (also referred to as a data transfer rate) depending on a load on the server 30 to which the job is transferred. When the data transfer rate decreases, the transfer monitor 120 causes the print server 10 to interrupt transmission as described later, thereby causing the load balancer 20 to release the transfer destination.

  FIG. 2 shows job transfer control in the print server 10. A timing chart showing transfer periods of a plurality of jobs (denoted as JOB-1 to 12 in the figure) is drawn in the upper part of FIG. The middle graph in FIG. 2 shows the transition of the load on the transfer destination server in a state where the transfer destination is fixed in the load balancer 20, and the lower graph shows the transition of the data transfer rate R.

At time t1, transfer of the job group 91 in which three jobs (JOB-1 to JOB-3) are batched, that is, transmission of a packet corresponding to each job is started. The three jobs in the job group 91 are handled collectively at the time of transfer because they are received continuously from the same client terminal, for example. At time t2 during the transfer of the job group 91, the transfer of the job group 92 including the two jobs is started. In the example, since the time from time t1 to time t2 is shorter than the persistence duration, the job group 92 is transferred to the same transfer destination as the previous job group 91. As a result, the print server 10 transmits a total of five jobs in parallel to any of the management servers 31, 32, 33 selected by the load balancer 20 .

  At time t3, transfer of a job group 93 consisting of four jobs is started, and at time t4, transfer of one job 94 is started. Since the transfer started first is continuing at time t4, the transfer destination management server performs a reception process for a total of 10 jobs (JOB-1 to 10).

  As the number of jobs received in parallel increases, the addition added to the transfer destination management server increases. In this example, when the load is smaller than a predetermined amount determined by the processing capabilities of the management servers 31, 32, 33, the data transfer rate R is substantially maintained at the normal data transfer rate Rs, and the load exceeds the predetermined value. As the rate further increases, the data transfer rate R decreases.

  At time t5 when the data transfer rate R decreases from the normal data transfer rate Rs to the threshold rate Rth, a job that has been spooled at this time and transfer has not started (this job is referred to as an “untransfer started job”). ) For 95 and 96, the print server 10 holds the transfer as necessary. The case where the hold is necessary is a case where the transfer end time of the non-transfer start jobs 95 and 96 is predicted to be advanced by holding. Whether or not the hold is necessary depends on the data amount of the non-transfer start jobs 95 and 96. Note that preparation for transmission such as generation of a packet can be advanced even if it is pending.

  In the transfer destination management server, after receiving the job, processing associated with the reception of the job is performed such as authenticating the job or transferring the job to the job server. In some cases, it responds to access from the printing apparatuses 41, 42, and 43. Therefore, even if the reception of the job is completed, the load is not always reduced, and the high load state may continue.

  When the print server 10 holds the transfer of the untransfer start jobs 95 and 96, from the time t6 when the transfer of all the jobs in the middle of the transfer ends when the hold starts, to the time t7 when the hold set time T2 elapses. Keep on hold. Then, the hold is released at time t7, and transfer of the untransfer start jobs 95 and 96 is started. Thereafter, transfer of the job spooled after time t5 is started.

The hold set time T2 is the same as the persistence duration (T1) set in the load balancer 20 or a length of time obtained by adding an appropriate margin to the persistence duration. When the transmission from the print server 10 is interrupted from the time t6 over the hold set time T2, the transfer destination is fixed in the load balancer 20 . That is, the load balancer 20 selects a transfer destination again for subsequent access from the print server 10.

When the load balancer 20 selects a transfer destination again, if the load of the management server (for example, the management server 31) selected last is smaller than the load of the other management servers (for example, the management servers 32 and 33), Then, the management server 31 is selected again as the transfer destination. When the load on one or both of the other management servers 32 and 33 is smaller than the load on the management server 31, the transfer destination is switched. In any case, when the transfer destination is selected again, the load on the transfer destination can be reduced compared to immediately before the transfer is suspended. If the management server in the low load state becomes the delivery destination, the transfer of the non-transfer start jobs 95 and 96 is completed earlier than the transfer of the non-transfer start jobs 95 and 96 to the management server in the high load state having a reduced data transfer rate. can do.

  In FIG. 3, the switching of the transfer destination by the load balancer (LB) 20 is schematically shown. The transfer destination selected by the load balancer 20 for the job group 92 is the management server 31. The job group 93 transmitted from the print server 10 at a time interval within the persistence duration T 1 following the job group 92 is also transferred to the management server 31. Similarly, the job 94 is also transferred to the management server 31. However, the job group 97 transmitted after the persistence time T1 has elapsed from the end of the transfer of the job 94 is transferred to the management server 33 different from the management server 31 to which the job 94 is transferred.

  Hereinafter, the configuration and operation of the print server 10 will be described in more detail.

  FIG. 4 shows the hardware configuration of the print server 10. The print server 10 includes a CPU (Central Processing Unit) 12 as a computer for executing a control program and various applications, a ROM (Read Only Memory) 13 for storing the control program, and a RAM (Random Access Memory) as a work area for executing the program. ) 14. The storage 15 is a hard disk drive or other large-capacity nonvolatile storage device, and stores programs, various data, received jobs, and the like. The communication interface 16 connects the print server 10 to the network. The removable memory drive 17 is used for installing a program from a removable memory, for example.

  FIG. 5 shows a functional configuration of the print server 10. The print server 10 includes a spooler 101, a transmission unit 102, and a transfer monitor 120. These components are functional elements realized by the CPU 12 executing a predetermined control program.

  The spooler 101 stores jobs from the client terminals 5, 5, 6, and 7 received by the communication interface 16 in a predetermined memory (spool 110 in FIG. 1), and manages jobs temporarily saved thereby. As the transmission by the transmission unit 102 progresses, the spooler 101 sequentially delivers the job to the transmission unit 102.

  The transmission unit 102 transmits a packet corresponding to a job to be transmitted to the load balancer 20 connected via the communication interface 16. The transmitted packet is transferred from the load balancer 20 to the transfer destination selected by the load balancer 20. The transmission unit 102 suspends the start of a transmission operation for sending a packet for an untransferred start job to the communication interface 16 in accordance with a suspension instruction from the transfer monitor 120. When resumption is instructed from the transfer monitor 120 after the suspension, the transmission unit 102 starts transmission of a non-transfer start job.

  The operation monitor 120 obtains the data transfer rate R, the hold instruction unit 124 that gives a hold instruction to the transmission unit 102, the calculation unit 126 that calculates the time related to the hold instruction, and the restart instruction to the transmission unit 102 A restart instruction unit 128 is provided.

  The acquisition unit 122 periodically acquires the data transfer rate R of transmission in the communication interface 16 during the period in which the data transfer indicated by the status of the transmission operation notified from the transmission unit 102 is performed. Then, every time it is acquired, the data transfer rate R is notified to the hold instruction unit 124.

  For acquiring the data transfer rate R, the acquisition unit 122 calculates the data transfer rate R using a benchmark test technique, and if a program for monitoring the network usage status is separately installed, the data is transferred from the program. There is a form in which the transfer rate R is captured. The acquisition cycle is set according to the persistence duration T1. For example, when the persistence duration T1 is 10 seconds, a period within a range of 0.1 to 1 second, which is 1 to 1/10, may be set. When calculating the data transfer rate R, the quotient obtained by dividing the number of transmission bits within the time corresponding to the acquisition cycle by the cycle may be used as the data transfer rate R, or for each divided time obtained by equally dividing the time corresponding to the cycle. The average value of the data transfer rates may be used as the data transfer rate R.

  The hold instruction unit 124 gives a calculation instruction to the calculation unit 126 when the data transfer rate R notified from the acquisition unit 122 is equal to or lower than the threshold (see FIG. 2). Upon receiving the instruction, the calculation unit 126 calculates the estimated required time TX and the total time TY, and notifies the hold instruction unit 124 of the calculation. The estimated required time TX is a time from the start to the end of transmission when it is assumed that an untransfer start job is transmitted at the data transfer rate R. The total time TY is a time obtained by adding the hold set time T2 to the estimated required time TXs that is a time from the start to the end of transmission when it is assumed that an untransfer start job is transmitted at the normal data transfer rate Rs. The hold instruction unit 124 gives a hold instruction to the transmission unit 102 when the total time TY is shorter than the estimated required time TX.

  The resume instruction unit 128 is notified that the hold instruction unit 124 has given the hold instruction to the transmission unit 102. The resumption instruction unit 128 gives a resumption instruction to the transmission unit 102 when the hold set time T2 has elapsed since the transfer of the job that was being transferred when the transmission of the untransfer start job was suspended.

  The flowchart in FIG. 6 shows an outline of the operation of the print server 10. The print server 10 receives jobs from the client terminals 5, 6, 7, and 8 (S01), spool processing for transferring jobs in order (S02), and transfer processing for sending jobs to the server 30 (S03). Repeatedly.

  FIG. 7 is a flowchart of the transfer processing routine.

  The transmission unit 102 determines a job to be transferred next in cooperation with the spooler 101 (S11). At this time, there are cases where a plurality of jobs are targeted at once. Subsequently, the transmission unit 102 starts transferring the job determined as the target (S12), and notifies the transfer monitor 120 of “start” as the status (S13). That is, it informs that the data transfer rate R should be monitored.

  The presence / absence of a hold instruction is checked (S14). If there is no hold instruction (NO in S14), the flow proceeds to step S21. A transmission process for sending out the packet of the job being transferred is performed (S21), and the presence or absence of an untransferred job is checked (S22). If there is an untransferred job, the flow returns to step S11. If there is no untransferred job, “end” is notified as a status to the transfer monitor 120 when the transfer is completed (S23), and the flow returns to the main routine of FIG.

  On the other hand, when there is a hold instruction (YES in S14), the transmission unit 102 continues the transmission process of sending out the packet of the job being transferred until there is no job being transferred (S15, S16). When the transmission of the packet of the job that was in the middle of transfer ends (YES in S16), the transmission unit 102 notifies the transfer monitor 120 of “pending” (S17). In response to this notification, the restart instruction unit 128 of the transfer monitor 120 sets a hold timer for counting the hold set time T2 (S18). When the counting of the hold timer is completed (YES in S19), the resumption instruction unit 128 gives a resumption instruction to the transmission unit 102 (S20). The flow returns to step S11, and transfer of an untransferred job is started.

  FIG. 8 is a flowchart showing the hold instruction operation of the transfer monitor 120.

  The acquisition unit 122 waits for the transmission unit 102 to start transferring a job (S31). When the transfer is started (YES in S31), the acquisition unit 122 acquires the data transfer rate R and notifies the holding instruction unit 124 (S32). The holding instruction unit 124 determines whether or not the data transfer rate R is equal to or less than a threshold value (S33). If the data transfer rate R is greater than the threshold (NO in S33), the flow returns to step S31. During the transfer, the acquisition of the data transfer rate R and the comparison with the threshold value are repeated.

  If the data transfer rate R is equal to or less than the threshold (YES in S33), the flow proceeds to step 34. Here, as a specific example, the normal data transfer rate Rs is set to “1 Gbps (1 gigabit per second)”, the threshold value is set to “0.6 Gbps”, and the acquired latest data transfer rate R is set to “0.56 Gbps”.

  In step S <b> 34, the calculation unit 126 that receives the calculation instruction from the holding instruction unit 124 calculates the data amount DQ of the untransfer start job based on the job information from the spooler 101. If there are a plurality of non-transfer start jobs 95 and 96 as in the example of FIG. 2, the sum of the data amount of each job is calculated as the data amount DQ. As a specific example, the calculated value of the data amount DQ is “160 Gbits”.

Subsequently, the calculation unit 126 calculates the estimated required time TX (S35). In the specific example, since the data amount DQ is “160 Gbit” and the data transfer rate R is “0.56 Gbps”, the estimated required time TX is about 280 seconds as shown in the following equation.
TX = 160 [Gbit] /0.56 [Gbit / sec.] ≈280 [sec.]
Subsequently, the calculating unit 126 calculates the estimated required time TXs and the total time TY (S36, S37). In the specific example, since the normal data transfer rate Rs is “1 Gbps”, the estimated required time TXs is about 160 seconds. Further, if the hold setting time T2 is set to “10 seconds” like the persistence duration time T1, and the estimated required time TXs is regarded as 160 seconds, the total time TY is 170 seconds.
TXs = 160 [Gbit] / 1 [Gbit / sec.] ≈160 [sec.]
TY = TXs + T2 = 160 [sec.] + 10 [sec.] = 170 [sec.]
The hold instruction unit 124 compares the calculated estimated required time TX and the total time TY (S38). In the specific example, the total time TY is shorter than the estimated required time TX (170 seconds <280 seconds). When the total time TY is shorter than the estimated required time TX as in this example (YES in S38), the hold instruction unit 124 causes the transmission unit 102 to hold the transfer start of the untransfer start job (S39). Thereafter, the flow returns to step S31.

  If the total time TY is longer than the estimated required time TX (NO in S38), the flow immediately returns to step S31. In this case, the transfer start of the non-transfer start job is not suspended, and the transfer of the non-transfer start job is started when the packet preparation is complete and transfer start is possible. The reason why the transfer start is not suspended is that the start of the transfer is delayed due to the hold, so even if the transfer can be performed at the normal data transfer rate Rs, the transfer is not started when the transfer is performed at a lower data transfer rate R. This is because the transfer can be completed early.

  According to the above embodiment, since the load balancer 20 selects one of the management servers 31, 32, 33 having the smallest load as a transfer destination, all the management servers 31, 32, 33 are not overloaded. When the transfer destination is switched, transfer at the normal data transfer rate Rs becomes possible.

  In the above-described embodiment, the load balancing algorithm of the load balancer 20 does not necessarily need to be an algorithm for selecting one of the transfer destination candidates with the smallest load. An algorithm that changes the transfer destination for each selection may be used. As long as the transfer destination is switched, there is a possibility that the communication environment is improved from a state where the data transfer rate is lowered to a state where the data transfer rate is not lowered.

  The network configuration of the printing system 1 can be changed as appropriate. The number of client terminals 5, 6, 7, and 8 connected to the print server 10 is not limited to “4” in the example of FIG. The print server 10 and other devices may be connected as devices that transmit jobs to the load balancer 20. The number of management servers 31, 32, 33 to which the load balancer 20 assigns jobs is not limited to “3” in the example.

1 Printing system 5, 6, 7, 8 Client terminal (client)
10 Print server (information equipment, first information equipment, first server)
20 Load balancer 30 Server 31, 32, 33 Management server (second information device, second server)
91, 92, 93, 97 Job group (request)
94 jobs (request)
95,96 Untransferred start job (request that has not started transmission)
101 Spooler 102 Transmitter R Data Transfer Rate Rs Normal Data Transfer Rate Rth Threshold Speed 122 Acquisition Unit 124 Reservation Instructing Unit 126 Calculation Unit 128 Resume Instructing Unit T1 Persistence Duration Time T2 Reservation Setting Time (Setting Time)
TX Estimated required time (first estimated required time)
TXs Estimated required time (second estimated required time)
TY Total time 41, 42, 43 Printing device

Claims (7)

An information device that communicates with a load balancer,
A transmission unit for transmitting a request from the client to the server to the load balancer;
An acquisition unit that periodically acquires a data transfer rate in the transmission during a period in which the request is transmitted to the load balancer by the transmission unit;
When the data transfer rate acquired by the acquisition unit is equal to or lower than a preset threshold rate smaller than the normal data transfer rate, the transmission unit holds the transmission for a request that has not started transmission. An indicator,
A preset set time longer than the persistence duration in the load balancer has elapsed since the transmission of the request that was performed when the transmission was instructed to the transmission unit by the hold instruction unit And a restart instruction unit that causes the transmission unit to release the suspension of the transmission and resume transmission to the load balancer. A spooler that spools a request to be transmitted by the transmission unit;
When the acquired data transfer rate is equal to or lower than the threshold rate, transmission in the case of transmitting a request that has not started transmission among the requests spooled by the spooler at the acquired data transfer rate A first estimated required time that is a time from the start to the end of the transmission, a second estimated required time that is a time from the start to the end of transmission when the request is transmitted at the normal data transfer rate, and the A calculation unit for calculating the total time with the set time,
When the calculated total time is shorter than the first estimated required time, the holding instruction unit causes the transmitting unit to hold transmission of the request that has not started transmission, and the calculated total time 2. The information device according to claim 1, wherein when the time is equal to or longer than the first estimated required time, the transmission unit is not instructed to hold the transmission of the request. A computer program executed in an information device that communicates with a load balancer,
In the computer that the information device has,
Processing to send a request from the client to the server to the load balancer;
A process of periodically acquiring a data transfer rate in the transmission during a period in which the request is transmitted to the load balancer;
When the acquired data transfer rate is equal to or lower than a preset threshold rate smaller than the normal data transfer rate, a process for deferring transmission for a request that has not started transmission;
After a lapse of a preset set time longer than the persistence duration in the load balancer after transmission of the request that was performed when transmission of the request that has not started transmission is suspended And a process for canceling the suspension of the transmission and restarting the transmission to the load balancer. On the computer,
A process for spooling a request to be transmitted, and a request that has not started transmission among the spooled requests when the acquired data transfer rate is equal to or lower than the threshold rate. The first estimated required time, which is the time from the start to the end of transmission when transmitting at the data transfer rate, and the time from the start to the end when transmitting the request at the normal data transfer rate And further executing a process of calculating a total time of the second estimated required time and the set time,
As the processing for deferring transmission, when the calculated total time is shorter than the first estimated required time, the transmission for the request that has not started transmission is deferred, and the calculated total time is The computer program according to claim 3, wherein if it is equal to or longer than the first estimated required time, a process that does not suspend transmission is executed. A first server that receives a print job from each of a plurality of client terminals, a plurality of second servers that transmit a print job to a printing apparatus, and a plurality of print jobs that are transmitted from the first server to the plurality of second servers A printing system having a load balancer to allocate,
The load balancer maintains the allocation destination setting in a state where a print job is transmitted from the first server to the second server determined as an allocation destination among the plurality of second servers without interruption for longer than the persistence duration time. And having a persistence function for canceling the maintenance of the allocation destination setting when the transmission of the print job from the first server is interrupted over the persistence duration,
The first server is
A transmission unit that transmits print jobs received from the plurality of client terminals to the load balancer;
An acquisition unit that periodically acquires a data transfer rate in the transmission during a period in which the transmission of the print job to the load balancer is performed by the transmission unit;
When the data transfer rate acquired by the acquisition unit is equal to or lower than a preset threshold rate smaller than the normal data transfer rate, the transmission unit holds the transmission of a print job that has not started transmission. A hold instruction section;
A preset set time longer than the persistence duration in the load balancer after the transmission of the print job that was performed when the hold instruction is given to the transmission unit by the hold instruction unit After the elapse of time, the printing system includes: a resumption instruction unit that causes the transmission unit to release the suspension of transmission and resume transmission to the load balancer. The first server is
A spooler for spooling a print job to be transmitted by the transmission unit;
When the acquired data transfer speed is equal to or lower than the threshold speed, a print job that has not started transmission among the print jobs spooled by the spooler is transmitted at the acquired data transfer speed. A first estimated required time that is a time from the start to the end of transmission in the case, and a second estimate that is a time from the start to the end of transmission when the print job is transmitted at the normal data transfer speed. A calculation unit for calculating a total time of the required time and the set time,
When the calculated total time is shorter than the first estimated required time, the hold instruction unit causes the transmission unit to hold transmission of the print job that has not started transmission, and the calculated total The printing system according to claim 5, wherein when the time is equal to or longer than the calculated first estimated required time, the transmission unit is not instructed to hold the transmission of the print job. A data transfer method for transferring data from a first information device to one of a plurality of second information devices using a load balancer that selects a transfer destination,
When the load balancer performs load distribution based on identification information that identifies the first information device as a transfer destination selection operation,
The first information device monitors the transition of the data transfer rate in the transmission of the data to the load balancer, and transfers the data when the data transfer rate falls below a threshold rate less than the normal data transfer rate. A data transfer method comprising: interrupting and resuming the transfer after a time has elapsed for the load balancer to reselect a transfer destination.

Images