JP2018025864A - Load control device, load control method, and load control processing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To properly control a response to an important request.SOLUTION: A load control device in an embodiment comprises: management means for managing a request which includes identification information for identifying a resource to be operated and which is for requesting an operation of the resource, and a latest reception time of the request, as requests before being added to a request queue; reception means for receiving a new request; calculation means for calculating a difference value which is a difference between the reception time of the received request and a reception time of a request including the same identification information as that included in the request being managed; and designation means for designating a request having a large difference value calculated by the calculation means, out of multiple requests having different identification information, as a request to be added to the request queue preferentially.SELECTED DRAWING: Figure 1

Description

  Embodiments described herein relate generally to a load control device, a load control method, and a load control processing program.

  With the spread of the Internet, various services can be used via a network. Examples are mail, homepage browsing, search, online transactions, IP phone calls, video on demand, and the like. Although these network services can be provided in various forms, in recent years, the use of a Web server has become the mainstream as an interface with a client.

  The basic mechanism of a service using a Web server (Web service) is as follows. First, a client transmits a request to which a URL (Uniform Resource Locator) for identifying content to be acquired is added to a Web server. When the Web server receives the request, the content for the URL in the request is sent back to the client as a response. The Web service is provided by repeating this request-response.

  Here, the entire server system that performs the Web service is referred to as a Web server, and a function for generating content according to a request on the Web server is referred to as a Web application.

  In the prior art, the server dynamically determines the priority of the response to the request for each IP address of the client that sent the request.

  Specifically, when a client accesses a Web page, a series of requests from the initial access to the server until the content in the page is acquired from the server, or a series of page transition requests during the session, the client's IP address Processing is preferentially performed in units.

Here, a load control system applied to the REST API (application programming interface) will be described. FIG. 7 is a block diagram illustrating a functional configuration example of a conventional load control system.
As shown in FIG. 7, the server 100 of the conventional load control system includes an application communication unit 101, a priority assignment unit 102, a priority prediction unit 103, a request queue management unit 104, a request execution unit 105, a resource operation unit 106, It has a resource DB (database) 107.

  REST API, for example, responses to requests that indicate operations (GET, POST, PUT, DELATE (get, create, update, delete) resources) for resources that are information such as web content articles and tags I do. This request includes a URI that uniquely identifies the target resource.

  The above operation realizes four processes of CRUD, that is, Create (create-POST / PUT), Read (read-GET), Update (update-PUT), and Delete (delete-DELETE) using HTTP methods. Is.

  The application communication unit 101 receives a request from the client via the network, and returns an appropriate response to the request to the client. When there are a plurality of requests before returning a response with respect to the above request, the priority assigning unit 102 assigns a priority that is the order in which the response is returned with priority.

FIG. 8 is a diagram showing an example of a high priority address table used in a conventional load control system. The priority prediction unit 103 stores the high priority address table shown in FIG. 8 in the internal memory.
In this high-priority address table, the start of a series of processes for displaying a Web page as a response (access to the root resource) and the IP address of the client that is the transmission source of the request started by the session are registered.

Thereafter, requests from clients corresponding to IP addresses registered in the high priority address table are given priority over requests from clients corresponding to IP addresses not registered in this high priority address table. Processing is performed.
The IP address registered in the high priority address table is deleted after a predetermined time after the response to the request from the client corresponding to the IP address is completed.

The request queue management unit 104 manages the order of responses for a plurality of requests.
The request execution unit 105 instructs the resource operation according to the order managed by the request queue management unit 104, and returns a response to the application communication unit 101.

The resource operation unit 106 operates resources in accordance with instructions from the request execution unit 105.
The resource DB 107 includes a storage medium such as a non-volatile memory, for example, and stores information regarding resources. This resource is operated by the resource operation unit 106 as described above.

JP 2008-059040 A

  As described above, when the priority of response is determined for each IP address of the client, if it passes through NAT (Network Address Translation) or proxy, it is important for other clients due to requests from some clients. An inappropriate priority may be set for a request.

  In the M2M / IoT (Internet of Things) platform compliant with oneM2M (Machine to Machine), M2M / IoT devices, applications, and acquired data are all managed in a tree structure. Receives operations, information changes, data from devices, etc., and manages and stores them as resources. Then, the server determines the operation target based on the URI included in the request to the API, and determines the operation content of the resource using the HTTP method included in the request.

  In the above M2M / IoT platform, when a request from a client corresponding to a certain IP address is once rejected on the server side, data as a response to an important request is not acquired for a long time depending on the request frequency on the client side This can lead to a situation that leads to a malfunction of the entire system.

  An object of the present invention is to provide a load control device, a load control method, and a load control processing program that can appropriately control a response to an important request.

  To achieve the above object, a first aspect of a load control device according to an embodiment of the present invention includes identification information for identifying an operation target resource, a request for requesting an operation of the resource, and a request for the request. Management means for managing the latest reception time as a request before being added to the request queue, reception means for receiving a new request, reception time of the received request, and identification included in the request to be managed A calculation unit that calculates a difference value that is a difference from a reception time of a request including the same identification information as the information, and the difference value calculated by the calculation unit among the plurality of requests that are managed and have different identification information A designation means for designating a large request as a request to be added to the request queue with priority. To provide a device for.

  According to a second aspect of the load control device configured as described above, in the first aspect, the request includes a type of the requested operation, and the calculation unit manages the reception time of the received request and the management. First calculation means for calculating a first difference value that is a difference between the latest reception time of the request including the same identification information as the identification information included in the request, and the reception time of the received request, A second difference value that is the difference between the latest reception time of a request that is managed and includes the same identification information as the identification information included in the received request and that requests the same operation as the operation requested by the received request; Second specifying means for calculating the request, wherein the specifying means calculates the plurality of requests managed by the first calculation means for the plurality of requests having different identification information. When the first difference value calculated is the same for the plurality of requests having different identification information, designated as a request to be preferentially added to the request queue in descending order of the first difference value. An apparatus is provided that designates the plurality of requests as requests to be added to the request queue preferentially in descending order of the second difference value calculated by the second calculation means.

  According to a third aspect of the load control device configured as described above, in the second aspect, the designation unit has the same calculated first difference value for the plurality of requests having different identification information, and When the calculated second difference value is the same for the plurality of requests having different identification information, the request is specified as a request to be added to the request queue with priority from the oldest reception time of the received request. An apparatus is provided.

  According to a fourth aspect of the load control apparatus configured as described above, in the first aspect, the request queue to which the received request is added provides an apparatus that is classified according to a type of operation requested by the request.

  In order to achieve the above object, an aspect of a load control method according to an embodiment of the present invention is a method applied to a load control device, including identification information for identifying a resource to be operated, and controlling the operation of the resource. The request to request and the latest reception time of this request are managed as a request before being added to the request queue, a new request is received, the reception time of the received request, and the request to be managed are Calculating a difference value that is a difference between a reception time of a request including the same identification information as the included identification information, and managing a request having a large calculated difference value among a plurality of requests having different identification information. A method for preferentially specifying a request to be added to the request queue is provided.

  An aspect of a load control processing program in an embodiment of the present invention is a program used in a computer that operates as a part of the load control device in the first aspect, and the computer is controlled by the management unit, the reception unit, and the calculation. And a program for functioning as the specifying means.

  According to the present invention, it is possible to appropriately control a response to an important request.

The figure which shows the function structural example of the load control system in embodiment of this invention. The block diagram which shows the function structural example of the server of the load control system in embodiment of this invention. The figure which shows an example of the frequency table used with the load control system in embodiment of this invention in a table format. The flowchart which shows an example of the process sequence from reception of the request by the load control system in embodiment of this invention to operation execution. The flowchart which shows an example of the procedure of a priority provision and order control process by the load control system in embodiment of this invention. The flowchart which shows an example of the procedure of the request addition process to a queue by the load control system in embodiment of this invention. The block diagram which shows the function structural example of the server of the conventional load control system. The figure which shows an example of the high priority address table used with the conventional load control system.

Embodiments according to the present invention will be described below.
FIG. 1 is a diagram illustrating an example of functions of a load control system according to an embodiment of the present invention.
As shown in FIG. 1, a load control system according to an embodiment of the present invention includes a plurality of clients and a server with a load control function (hereinafter simply referred to as a server) 10 that is a load control device. It is a system connected via. The clients are, for example, the application 1-1, the device 1-2, and the gateway 1-3. The gateway 1-3 is provided between the device 1-3b and the network 3.

  In the present invention, the priority of the response to the request is determined so that the server 10 can process an important request first based on the URI, HTTP method, and request frequency included in the request from the client to the API.

FIG. 2 is a block diagram illustrating a functional configuration example of a server of the load control system according to the embodiment of the present invention.
As shown in FIG. 2, the server (M2M / IoT platform) 10 of the load control system according to the embodiment of the present invention includes an application communication unit 11, a priority assignment unit 12, a frequency table 13, a request queue management unit 14, and a request execution. Section 15, resource operation section 16, and resource DB 17.

  The application communication unit 11 receives (receives) a request from the client via the network, and returns an appropriate response to the request to the client. The application communication unit 11 includes an M2M / IoT device.

  The priority assigning unit 12 prioritizes the request with respect to the request queue when there are a plurality of requests before being added to the request queue with priority (hereinafter, simply referred to as a request queue). And give priority that is the order of addition.

FIG. 3 is a diagram showing an example of a frequency table used in the load control system according to the embodiment of the present invention in a table format.
The frequency table 13 can be realized by a storage medium such as a nonvolatile memory, for example, and associates the tree URL included in the received request, the resource operation type (CRUD / ALL), and the latest reception time (last_received_time) of the request. , Register as a request before adding to the request queue and manage this. The tree URL is identification information for identifying a resource to be operated.
Here, the type of operation is CRUD realized by the HTTP method, that is, one of Create (create-POST / PUT), Read (read-GET), Update (update-PUT), and Delete (delete-DELETE). And Hereinafter, a request in which any of the above CRUD operations is specified is referred to as a method-specific request.

The request last_received_time registered in the frequency table 13 is obtained when a new request for specifying the tree URL specified by the registered request, the same tree URL as the operation type, and the same operation type is received. It is updated at the reception time of this request.
In other words, the frequency table 13 manages to which tree URL the received request is, what operation is requested for the resource, and is related to various tree URLs and types of operations. Manage the reception time of the latest request.
The depth of the hierarchy above the wild card of the tree URL in the frequency table 13 can be set in advance. In the example shown in FIG. 3, the depth of this hierarchy is 2.

  Here, as newly received requests, for example, “Request1: reception time T1 CREATE incse / ae0 / container0 / contentInstance1111”, “Request2: reception time T2 UPDATE incse / ae0 / container1 / contentInstance2222”, “Request3: Three requests of reception time T3 DELETE incse / ae0 / container0 / contentInstance3333 are sequentially accepted, and the tree URLs “incse / ae0 / container0 ...” (Request1, Request3), “incse / ae0 / container1 ...” of each request. The update of the frequency table 13 in the case where the character string is common in the shallow layer (up to the second layer in this case) of (Request 2) and the character string is partially different in the deep layer (here, the third layer) will be described.

  As shown in FIG. 3, in the frequency table 13, the tree URL “incse / ae0 / *” having a depth of 2 above the wild card is managed, and each of the above requests Request1, Request2, and Request3 is sequentially received. In the frequency table 13, the last_received_time column in the row of the type ALL of the tree URL “incse / ae0 / *” corresponding to the second layer of the tree URL of each request in the frequency table 13 indicates the type of operation in the request. Regardless, it is updated sequentially at the reception time of the latest received request.

  In addition, the last_received_time column of each row for each method of the tree URL “incse / ae0 / *” in the frequency table 13 indicates an operation corresponding to each row for each method each time Request1, Request2, and Request3 are sequentially received. It is updated at the reception time of the request specifying the type.

Specifically, when the above Request1, Request2, and Request3 are sequentially received, the reception time T1 of Request1 is the last_received_time column in the type C row of the tree URL “incse / ae0 / *” on the frequency table 13. In addition, the reception time T2 of Request2 is the last_received_time column of the type U row of the same tree URL “incse / ae0 / *” on the frequency table 13, and the reception time T3 of Request3 is the same tree URL on the frequency table 13. This is reflected in the last_received_time column of the type D row of “incse / ae0 / *”.
Here, the last_received_time column corresponding to the row of the type ALL of the tree URL “incse / ae0 / *” in the frequency table 13 is updated to the reception time T3 of the last Request3.

Further, the frequency table 13 may be configured to manage a tree URL having a depth of 3 or more.
For example, a first tree URL “incse / ae0 / container0 / *” having a depth of 3 above the wild card and a second character string that is different from the first tree URL in the third hierarchy from the top. The tree URL “incse / ae0 / container1 / *” may be managed in the frequency table 13. In this case, registration information related to the type ALL and registration information for each method are managed individually for each tree URL.

When the above Request1, Request2, and Request3 are newly received sequentially, the Request1 reception time T1 is the last_received_time of the type C row of the first tree URL “incse / ae0 / container0 / *” on the frequency table 13. The reception time T3 of Request3 is reflected in the last_received_time column of the type D row of the same first tree URL “incse / ae0 / container0 / *” on the frequency table 13.
Here, in the frequency table 13, the last_received_time column corresponding to the row of the type ALL of the first tree URL “incse / ae0 / container0 / *” is the reception time T3 of Request3 received later among Request1 and Request3. Updated.

Further, the reception time T2 of Request2 is reflected in the last_received_time column of the type U row of the second tree URL “incse / ae0 / container1 / *” on the frequency table 13.
Here, the last_received_time column corresponding to the row of the type ALL of the second tree URL “incse / ae0 / container1 / *” in the frequency table 13 is updated to the reception time T2 of Request2.

The request queue management unit 14 manages the request queue for requests registered in the frequency table 13 and the order in which requests are added to the request queue.
The request execution unit 15 instructs the resource operation according to the request queue managed by the request queue management unit 14 and returns a response to the application communication unit 11.

The resource operation unit 16 operates resources according to instructions from the request execution unit 15.
The resource DB 17 includes, for example, a storage medium such as a nonvolatile memory, and stores information related to resources. This resource is operated by the resource operation unit 16 as described above.

  In the example illustrated in FIG. 1, the server 10 includes a request execution unit 15, a resource operation unit 16, and a resource DB 17. The configuration in which the request execution unit 15, the resource operation unit 16, and the resource DB 17 are provided one by one. did. However, the present invention is not limited to this, and the request execution unit 15, the resource operation unit 16, and the resource DB 17 are separated from the server 10, and a plurality of request execution units 15 and a plurality of resource operation units 16 corresponding to the request execution units 15 are provided. It may be provided.

  As a configuration, for example, a device including the request execution unit 15 and the resource operation unit 16 is configured as another application server different from the server 10, and this application server corresponds to the required number of request execution units 15 and resource operation units 16. There may be a configuration in which the number of servers 10 and each application server are connected to a common resource DB 17 via a network.

  In this way, the load on the resource operation can be distributed by dividing the instruction by the request execution unit 15 and the resource operation by the resource operation unit 16 in accordance with the instruction, for example, according to the type of request. it can.

Next, procedures of various processes by the load control system will be described.
Overall, the load control system adds requests to the request queue based on the tree URL unit and the frequency of requests for each operation, and sequentially retrieves requests from the request queue and processes responses.

FIG. 4 is a flowchart illustrating an example of a processing procedure from request reception to operation execution by the load control system according to the embodiment of the present invention.
First, the priority assigning unit 12 of the server 10 receives a new request received from the client by the application communication unit 11 (S1). The priority assigning unit 12 sets the time counted by the built-in timer at the time when the request is received as the current time when the new request is received (hereinafter sometimes referred to as the current time at the time of reception). To remember.
This request includes a tree URL and the type of operation requested. Hereinafter, a request in which any of the above CRUD operations is specified is referred to as a method-specific request.

  Here, past method-specific requests in which the same tree URL is specified in this method-specific request are registered in the frequency table 13 for each operation type in association with the latest reception time of this request. Suppose that

  In addition, it is assumed that a request specifying the same tree URL as the tree URL in which the request for each method to be registered is specified and specifying “ALL” as the operation type is registered in the frequency table 13. Hereinafter, this request is referred to as a request (ALL) and is distinguished from the above method-specific request. A request (ALL) and a method-specific request may be simply referred to as a request.

  This request (ALL) is a request regardless of the type of operation. The last_received_time in this request (ALL) is the latest last_received_time that is already registered in the frequency table 13 and is associated with various method-specific requests that specify the same tree URL as the tree URL specified by this request (ALL). is there.

  When a new method-specific request specifying the same tree URL as that specified by this request (ALL) is further registered in the frequency table 13, the last_received_time of this request (ALL) is further registered as described above. Updated to match last_received_time of request by method.

  Next, the priority assignment unit 12 performs priority assignment and order control processing for adding the request registered in the frequency table 13 to the request queue (S2). Details of the priority assignment and order control processing will be described later.

  Next, the request execution unit 15 performs resource operations on the requests added to the request queue in the order of the addition, and uses the result of the execution as a response to the client via the application communication unit 11. (S3).

Next, the details of the priority assignment and order control process, which is S2 described above, will be described.
FIG. 5 is a flowchart illustrating an example of a procedure of priority assignment and order control processing by the load control system according to the embodiment of the present invention.
First, the priority assigning unit 12 selects a request (ALL) that specifies the same tree URL as the tree URL specified by the method-specific request received in S1 from the frequency table 13, and uses the last_received_time specified by the request (ALL). Get some last_received_time (ALL).

In addition, the priority assigning unit 12 selects, from the frequency table 13, a past request for each method that specifies the tree URL specified by the method-specific request received in S1, the same tree URL as the operation type, and the same operation type, Last_received_time (by method) that is the last_received_time specified by this request is acquired (S21).
The built-in timer of the priority assignment unit 12 acquires the current time at the time of reception stored in the internal memory in S1 (S22).

Next, the priority assigning unit 12 designates a tree URL that is the same as the tree URL specified by the new request based on the current reception time acquired in S22 and the last_received_time (ALL) acquired in S21. A frequency value F_a relating to a request of any type is calculated according to the following equation (1).
The priority assigning unit 12 also uses the tree URL specified by the new request and the same type of operation based on the current reception time acquired in S22 and the last_received_time (for each method) acquired in S21. The frequency value F_m related to the request designating the operation type is calculated according to the following equation (2). The priority assigning unit 12 sets these calculated values in the internal memory (S23).

F_a = (current time at reception)-last_received_time (ALL) ... Formula (1)
F_m = (current time at reception)-last_received_time (by method)… Formula (2)
Next, the priority assigning unit 12 registers and updates the request received in S1 in the frequency table 13 (S24). This registered request includes the tree URL, the type of operation, and last_received_time as the current reception time stored in the internal memory in S1.

Here, as described above, the priority assigning unit 12 specifies the same tree URL as the tree URL specified by the newly registered method-specific request in the frequency table 13 and is already registered in the frequency table 13. Update the last_received_time of the existing request (ALL) to match the last_received_time of the reflected method-specific request.
Next, the priority assigning unit 12 sequentially adds the requests registered in the frequency table 13 to the request queue through a predetermined process (S25). Details of this processing will be described later.

  Next, a process for sequentially adding requests registered in the frequency table 13 to the request queue will be described.

FIG. 6 is a flowchart illustrating an example of a procedure for adding a request to a queue by the load control system according to the embodiment of the present invention.
First, when there is no request in the request queue (NO in S25-1), the priority assigning unit 12 sequentially puts the requests registered in the frequency table 13 into the request queue (S26-2). That is, the request registered in the frequency table 13 is immediately executed.

  Further, when there is a request in the request queue (YES in S25-1) and this request queue is full, that is, when the predetermined number has been reached (NO in S25-3), the priority assignment unit 12 Does not register a new request in the request queue, and after the server 10 sleeps, performs a process of holding for a predetermined time or refusing the request to re-evaluate whether or not the request queue is full Or other processing is performed (S25-4).

  If the request queue is not full (YES in S25-3), the priority assigning unit 12 is registered in the frequency table 13 and has a large frequency value among a plurality of requests (ALL) that specify different tree URLs. This request is added to the request queue so that various method-specific requests that specify the same tree URL as the tree URL specified by the request (ALL) from which F_a is calculated are executed first (S25-5). ).

  In addition, when the frequency value F_a is the same for a plurality of requests specifying different tree URLs registered in the frequency table 13 (YES in S25-6), the priority assigning unit 12 determines the frequency of the same value. Of the various method-specific requests that specify the same tree URL as the tree URL specified by multiple requests (ALL), the frequency value F_m calculated as described above for this request is large. This request is added to the request queue so that various method-specific requests are executed in order (S25-7).

  In addition, as described above, the frequency value F_a is the same for a plurality of requests specifying different tree URLs registered in the frequency table 13, and various methods related to the plurality of requests (ALL). When the frequency value F_m is the same for all the other requests (YES in S25-8), the priority assigning unit 12 has various types of old_last_received_time registered in the frequency table 13 among these requests. The request is added to the queue so that the method-specific requests are executed first, and various method-specific requests with the last_received_time itself later are executed (S25-9).

  As described above, when receiving a new request, the load control system according to the embodiment of the present invention processes this request in units of requests in the order of the latest request time, regardless of the type of request issuer. Therefore, the service can be continued while suppressing the malfunction of the entire platform. In this case, it is not necessary to modify the request issuing client.

  Also, there may be a plurality of request queues. In this case, for example, requests specifying a CUD operation in CURD are added to the first request queue in descending order of frequency value F_m, and a request specifying an R operation is a frequency. The prioritization method may be made different for each request queue by adding it to the second request queue in ascending order of the value F_m.

  In addition, the method described in each embodiment is, for example, a magnetic disk (floppy (registered trademark) disk, hard disk, etc.), optical disk (CD-ROM, etc.) as a program (software means) that can be executed by a computer (computer). It can be stored in a recording medium such as a DVD, MO, etc., semiconductor memory (ROM, RAM, flash memory, etc.), or transmitted and distributed by a communication medium. The program stored on the medium side includes a setting program that configures software means (including not only the execution program but also a table and data structure) in the computer. A computer that implements this apparatus reads a program recorded on a recording medium, constructs software means by a setting program as the case may be, and executes the above-described processing by controlling the operation by this software means. The recording medium referred to in this specification is not limited to distribution, but includes a storage medium such as a magnetic disk or a semiconductor memory provided in a computer or a device connected via a network.

  DESCRIPTION OF SYMBOLS 11 ... Application communication part, 12 ... Priority giving part, 13 ... Frequency table, 14 ... Request queue management part, 15 ... Request execution part, 16 ... Resource operation part, 17 ... Resource DB.

Claims (6)

A management unit that includes identification information for identifying an operation target resource, and that manages a request for requesting the operation of the resource and a latest reception time of the request as a request before being added to the request queue;
A receiving means for receiving a new request;
Calculation means for calculating a difference value that is a difference between the reception time of the received request and the latest reception time of the request that includes the same identification information as the identification information included in the request that is managed;
A designation unit that preferentially designates a request having a large difference value calculated by the calculation unit as a request to be added to the request queue among a plurality of requests that are managed and have different identification information. A characteristic load control device. The request includes the type of operation requested.
The calculating means includes
First calculation means for calculating a first difference value that is a difference between the reception time of the received request and the latest reception time of the request that includes the same identification information as the identification information included in the request. ,and,
The reception time of the received request and the latest reception time of the request that requests the same operation as the operation requested by the received request, including the same identification information that is managed and included in the received request. A second calculating means for calculating a second difference value that is a difference between
The designation means is:
For the plurality of requests with different identification information to be managed, specify the requests as priority requests to be added to the request queue in descending order of the first difference value calculated by the first calculation means,
When the first difference values calculated for the plurality of requests having different identification information are the same, the second difference values calculated by the second calculation unit for the plurality of requests are in descending order. The load control device according to claim 1, wherein the load control device is designated as a request to be added to the request queue with priority. The designation means is:
The calculated first difference value is the same for the plurality of requests having different identification information, and the calculated second difference value is the same for the plurality of requests having different identification information. In this case, the load control device according to claim 2, wherein the load control device is specified as a request to be added to the request queue with priority from the oldest reception time of the received request. The load control apparatus according to claim 1, wherein the request queue to which the received request is added is classified according to an operation type requested by the request. A method applied to a load control device,
Including identification information for identifying an operation target resource, and managing a request for requesting the operation of the resource and the latest reception time of the request as a request before being added to the request queue,
Receive a new request,
Calculating a difference value that is a difference between the reception time of the received request and the reception time of the request that includes the same identification information as the identification information included in the request that is managed;
A load control method comprising: specifying a request having a large calculated difference value as a request to be preferentially added to the request queue among a plurality of requests that are managed and have different identification information. A program used in a computer that operates as a part of the load control device according to claim 1,
The computer,
A load control processing program for causing the management unit, the receiving unit, the calculating unit, and the specifying unit to function.