JP2016149664A - Processing system and processor switching method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a virtualization circumstance processing system capable of preventing generation of hindrance at switching a processor with unfixed functions loaded by a virtualization.SOLUTION: A processing system 1000 has a control unit 7 that controls a request distribution apparatus 1, which distributes a piece of processing to either of an old transfer apparatus 2 and a new transfer apparatus 3, to distribute the processing so that the processing ratio of the new transfer apparatus 3 relative to that of the old transfer apparatus 2 gradually increases. When the resource utilization rate of the new transfer apparatus 3 exceeds a predetermined threshold level, the control unit controls the new transfer apparatus 3 to stop from increasing the processing ratio.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a processing device switching technique in a processing system including a plurality of processing devices (server, transfer device, etc.).

  Conventionally, in a system including a plurality of processing devices, it may be necessary to switch (replace) the processing device to a new processing device for maintenance reasons. Hereinafter, “old” is added to the name of the processing device before switching (for example, “old processing device”, “old transfer device”, etc.), and “new” is added to the processing device after switching. (For example, “new processing device”, “new transfer device”, etc.).

  As a scene, for example, a server system that processes a request from a client (client computer) can be considered. When switching any processing device (server, transfer device, etc.) in the server system, for example, if it is not a virtual environment, the performance required for each processing device can be easily identified. When switching to, the performance required for the new processing apparatus can be specified in advance and switched easily.

  Non-Patent Document 1 describes a device performance verification method in a virtual environment. As a processing device used in a virtual environment, for example, a so-called IA (Intel (registered trademark) Architecture) server is conceivable. The IA server is a server in which an Intel microprocessor or a compatible product is mounted on a CPU (Central Processing Unit). In a processing system using a plurality of processing devices such as an IA server, a plurality of virtualization functions (hereinafter simply referred to as “functions”) are often mounted on one processing device in a virtual environment.

Toru Miyahara, “Series: How to Design and Build a Virtual Environment That Does Not Fail” Pre-verification and Operation Management Important in Virtualization, [online], February 22, 2011, [Search February 3, 2015], Impress Corporation, Internet <URL: http://thinkit.co.jp/story/2011/02/22/2014/page/0/1>

  Even in a virtualized environment, the number of performance test patterns is limited for processing devices with fixed functions (number of functions, types and combinations are fixed, and so on). It is easy to specify the performance required for the apparatus and realize switching from the old processing apparatus to the new processing apparatus.

  However, in a virtualized environment, with regard to processing devices whose functions are not fixed by virtualization, the number of performance test patterns in advance becomes enormous, and the performance test specifies the performance required for the new processing device. It is not easy to switch from a new processing device to a new processing device. For example, in a processing system using a plurality of IA servers, many IA servers from different manufacturers are used, and the functions installed by virtualization in each IA server are not fixed, so it is necessary for a new IA server for switching. It is very difficult to specify the performance (throughput etc.) in advance.

  And when switching from the old processing device to the new processing device, if a load exceeding the performance is applied to the new processing device, the processing delay of the new processing device becomes a bottleneck and the operation of the entire processing system becomes slow. It will cause trouble.

  Therefore, an object of the present invention is to suppress the occurrence of trouble at the time of switching for a processing apparatus in which a function installed by virtualization is not fixed in a processing system in a virtual environment.

  In order to solve the above-described problems, the present invention provides a processing system using a processing device in which a function mounted by virtualization is not fixed in a virtual environment, and an old processing device that performs predetermined processing; A new processing device that newly takes over processing performed in the old processing device, a virtual switch that distributes processing to be performed to either the old processing device or the new processing device, and the old processing device for the virtual switch Control so that the processing ratio of the new processing device gradually increases with respect to the new processing device, and before the complete transition of processing from the old processing device to the new processing device, the resource usage rate of the new processing device is predetermined. And a control device that stops the increase in the processing ratio when the threshold value is exceeded.

  According to this, when the resource usage rate of the new processing device exceeds a predetermined threshold, the increase of the processing rate of the new processing device is stopped, so that a situation exceeding the performance on the new processing device is applied. It is possible to reliably avoid the occurrence of troubles at the time of switching.

  In the present invention, the old processing device and the new processing device are commonly assigned an IP (Internet Protocol) address and / or a MAC (Media Access Control) address used for data transmission and reception. However, it is preferable that one of the plurality of output ports is used for data transmission to the old processing device and the other one is used for data transmission to the new processing device.

  According to this, the IP address and / or the MAC address used for data transmission / reception can be commonly assigned to the old processing device and the new processing device, and the work accompanying the switching can be reduced.

  ADVANTAGE OF THE INVENTION According to this invention, trouble generation at the time of switching can be suppressed regarding the processing apparatus in which the function mounted by virtualization is not fixed in the processing system of a virtualization environment.

It is a whole block diagram of the processing system of this embodiment. (A) is a partial view of a processing system. (B) is a figure which shows the example of a flow table. (C) is a figure which shows the example of a group table. 1 is a partial view of a processing system. It is a flowchart which shows the process of a control apparatus. (A) is explanatory drawing at the time of using LB and a router as a comparative example. (B) is explanatory drawing at the time of using L2SW and VRRP as a comparative example.

  Embodiments of a processing system according to the present invention will be described below with reference to the drawings. In this embodiment, OpenFlow technology is used as a premise. OpenFlow is a technology that centrally manages devices in a network and performs transfer control using a single control device. Standard specifications have been established by the Open Networking Foundation, an industry group.

  As shown in FIG. 1, the processing system 1000 includes a request distribution device 1, an old transfer device 2, a new transfer device 3, a response distribution device 4, a server 5, a performance determination device 6, and a control device 7. Responds to requests from the client 100. Here, at least the old transfer device 2 and the new transfer device 3 are processing devices in which functions mounted by virtualization are not fixed (the number, type, and combination of functions are fixed) in a virtual environment.

Below, the case where it switches from the old transfer apparatus 2 which was operating originally to the new transfer apparatus 3 is demonstrated. The old transfer device 2 and the new transfer device 3 are, for example, a firewall, a load balancer, and the like, and have the same function such as function A. In FIG. 1, only “function A” is shown as a function, but actually, a plurality of functions are often installed.
The request distribution device 1 and the response distribution device 4 can be realized by an OpenFlow switch.
The control device 7 can be realized by an OpenFlow controller.

Next, an IP address, a MAC address, and the like that each device has will be described.
For the client 100, the IP address (global IP address) is “A”, and the MAC address (global MAC address) is “A”. These two “A” s are simplified symbols and do not mean that they are exactly the same as each other. The IP address “A” and the IP address “A” are the same, and the IP address “A” and the IP address “B” are different symbols (hereinafter the same).

  For the old transfer device 2 and the new transfer device 3, the IP address (global IP address) on the client 100 side is “B”, and the MAC address (global MAC address) is “B”. The IP address (local IP address) on the server 5 side is “C”, and the MAC address (local MAC address) is “C”. The old transfer device 2 and the new transfer device 3 use NAT (Network Address Translation) or NAPT (Network Address and Port Translation) to convert a global IP address and a local IP address, and a global MAC address and a local MAC address. Perform the conversion.

  The response distribution device 4 transfers the data received from the old transfer device 2 or the new transfer device 3 to the server 5, and responds to the response received from the server 5 by the old transfer device 2 or the new transfer device 3 (the transmission source and Return to the same).

  For the server 5, the IP address (local IP address) is set to “D”, the MAC address (local MAC address) is set to “D”, and the IP address (local IP address) of the initial setting (Default) GW (GateWay). Is “C”.

  When there is a request from the client 100, the request distribution device 1 distributes the request to the old transfer device 2 or the new transfer device 3. Hereinafter, assuming that the old transfer device 2 is assigned, the old transfer device 2 transfers the request to the server 5 via the response distribution device 4. The server 5 performs a predetermined process and returns a response to the response distribution device 4. The response is transferred to the old transfer device 2 and the request distribution device 1 and reaches the client 100.

The new transfer device 3 periodically sends resource (CPU, storage unit, etc.) usage rate information to the performance determination device 6.
The performance determination device 6 determines whether or not the resource usage rate of the new transfer device 3 has exceeded a predetermined threshold (for example, 90%). An instruction is given to increase the processing ratio of the new transfer apparatus 3 and if it exceeds, for example, the control apparatus 7 is notified that the ratio has been exceeded or the processing ratio of the new transfer apparatus 3 to the old transfer apparatus 2 Instruct.

The control device 7 includes a control unit 71 and a storage unit 72.
The storage unit 72 stores request distribution information 721 and response distribution information 722, and can be realized by, for example, a RAM, a ROM, an HDD, or the like.
The request distribution information 721 is used when the request distribution apparatus 1 distributes a request from the client 100 to either the old transfer apparatus 2 or the new transfer apparatus 3 (details will be described later in FIG. 2).
The response distribution information 722 is used when the response distribution device 4 distributes the response from the server 5 to either the old transfer device 2 or the new transfer device 3 (details will be described later in FIG. 3).

  The control unit 71 uses the OpenFlow protocol for the request distribution device 1 so that the processing ratio of the new transfer device 3 to the old transfer device 2 gradually increases based on an instruction from the performance determination device 6. If the resource usage rate of the new transfer device 3 exceeds a predetermined threshold before the complete transfer of processing from the old transfer device 2 to the new transfer device 3, the processing ratio is increased. Is stopped (details will be described later in FIG. 4).

  Next, referring to FIG. 2, even if the old transfer device 2 and the new transfer device 3 have the same IP address and MAC address, the request distribution device 1 sends a request from the client 100 to one of them. Explain the points that can be allocated appropriately.

  As shown in FIG. 2, the control device 7 includes a flow table 721 a (b) and a group table 721 b (c) as the request distribution information 721, and sets them in the request distribution device 1. (Provide). Then, when receiving a request from the client 100 at the port # 1, the request distribution apparatus 1 first refers to the flow table 721a and recognizes that the processing of the group a is performed because the reception port is “# 1”. Next, referring to the group table 721b, it recognizes that 95% of the request is transferred to the old transfer device 2 via the port # 2, and 5% is transferred to the old transfer device 2 via the port # 3. And process based on it.

  In this way, even if a common IP address and MAC address are assigned to the old transfer device 2 and the new transfer device 3, the request distribution device 1 can appropriately apply the request from the client 100 to any of them. Can be sorted. Therefore, it is not necessary to change the settings of the client 100, the old transfer device 2, etc., and the work involved in switching from the old transfer device 2 to the new transfer device 3 can be reduced.

  Next, referring to FIG. 3, even if the old transfer device 2 and the new transfer device 3 have the same IP address and MAC address, the response from the server 5 can be appropriately distributed to either of them. Will be described.

  As shown in FIG. 3, the control device 7 has a flow table 722 a as response distribution information 722, and sets (provides) it in the response distribution device 4. The flow table 722a is a conversion table related to IP addresses. Here, IP addresses “X” and “Y” that can be newly used are secured.

  Then, for example, the response distribution device 4 receives data D1 (source (SRC) IP address “C”, source MAC address “C”, destination (DST) IP address “D”, destination) at port # 1. When the MAC address “D” is received, the source IP address is rewritten from “C” to “X” with reference to the flow table 722a and sent to the server 5 as data D2.

  When the server 5 receives the data D2, the server 5 performs predetermined processing and sends the data D3 in which the transmission source and the transmission destination of the IP address and the MAC address are reversed to the response distribution device 4. The port used for data transmission / reception with the server 5 in the response distribution device 4 is “# 7”.

  Upon receiving the data D3, the response distribution device 4 refers to the flow table 722a, rewrites the destination IP address from “X” to “C”, and transfers the old transfer as data D4 via port # 1. Send to device 2. Here, the port “# 1” row in the flow table 722a means that the destination IP address is rewritten from “X” to “C”.

In this way, even if a common IP address and MAC address are assigned to the old transfer device 2 and the new transfer device 3, the response from the server 5 is appropriate for one of them (the one that issued the request). Can be sorted. Therefore, it is not necessary to change the setting of the server 5, and work associated with switching from the old transfer device 2 to the new transfer device 3 can be reduced. This also makes it possible to easily establish a connection between the old transfer device 2, the new transfer device 3, and the server 5. The flow table 722a is not limited to an IP address, and may be a TCP (Transmission Control Protocol) / UDP (User Datagram Protocol) port number.
The same processing is performed on data transmitted from the new transfer device 3 to the response distribution device 4 using the port # 6 of the response distribution device 4.

Next, the processing of the control device 7 will be described with reference to FIG.
After introducing the new transfer device 3 instead of the old transfer device 2 and starting to use them together, the control unit 71 of the control device 7 gradually increases the processing ratio of the new transfer device 3 to the old transfer device 2 ( Step S1).

Next, in step S2, the control unit 71 determines whether or not the processing ratio of the new transfer device 3 has reached 100%. If Yes, the process proceeds to step S3. If No, the process proceeds to step S4.
In step S3, the control unit 71 displays on the display unit (not shown in all drawings) that the transition (switching) from the old transfer device 2 to the new transfer device 3 has been completed, and ends the process.

  In step S4, the control unit 71 determines whether or not the resource usage rate of the new transfer device 3 exceeds a predetermined threshold (for example, 90%) based on an instruction from the performance determination device 6. Proceed to S5, and if No, return to Step S1.

  In step S5, the control unit 71 displays a warning to the effect that the resource usage rate of the new transfer device 3 has exceeded a predetermined threshold on the display unit (not shown in all drawings), and ends the process. That is, when the resource usage rate of the new transfer device 3 exceeds a predetermined threshold (Yes in step S4), the processing rate of the new transfer device 3 does not increase any more because the process does not return to step S1. If the control unit 71 receives an instruction from the performance determination device 6 regarding a decrease in the processing ratio of the new transfer device 3, the control unit 71 executes the instruction.

  As described above, according to the processing system 1000 of the present embodiment, a request distribution device that is an OpenFlow switch with respect to the old transfer device 2 and the new transfer device 3 in which the functions mounted by virtualization are not fixed in a virtual environment. 1 and the response distribution device 4 are used to control the distribution so that the processing ratio of the new transfer device 3 to the old transfer device 2 gradually increases, and complete processing from the old transfer device 2 to the new transfer device 3 If the resource usage rate of the new transfer device 3 exceeds a predetermined threshold before the transition, it is possible to suppress the trouble at the time of switching by stopping the increase in the processing ratio.

  Further, the request distribution device 1 and the response distribution device 4 which are OpenFlow switches use one of their own output ports for data transmission to the old transfer device 2 and the other one as a new transfer device. 3 can be used in common to assign an IP address and / or a MAC address used for data transmission / reception to the old transfer device 2 and the new transfer device 3. Work associated with switching to the device 3 can be reduced.

(Comparative example)
Next, a comparative example will be described. As shown in FIG. 5A, when a load balancer (LB) is used instead of an OpenFlow switch as a request distribution device and a router instead of an OpenFlow switch is used as a response distribution device, the old transfer device and the new transfer device The IP address and the MAC address must be made separate (B1 and B2, C1 and C2), and the work accompanying switching from the old transfer device to the new transfer device increases.

  Further, as shown in FIG. 5B, when the request distribution device and the response distribution device use L2SW instead of the OpenFlow switch, the IP address and MAC address of the old transfer device and the new transfer device can be made common. Therefore, it becomes necessary to add a function of VRRP (Virtual Router Redundancy Protocol) to the old transfer device and the new transfer device, and the work accompanying switching from the old transfer device to the new transfer device also increases.

Although description of this embodiment is finished above, the aspect of the present invention is not limited to these.
For example, the target to be switched is not limited to the transfer device, but may be another processing device such as an IA server.

Further, the request distribution device 1 and the response distribution device 4 are not limited to OpenFlow switches, and can be realized by general virtual switches that can perform the same operation.
In addition, specific configurations and processes can be appropriately changed without departing from the gist of the present invention.

1 Request distribution device (virtual switch)
2 Old transfer device (processing device)
3 New transfer equipment (processing equipment)
4 Response distribution device (virtual switch)
DESCRIPTION OF SYMBOLS 5 Server 6 Performance determination apparatus 7 Control apparatus 71 Control part 72 Memory | storage part 100 Client 721 Request distribution information 721a Flow table 721b Group table 722 Response distribution information 722a Flow table 1000 Processing system

Claims (4)

In a virtualized environment, a processing system using a processing device in which functions mounted by virtualization are not fixed,
An old processing device for performing predetermined processing;
A new processing device that newly takes over the processing performed in the old processing device;
A virtual switch that distributes the processing to be performed to either the old processing device or the new processing device;
The virtual switch is controlled so that the processing ratio of the new processing device to the old processing device is gradually increased, and before the complete transition of processing from the old processing device to the new processing device. A control device that stops the increase in the processing ratio when the resource usage rate of the new processing device exceeds a predetermined threshold;
A processing system comprising: The old processing device and the new processing device are commonly assigned an IP (Internet Protocol) address and / or a MAC (Media Access Control) address used for data transmission / reception,
The virtual switch uses one of its plurality of output ports for data transmission to the old processing device and the other one for data transmission to the new processing device. The processing system according to 1. In a virtual environment, a processing device switching method by a processing system using a processing device in which the function mounted by virtualization is not fixed,
The processing system includes:
An old processing device for performing predetermined processing;
A new processing device that newly takes over the processing performed in the old processing device;
A virtual switch that distributes the processing to be performed to either the old processing device or the new processing device;
A control device, and
The controller is
The virtual switch is controlled so that the processing ratio of the new processing device to the old processing device is gradually increased, and before the complete transition of processing from the old processing device to the new processing device. When the resource usage rate of the new processing device exceeds a predetermined threshold value, the increase in the processing rate is stopped. The old processing device and the new processing device are commonly assigned an IP address and / or a MAC address used for data transmission / reception,
The virtual switch uses one of its plurality of output ports for data transmission to the old processing device and the other one for data transmission to the new processing device. 4. The processing apparatus switching method according to 3.

Images