JP5947974B2 - Information processing apparatus, information processing apparatus exchange support system, and exchange support method - Google Patents

Description

  The present invention relates to an information processing apparatus, an information processing apparatus replacement support system, and a replacement method, and more particularly to an information processing apparatus that includes a storage and is used as a distributed cloud that provides an application via a network. The present invention relates to a support system and method for exchanging devices.

  Information processing apparatuses such as servers that require high availability have a redundant system configuration and perform regular maintenance in preparation for failures or failures due to external factors. One of the maintenance tasks is the work of replacing a failed part or the entire device. Since the storage composed of hard disks has a mechanical mechanism, a failure occurrence rate is high, and various technologies for increasing availability such as RAID (Redundant Arrays of Independent Disks) are known. Therefore, the storage hard disk is replaced relatively frequently, and the entire apparatus may be replaced in addition to RAID reconfiguration in units of parts. In the exchange of a device having a storage, it is necessary to copy the storage data in the device to the storage of the new device.

  There are two types of storage copy methods. One is a method of copying at a remote location, and the other is a method of bringing a new device to a customer and copying it.

  An example of a method of copying at a remote location is shown in Patent Document 1. According to the invention of Patent Document 1, the storage system includes a primary storage system at a user site, a secondary storage system via a network, and a replacement storage system at another site via the network. When the user site shuts down, copy from the primary storage to the secondary storage, make the contents identical, and if the primary storage failure is found, copy the secondary storage data to the replacement storage in another location How to do is shown. By copying all normal shutdown disk images to the replacement storage remotely, for example, when a failure occurs in the primary storage, an alternative storage is prepared at night, and the primary storage is replaced the next morning. Yes.

  An example of a method of copying at a customer site is shown in Patent Document 2. In the invention of Patent Document 2, when it is determined that the storage controller has failed and the device needs to be replaced, the new device accesses the storage of the original device and continues the service, while the original device is in the background. Shows how to copy all of the storage to a new device. When copying of the new device to storage is complete, the replacement is completed by disconnecting the original device.

JP 2006-293447 A JP 2008-97318 A

  The form of computing may be determined depending on the performance and cost of the external network. The external network is composed of a core network of the backbone network and an access network of the edge network. With the spread of optical fiber in the core network close to the data center side, the bandwidth of the backbone network increases, so the form of cloud computing is generally used for computing resources such as servers and storage via the network It came to be used. Cloud computing has a centralized architecture that concentrates data and processing in a data center (core data center: CDC). Virtualize large-scale computing resources and flexibly provide the necessary amount of resources as a resource pool. Furthermore, there is an advantage that maintenance cost can be reduced by centralized management. In centralized cloud computing, when a smartphone is used as an example of a mobile terminal, all data used by the smartphone is collected and processed in a data center. In a centralized cloud, resources such as servers and storage are concentrated in one place, so that maintenance operation management can be concentrated in one place. For example, when replacing a single storage device or device, the maintenance staff does not need to move because there is only one base.

  However, in recent years, communication with data centers has become tight as the functions of mobile terminals such as smartphones increase and the number of sensor devices installed increases. In addition, the infrastructure development of broadband wireless communication means such as a WiFi router is progressing, and the broadband access network on the edge side close to the portable terminal is progressing. Therefore, distributed cloud computing is being studied in which data is not concentrated in a data center but is widely distributed in the vicinity of mobile terminals to process data.

  There are differences in the network between the two methods of copying the above storage, that is, remote copy or customer copy. The remote copy method has a network delay due to a long distance via a public line, while the copy method at the customer site uses a closed network and has no network delay compared to an old machine. . For this reason, as a replacement method with continued service, copying at a customer site is common. This is because a closed network has two advantages: high security and no additional network delay added to service responsiveness.

  When exchanging information processing devices in a distributed cloud environment, the replacement method for copying at the customer site starts after the information processing device is installed, so if the storage capacity is large, the copy time is long and the delivery person installs it. There is a problem that an old device cannot be taken home in a short time later. As a result, it is necessary to visit the customer twice for installation and withdrawal, and the maintenance costs associated with delivery are high in places with poor transportation. In addition, if the information processing apparatus is placed in the restricted access area at the customer site, it is assumed that the customer is present during the exchange, and the burden on the customer site cannot be ignored.

  On the other hand, the method of copying at a remote place has the advantage that the stay at the customer site is short and the installation and withdrawal can be performed once because parts or information processing devices for which storage copying has been completed in advance are delivered. . However, in the method of copying the snapshot image that was shut down normally to the replacement storage, the storage contents are old in time and the storage contents are restored to the old storage time. There is a problem that the service cannot be performed.

  In the inventions described in Patent Document 1 and Patent Document 2, both storage copies are performed in one stage, and all contents are copied. For this reason, when it is assumed that the information processing apparatus is exchanged in a distributed cloud environment, the above-described problems of copying at a customer site and copying at a remote place cannot be solved.

Therefore, in the present invention, when replacing the entire information processing apparatus,
(1) Make the stay at the customer site as short as possible and install and remove in one time.
(2) An object of the present invention is to provide an exchange method that ensures both service responsiveness and continuity, and that the service never stops, and a system that realizes it.

An example of a representative configuration of the present invention is an information processing apparatus used as a distributed cloud that provides an application via a network, and the information processing apparatus is a central data center. Is an information processing apparatus that constitutes a small data center in a distributed cloud environment that processes data in cooperation with the information processing apparatus, and the information processing apparatus includes at least one information processing unit having a controller and a storage, A file system having a pair of primary files and secondary files in file units or directory units, and the controller has a file system management unit for managing the file system, and the file system management unit is made redundant Management table, update management table, A line replaceable trigger unit, and a off-line exchange triggering unit, the exchange of the second information processing unit and the first information processing unit having the same function in those carried out online and offline, the online replacement trigger unit And the off-line exchange trigger unit has a function of controlling an on-line exchange trigger for performing on-line or off-line exchange of the first information processing unit and the second information processing unit, an off-line exchange trigger, and the redundancy The correspondence management table manages the correspondence between the primary file and the secondary file of the file system , writes to the corresponding secondary file when the primary file is written, and the update management table stores the offline exchange It said by the trigger part o After line replaceable trigger is set, the has a function of managing a file system information about updating, the file system management unit, with the set of online replacement trigger by the online replacement trigger unit, the update management table the basis of the information, the basis of the file system information on updating after the offline exchange trigger is set, only the file system updated the primary file in the first information processing unit, the second It has a function of copying to the information processing unit.

  According to the present invention, in an information processing device used as a distributed cloud that provides an application via a network, when the information processing device is replaced, the storage is copied in advance on the network at a remote location, After transporting the device to the device installation location, the replacement time at the device installation location is reduced by copying only some primary files without copying all the storage differences between the old and new devices during the transport to the new device. Moreover, the apparatus can be replaced without stopping the application.

It is a figure which shows the structural example of (micro | micron | mu) DC based on the 1st Example of this invention. It is a figure which shows the structure of a controller (CTR) in a 1st Example. It is a figure which shows the example of execution of the application in distributed cloud computing. It is an example of the outline | summary of replacement | exchange of information processing apparatus in a 1st Example. It is a time chart explaining the example of the operation | movement of replacement | exchange of information processing apparatus based on a 1st Example. FIG. 5 is a diagram illustrating a relationship between old and new μDCs corresponding to FIG. 4. It is a table which shows the necessity of file copy when the information processing apparatus arrives at the spot. It is an example of the block diagram of μDC concerning the 2nd example of the present invention. It is a time chart explaining the example of operation | movement of replacement | exchange of information processing apparatus based on a 2nd Example. It is a figure which shows the structural example of (micro | micron | mu) DC concerning the 3rd Example of this invention. It is a figure which shows the structural example of (micro | micron | mu) DC based on the 4th Example of this invention. It is a time chart explaining the example of operation | movement at the time of the failure | occurrence | production occurrence of several information processing unit (micro | micron | mu) DC based on a 4th Example.

  In the present invention, in order to process data near the terminal (site) without consolidating data in the data center, a plurality of small data centers installed in the vicinity of the terminal and the central data center process the data. Adopt a wide-area distributed cloud. In this decentralized cloud, storage units and devices are distributed, and when replacing, it is necessary for the delivery staff or maintenance staff to deliver to the installation locations of the micro DCs that are distant from each other and replace them. is there. Here, in the present invention, a place where the micro DC is arranged in the vicinity of the terminal is particularly referred to as “on-site”.

  The point of the present invention is that when exchanging the information processing apparatus, a large part of the storage of the old apparatus is copied in advance to the storage of the new apparatus in the data center in advance by remote copy, and the old apparatus is based on the information in the update management table in the field. Is to copy only a part of it to the storage of the new device. In other words, the storage copy at the time of device replacement is made in two stages, and the second copy is made by copying only a part, thereby realizing a reduction in work time at the site. For example, only a part of the difference between the primary storage and the alternative storage generated when the storage of the new device is transported to the customer is copied on site. It also ensures continuity of service to users.

As an example of a specific configuration as a system for solving the above problems,
When the primary file is written, the secondary file is written, the redundancy management table that manages the correspondence between the primary and secondary files, and the update management table that manages whether the file has changed In an information processing apparatus that provides an application via a network, when exchanging from the first information processing apparatus to the second information processing apparatus, the primary file and the secondary file of the first information processing apparatus are The first copy procedure for copying to the second information processing device, the start procedure indicating the start of the update management table after the first copy procedure, and the update management table of the first information processing device are dirty and redundantly managed. The second copy procedure to copy the file whose table is primary to the second computer And after the replacement from the first information processing device to the second information processing device, the update information table of the first information processing device is dirty and the redundancy management table is the secondary file. A third copy procedure for copying from the primary file of the processing apparatus;
As another system configuration example, the information processing apparatus has a storage copy priority table defined by the first information processing apparatus instead of the update management table, and is stored in the second copy procedure. Referring to the copy priority table, change the copy order.

  Furthermore, as another system configuration example, when a file is written to the primary, the file is written to the secondary, the redundancy management table that manages the correspondence between the primary and secondary in units of files, and the file in units of files In an information processing apparatus having an update management table for managing whether there is a change and providing an application via a network, the information processing apparatus includes an information processing unit including a CPU and a storage, and a plurality of information processing units A second computer having a secondary file of the first information processing unit when a failure occurs in the first information processing unit and the first information processing unit is not used. In the information processing unit, the redundancy management table of the second information processing unit Procedure for changing secondary bit to primary bit, procedure for registering secondary file in redundancy management table of third information processing unit after starting update of primary file in second information processing unit, and second information A procedure for copying the secondary file from the primary file of the processing unit to the third information processing unit;

  DESCRIPTION OF EMBODIMENTS Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the drawings. Note that components having the same function are denoted by the same names and reference symbols throughout the drawings for describing the embodiments for carrying out the invention, and the repetitive description thereof will be omitted.

  As a first embodiment of the present invention, an exchange support system that supports exchange of information processing apparatuses in a distributed cloud environment will be described. That is, it is a form of a wide-area distributed cloud in which a plurality of small data centers (micro DCs) installed in the vicinity of a terminal and a central data center (CDC) process data in cooperation.

FIG. 1A and FIG. 1B show the main part of an exchange support system for supporting exchange of information processing apparatuses in a distributed cloud environment according to the first embodiment.
FIG. 1A is an example of a configuration diagram of a μDC capable of speeding up replacement of an information processing apparatus according to the first embodiment of the present invention.

  The micro DC (μDC) of this embodiment has a mechanism / function for reducing storage copy by paying attention to redundancy in order to perform the minimum necessary copy in the field.

  Reference numeral 100 denotes a μDC 1 before replacement installed in the field, and reference numeral 200 denotes a replacement μDC 2, both of which have the same function. However, the new μDC2 for replacement may have a new function in addition to the functions such as μDC1. μDC1 and μDC2 are connected by a network 300.

  The μDC1 (100) includes a controller (CTR) 120 using a server or the like, a storage (STR) 130, and a router (RTR) 140. Hereinafter, a functional unit having a controller (CTR) and a storage is referred to as an information processing unit (UNT) 110. However, the information processing units 110 may be physically configured on the same base, or may be configured on different bases. The controller (CTR) 120 includes a memory (MEM) 121, a CPU 124, a network interface (NIC) 125 that communicates with a router, and a host bus adapter (HBA) 126 that communicates with a storage. The memory 121 includes a file system management unit (FSM) 122 that manages a file system and a buffer (BUF) 123 that temporarily holds inputs from sensors and portable terminals. The file system management unit (FSM) 122 associates each file with an I node (INODE).

  Similarly, the μDC2 (200) includes an information processing unit 210, CTR220, MEM221, CPU224, NIC225, HBA226, STR230, and RTR24.

  The storages 130 and 230 may be built in the CTRs 120 and 220 or may be installed independently. The interfaces of the storages 130 and 230 are not particularly limited to SATA (Serial Advanced Technology Attachment), SAS (Serial Attached SCSI), iSCSI (Internet SCSI), or Fiber Channel (FC). The device may be either HDD or SSD. The storages 130 and 230 are provided with a logical unit LU (Logical Unit) as one or a plurality of logical storage areas on a physical recording area. A plurality of files are provided in the LU. FIG. 1A shows an example in which there are six files in one LU 1 of the storage 130.

  In FIG. 1A, for simplicity of explanation, a single information processing unit is used, but a plurality of information processing units may be used. In order to reduce costs, the storage does not use RAID. This is because, when there are a plurality of storage units composed of small-capacity HDDs, providing each controller with a RAID controller increases the cost. Therefore, the μDC (100, 200) does not provide a RAID controller in each information processing unit, and the CPUs 124, 224 write data in two locations, primary and secondary. The write command is not limited, but, for example, Amazon S3 (Simple Storage Service) REST API is used.

  In the present invention, the unit of redundancy is set to a fine granularity as to how the primary and secondary are provided. When a sensor or mobile terminal moves, it connects to a nearby μDC at the destination, so that the micro DC to be connected changes for each user and application, and accordingly, a file or directory that is a component of the application We decided to provide primary and secondary in units (hereinafter, file units).

  In FIG. 1A, in order to set primary and secondary in units of files, a redundancy management table (RMT) 1222 that newly associates the file system management unit (FSM) 122 with the INODE 1221 and newly represents the primary and secondary for each file. Is provided. As an example of the redundancy management table (RMT) 1222, pairs of INODE 1 and 2, 3 and 4, 5 and 6 indicate primary (P) and secondary (S) files, respectively. The redundancy management table (RMT) 1222 indicates whether each file is primary or secondary, and which INODE 1221 is a pair. The primary and secondary are physically separate devices, and the reliability at the time of failure is improved. Therefore, the primary and secondary devices are used as separate devices for operation, but are not limited.

  Further, in order to manage the reduction of storage copy in the field, an offline difference management table (ODMT) 1223 as an update management table, which is a feature technique of the present invention, an online exchange trigger (NTRG) unit 1224, an offline exchange trigger An (FTRG) unit 1225 is provided.

  FIG. 1B is a diagram showing a configuration of the controller (CTR) 120 in FIG. 1A. The CTR includes a μDC monitoring unit 1211, a redundancy management unit 1212, an application management unit 123, a μDC switching control unit 1214, and an offline management unit 1215 as main functions realized by executing a computer program on the CPU. I have. Reference numeral 127 denotes an interface. The μDC monitoring unit 1211 monitors the state of each μDC. When the number of failure occurrences of a node exceeds a certain number, if it is determined that the μDC itself needs to be replaced, it makes a replacement request to the maintenance station. The μDC has a redundant configuration internally, and continues service by degenerating even when a node failure occurs. The redundancy management unit 1212 manages the redundant configuration of the μDC. The application management unit 123 manages an application executed in its own μDC as a service for a nearby terminal. The μDC switching control unit 1214 performs control when replacing the old and new μDC. The offline management unit 1215 performs control when the μDC is offline when the μDC is replaced.

  The first embodiment of the present invention assumes a computing environment that processes data in the vicinity of a terminal as shown in FIG. That is, this example is a wide-area distributed cloud form in which a plurality of μDCs 100 and 200 installed in the vicinity of a terminal and a data center (CDC) 400 are connected by a network 300 and process data in cooperation. The CDC 400 includes an application manager (APMGM) 410, and each μDC 100, 200 processes data of a terminal 500 in the vicinity thereof.

  A small data center (CDC) 400 includes a network, a CTR, and a storage, and can communicate with an external network, and thus has a small data center function. The μDCs 100 and 200 operate as application CTRs and provide applications via the network 300 from a mobile terminal 500 such as a smartphone. For this reason, it takes a service form as SaaS (Software as a Service). In order to allow each user of the portable terminal 500 to access the nearest micro DC in the most location, the data center includes an APMGM 410 and manages which application is distributed to which micro DC. The mobile terminal 500 inquires a priority list of μDCs to be connected from the data center, and connects to nearby μDCs according to the list. Not only is the response speed of the mobile terminal fast, but also reduces the communication cost because the data communication volume of the core network reaching the data center is reduced by processing as close as possible and sending metadata to the data center. Can do.

  FIG. 3 shows a conceptual diagram of information processing apparatus replacement in the distributed cloud environment in the first embodiment.

  This example stores a central data center (CDC) 400 that performs application deployment and μDC management, a micro DC 100A that is used in the vicinity of the terminal in the field 700, and a micro DC 100B that is used for replacement. Maintenance station 600. Each μDC has a redundant configuration, and even if a failure occurs on the one hand, processing can be continued on the other hand. Here, an example in the case of exchanging the micro DC device at a regular or predetermined date and time is shown. Further, transportation means such as a truck is used for the transportation means 900 of μDC between the maintenance station 600 and the site 700.

  The two μDCs to be exchanged are physically different but functionally the same. Therefore, in the following, for convenience of explanation, the micro DC 100A being used is denoted as μDC 1 or μDC 1-Old, and the new micro DC 100B for replacement is denoted as μDC 2 or μDC 1-New.

  Next, based on FIG. 4, the operation | movement of replacement | exchange of information processing apparatus based on a 1st Example is demonstrated. FIG. 4 is a time chart for explaining an example of the switching operation by the controller (CTR) 120 for exchanging the device (μDC).

  The components for exchanging the equipment are μDC1-Old, which is a micro DC before the replacement at the site 700, μDC1-New, a new micro DC of the maintenance station 600 having functions and performance equivalent to this μDC1-Old, the center A data center (CDC) 400 of a mobile phone, and a terminal user (users) 500 such as a smartphone. There are several possible configurations of the micro DC, but the simplest form includes at least one node having a CTR (server) and storage, and an IP address as an identifier of the node.

  Hereinafter, it demonstrates along the time chart of FIG. In FIG. 4, (1) to (9) indicate main processing for device replacement.

  (1) A new μDC2 of the maintenance station 600 notifies the data center (CDC) 400 of an IP address as node information (time t0), and makes an exchange mode request with the exchange target μDC1. In the CDC 400, the old and new μDC1 and μDC2 are associated with each other.

  The node correspondence list is notified from the CDC to the μDC1 and μDC2.

  (2) The maintenance station 600 remotely copies the storage of μDC1 to μDC2. The remote copy performs secure communication via the Internet by using a known technology such as VPN (Virtual Private Network).

  (3) From the CDC 400, deploy the same application deployed on the node of μDC1 to μDC2.

  (4) When the application deployment and the storage copy are completed, the offline exchange trigger (FTRG) unit 1225 sets an offline exchange trigger for difference detection from μDC2 to μDC1 (time t1). When the off-line exchange trigger is set, the μDC 1 records whether or not there has been writing to the storage in units of files. Then, the replacement μDC 2 is physically transported to the site 700. μDC2 is transported using physical transportation such as trucks.

(5) After arriving at the site 700 (time t2), the node information (IP address) of μDC2 is notified to the CDC 400, and the CDC 400 performs node association between μDC1 and μDC2.
Those that do not change are C bits (Clean: 0).

(6) The online exchange trigger (NTRG) unit 1224 sets the online exchange trigger to the μDC1 and stops the application of the μDC1. Then, only the dirty primary file that is a part of the files in which the difference is detected in the above (4) is copied from the μDC1 to the μDC2. By limiting the copy to a part, the copy time at the site 700 can be reduced. The copy target will be described later.
Thereafter, the application of μDC2 is activated.

(7) While copying the storage difference at the site 700, the input to the μDC1 is accumulated in the buffer.
After the copy of (6) above, the input buffer is copied to μDC2, the primary device at the site is set to μDC2, and the service to the user 500 is started. Upon completion of the exchange (time t3), the primary device is notified to the CDC 400 as μDC2. In this way, the service to the user 500 is temporarily interrupted for a very short time associated with switching between the old and new μDCs.

  (8) The CDC 400 notifies the user 500 to change the connection destination from μDC1 to μDC2. The user 500 changes the connection destination to μDC2. The input is switched to μDC2, and the service is continued by processing sequentially from the input buffer (7).

  (9) After completion of the replacement, in the background, the remaining part that is the difference in storage and not copied in (6) is copied.

  According to the present embodiment, as shown in the flow of FIG. 4, in order to replace the μDC periodically or at a fixed date and time, remote copy of the storage and deployment of the application to the same device are performed in advance at the maintenance station, and then to the site. Make a minimum storage copy after transportation and continue the service with a new μDC. Since the μDC for replacement is prepared and delivered in advance, it is only necessary to go to the site only once and the staying time is short. Regarding the file in which the difference is detected, the storage copy is performed twice, but the target of the storage copy differs between the first time and the second time, which is a feature of the first embodiment of the present invention.

  This embodiment has a mechanism for performing storage copy reduction by paying attention to redundancy in order to perform the minimum necessary copy at the site 700 and speed up the replacement of the apparatus.

  Here, the redundancy of the micro DC will be described. FIG. 5 is a diagram showing the relationship between old and new μDCs corresponding to FIG.

  When the μDC2 is set offline in (4) of the exchange flow in FIG. 4 and the offline exchange trigger (FTRG) of the μDC1 is set, only the file updated in the μDC1 offline difference management table (ODMT) in FIG. (Dirty: 1) is set ((3) and (4) of the ODMT in FIG. 5), and the C bits (Clean: 0) are not changed (ODMT (1) and (2) in FIG. 5). . Next, μDC2 is transported from the maintenance station to the site. When the μDC2 goes to the site and the online exchange trigger (NTRG) of the μDC1 is set in (6) of the exchange flow in FIG. 4, the necessary files are copied after copying to the μDC2 of the file system management unit (FSM). .

  In FIG. 5, only the file having the D bit in the offline difference management table (ODMT) and the P bit in the redundancy management table (RMT) (ODMT (3) in FIG. 5) is copied from μDC1 to μDC2. A file whose offline difference management table (ODMT) is D bits and whose redundancy management table (RMT) is S bits (ODMT (4) in FIG. 5) is not copied.

  In the example of FIG. 5, the file 5 in the LU is copied from μDC1 to μDC2, but the file 6 in the LU is not copied. Since the file 6 has the same content as the file 5, the file 6 is generated by copying from the file 5 of the μDC 2 after exchanging the μDC 1 and the μDC 2. Since the offline difference management table (ODMT) 123 is a C-bit file (ODMT (1) and (2) in FIG. 5) is not updated, there is no need to copy the file.

  As described above, when a new file is generated after the offline exchange trigger (FTRG) of μDC1 is set, the primary and secondary of the redundancy management table (RMT) are added corresponding to the INODE of the new file. The D bit is set in the offline difference management table (ODMT) corresponding to the corresponding INODE. When the online exchange trigger (NTRG) is set, only the primary file of the redundancy management table (RMT) among the new files is copied from μDC1 to μDC2. The secondary file is copied from the primary file in the μDC 2 after the exchange.

  A table 1300 in FIG. 6 indicates whether or not file copying is required upon arrival at the site. In FIG. 6, file 1 of case 1 has RMT as primary (P) and ODMT as Clean (0). In this case, when it arrives at the site, it is not copied to μDC2. In file 2 of case 2, RMT is secondary (S) and ODMT is Clean (0), and in this case also, it is not copied to μDC2. The file 5 of case 3 is RMT is primary (P) and ODMT is Dirty (1). In this case, the file 5 is copied to the μDC 2 upon arrival at the site. The file 6 of the case 4 is RMT secondary (S) and ODMT Dirty (1). In this case, the file 6 is not copied to the μDC 2 upon arrival at the site. That is, only the file 5 of the case 3 needs to be copied at the site.

  In FIG. 1A, since each output of the online exchange trigger (NTRG) unit 1224 and the offline exchange trigger (FTRG) unit 1225 is in a format representing a state, it is necessary to see the change point on the software, but represents the trigger. The format, that is, 1 may be written to return to 0, and the implementation is not limited as long as the trigger can be determined.

  When the trigger is set by the online exchange trigger (NTRG) unit 1224 in (6) of the exchange flow in FIG. 4, the copy service starts from μDC1 to μDC2, and at the same time, the application service stops and the input buffer (BUF) in the memory Session information and input data are stored. Along with this, response processing is performed so that the user side does not hang up. For example, when the user is a smartphone, the session is maintained, and a message indicating that it is temporarily interrupted is responded as a Web screen so that input from the user is not performed. In the case where data is unilaterally transmitted to the CTR side such as a sensor, no response is required. After the storage copy to the μDC2 is completed, the copy is made from the input buffer (BUF) of the μDC1 to the input buffer of the μDC2. The service to the user 500 is temporarily interrupted only for a short time associated with the switching between the old and new μDC, but the input from the user is accumulated in the buffer of the μDC1. The μDC2 processes the input from the buffer first after the storage and the buffer copy. As described above, after the user switches the connection to the μDC2, the application is resumed, and the input data is processed from the input buffer (BUF) of the μDC2, thereby continuing the application without losing the input data during the service interruption. can do.

  Normally, when a file is updated, that is, if there is a Dirty file in the offline difference management table (ODMT), it is natural to copy all the files. However, in this embodiment, even if it is Dirty, the secondary file can be generated autonomously from the primary, and therefore, copying at the time of replacement is not performed. As a result, there is an advantage that the time required for storage copy at the site at the time of replacement can be greatly reduced, and the maintenance cost can be reduced by reducing the stay time and the number of stays at the site. In addition, according to the present embodiment, since the work time at the site is shortened, it is easy to retain user input data when exchanging μDC, and there is an advantage that service for the user can be continued without stopping the application. .

  Next, FIG. 7 shows a configuration example of the μDC (100A, 100B) according to the second embodiment of the present invention. In this example, a management table indicating the priority of storage copy, which is an extension of the offline difference management table (ODMT) of the first embodiment, is used as the update management table. That is, in the second embodiment, a priority management table (PMT) 1226 is provided instead of the offline difference management table (ODMT). The storage copy priority is used to give priority to the storage copy at the time of replacement. In FIG. 7, the priority is a number from 1 to 8, but is not particularly limited. The priority is determined by the file system management unit. For example, the priority of a file that has not been accessed for a certain period of time or an older version file is lowered.

  The operation of the second embodiment will be described below along the time chart of FIG. The difference from the time chart of FIG. 4 is that the priority is set in units of files after the offline exchange trigger is set (time t1). The priority may be set in the field after the offline exchange trigger is set, or may be defined and automatically processed before the offline exchange trigger is set. Furthermore, an online exchange trigger is set in μDC1, and only the set high priority file is copied from μDC1 to μDC2.

  Even in this example, the copy time at the site 700 can be reduced by limiting the copy to a part. After changing the connection destination from μDC1 to μDC2, a low priority file is copied from μDC1 to μDC2 in the background.

  That is, in the present embodiment as well as the first embodiment, it is possible to reduce the time of storage copy at the site at the time of replacement, and there is an advantage that the service for the user can be continued without stopping the application.

In the first embodiment and the second embodiment, for the sake of simplification, an example in which the number of information processing units of μDC in each information processing apparatus is one has been described, but the information processing unit ( It is also possible to deal with a case where the number of UNT) is plural.
FIG. 9 shows a configuration example of the μDC 100 according to the third embodiment of the present invention. The information processing apparatus μDC100 includes an information processing unit 1 (UNT1) 110A and an information processing unit 2 (UNT2) 110B. Each unit has a controller (CTR) and a storage (STR), and the information processing apparatus And a common router (RTR) 140.

  In the third embodiment, the number of information processing units (UNT) is a redundant configuration of two files. Of the four files in each LU, files 1 and 3 have primary information processing unit 1 (UNT1) and secondary information processing unit 2 (UNT2), and files 2 and 4 have primary information processing unit 2 (UNT2) and secondary information. It is assumed that the information processing unit 1 (UNT1). Redundancy management tables (RMT) 1222A and 1222B are managed as information processing unit / I node pairs (information processing unit, I node) information for each file. For example, in the redundancy management table (RMT) 1222A of the information processing unit 1 (UNT1), since the other file is in the information processing unit 2 (UNT2), the information of the information processing unit in units of files is 2. The I node number is the same regardless of the information processing unit, but it may not be the same.

  In the present embodiment, as in the first embodiment, it is possible to reduce the time of storage copy at the site at the time of replacement, and there is an advantage that the service for the user can be continued without stopping the application.

  Up to now, an example of device exchange in the fine-grained file system management unit (FSM) in units of files has been shown. A method that does not copy the secondary can also be applied to the operation when a failure occurs.

  As a fourth embodiment, FIG. 10 shows a configuration example of the μDC 100 including three information processing units (110A, 110B, and 110C) according to the fourth embodiment of the present invention. The time chart of FIG. 11 shows the operation when a failure occurs.

  Each file of the LU is duplicated in another information processing unit in the μDC 100 as in the third embodiment. When a failure occurs in one of the files, the redundancy management table (RMT) 1222 is updated and the service is immediately continued using the primary storage, and the secondary storage is copied in the background while the service is resumed. . In this example, as shown in FIG. 11, in the normal state, the primary file (P) and the secondary file (S) function normally in both the information processing unit 1 (110A) and the information processing unit 2 (110B). Shall.

  Here, it is assumed that a failure has occurred in the information processing unit 1 (110A) ((1) in FIG. 11) and the data center (CDC) 400 has been notified. Thereby, according to the redundancy management table (RMT) 1222 of the information processing unit 1 (110A), the information processing unit 2 (110B) has the primary file (S) for the files of the I node 2 and the I node 4. Therefore, application service execution is not affected. Regarding the files of the I node 1 and the I node 3 of the information processing unit 1, since the file of the information processing unit 2 stores the same contents as the secondary file (S), this (S of the failure node) is designated as the primary (P ) To quickly resume application services. Specifically, in (2) of FIG. 11, the information in the redundancy table (RMT) of information processing unit 2 (110B) is changed (S of I node 1 and I node 3 is changed to P). . Then, the application executed by the information processing unit 1 may be changed to be executed by the information processing unit 2. Then, after notifying the data center (CDC) 400 of the service restart, the information processing unit 2 (110B) can immediately restart the application service ((3) and (4) in FIG. 11). On the other hand, the secondary (S) file may be copied to the new information processing unit 3 (110C), and the related application may be deployed in the background ((5) in FIG. 11). Here, for the sake of simplicity, only the information processing unit 1 and the information processing unit 2 are used. However, even if a plurality of information processing units have a relationship between primary and secondary of a file, they can be handled in the same manner.

  As described above, the service restart time can be shortened by managing primary and secondary in units of files and copying only some files in the event of a μDC replacement or failure based on information in the update management table. . The μDC operation management middleware monitors the storage copy status at the time of replacement or failure, but it can be determined by observing this copy status that the secondary is not copied.

  100: μDC 1 before replacement 110: Information processing unit 120: Controller (CTR) 121 ... Memory (MEM) 122 ... File system management unit (FSM) 123 ... Buffer (BUF) 124: CPU 125 Network interface (NIC), 126 ... Host bus adapter (HBA), 130 ... Storage (STR), 140 ... Router (RTR), 200 ... μDC2 for replacement, 300 ... Network, 1223 ... Offline difference management table (ODMT), 1224 ... Online exchange trigger (NTRG) part, 1225 ... Offline exchange trigger (FTRG) part.

Claims (15)

An information processing apparatus used as a distributed cloud that provides an application via a network,
The information processing apparatus is an information processing apparatus constituting a small data center in a distributed cloud environment that processes data in cooperation with a central data center,
The information processing apparatus includes:
Comprising at least one information processing unit having a controller and a storage;
The storage includes a file system having a primary file and a secondary file of a pair in file units or directory units,
The controller is
A file system management unit for managing the file system;
The file system management unit includes a redundancy management table, an update management table, an online exchange trigger unit, and an offline exchange trigger unit.
In the replacement of the first and second information processing units having the same function online and offline,
The online exchange trigger unit and the offline exchange trigger unit have a function of controlling an online exchange trigger and an offline exchange trigger for exchanging the first information processing unit and the second information processing unit online or offline. Have
The redundancy management table manages the correspondence between the primary file and the secondary file of the file system, and writes to the secondary file corresponding to when the primary file is written,
The update management table has a function of managing information related to the update of the file system after the offline exchange trigger is set by the offline exchange trigger unit,
The file system management unit
With the setting of the online exchange trigger by the online exchange trigger unit, the first information processing based on the information on the update of the file system after the offline exchange trigger is set based on the information of the update management table only the file system updated the primary file unit, the information processing apparatus characterized by having the ability to copy to said second information processing unit. In claim 1,
The file system management unit of the second information processing unit is
The information processing apparatus characterized in that said updated the secondary file other than the primary file, autonomously generated in a storage of the second information processing unit. In claim 2,
The update management table is an offline difference management table;
The file system management unit
Along with the set of the offline exchange trigger, the difference between the storages of the first information processing unit and the second information processing unit is detected and recorded in the offline difference management table,
In accordance with the set of online exchange triggers, only the primary file is copied to the second information processing unit with respect to the storage difference. In claim 2,
The update management table is a priority table;
The file system management unit
In accordance with the set of the online exchange trigger, only the primary file of the first information processing unit is copied to the second information processing unit based on information in the priority table. . In claim 4,
The information processing apparatus, wherein the priority table lowers the priority of a file that has not been accessed for a certain period of time or an old version of the file. In claim 2,
The information processing apparatus and the first information processing unit, the second information processing unit, characterized in that provided in said one information processing apparatus. In claim 2,
Wherein the one information processing apparatus, the first information processing unit having the same function, the second information processing unit, and includes a third information processing unit,
Exchanged from the first information processing unit to the second information processing unit and the third information processing unit,
The online exchange trigger unit and the offline exchange trigger unit have a function of controlling an online exchange trigger for performing the exchange of the first to third information processing units online or offline, an offline exchange trigger ,
Each of the information processing units is
When the primary file is written, the secondary file is written,
When a failure occurs in the first information processing unit and the first information processing unit is not used,
In the second information processing unit having a secondary file of the first information processing unit,
A procedure for changing the secondary bit of the redundancy management table of the second information processing unit to the primary bit, and after starting the update of the primary file in the second information processing unit,
A procedure for registering a secondary file in the redundancy management table of the third information processing unit;
An information processing apparatus comprising a procedure of copying a primary file of the second information processing unit to the third information processing unit as a secondary file. An information processing apparatus exchange support system used as a distributed cloud that provides applications via a network,
The information processing apparatus is an information processing apparatus constituting a small data center in a distributed cloud environment that processes data in cooperation with a central data center,
When exchanging the first information processing device constituting the small data center with a second information processing device having the same function as the first information processing device,
Via the network connecting the said second information processing apparatus of the central data center and the first information processing apparatus of the small data center, the files on the storage of the first information processing apparatus Copy to the storage of the second information processing device,
After transporting the second information processing apparatus to the small data center, only a part of the storage difference between the first information processing apparatus and the second information processing apparatus generated during the transport is stored in the first information processing apparatus. Copy to the second information processing device, perform the exchange,
The remaining part of the difference between the storages is autonomously generated in the storage of the replaced second information processing apparatus. In claim 8,
The first information processing apparatus and the second information processing apparatus are:
Controller and comprising at least one of the information processing unit having said storage,
The storage includes a file system having a primary file and a secondary file of a pair in file units or directory units,
The controller is
A file system management unit for managing the file system;
The file system management unit includes a redundancy management table, an update management table, an online exchange trigger unit, and an offline exchange trigger unit.
The replacement of the first information processing device and the second information processing device is performed offline during the transportation, and is performed online after the transportation,
The online exchange trigger unit and the offline exchange trigger unit have a function of controlling an online exchange trigger and an offline exchange trigger for exchanging the first information processing apparatus and the second information processing apparatus online or offline. Have
The redundancy management table manages the correspondence between the primary file and the secondary file of the file system, and writes to the secondary file corresponding to when the primary file is written,
The update management table has a function of managing information related to the update of the file system of the first information processing apparatus after the offline exchange trigger is set by the offline exchange trigger unit,
The first information processing apparatus refers to the update management table with the setting of the online exchange trigger by the file system management unit, and based on information related to the update of the file system after the offline exchange trigger is set. exchange support system of the information processing apparatus, characterized in that only the primary file of the updated file system, has the ability to copy to said second information processing apparatus. In claim 9,
The update management table is an offline difference management table;
With the setting of the offline exchange trigger, the difference between the storages of the information processing units of the first and second information processing devices is detected and recorded in the offline difference management table,
As the online exchange trigger is set, regarding the storage difference, the offline difference management table of the first information processing apparatus is dirty, and the primary file of the redundancy management table is used as the second information processing apparatus. Copy to
After the exchange,
The offline difference management table of the first information processing apparatus is dirty, and the secondary file of the redundancy management table is copied from the primary file of the second information processing apparatus to the second information processing apparatus as a secondary file. An information processing apparatus replacement support system. In claim 10 ,
The update management table is a priority table;
With the setting of the offline exchange trigger , based on the information in the priority table , start the function of the offline difference management table,
The offline difference management table of the first information processing apparatus is dirty, and the primary file of the redundancy management table is copied to the second information processing apparatus ,
An information processing apparatus replacement support system, wherein after the replacement, the remaining part of the storage is autonomously generated in the second information processing apparatus. A method of supporting the exchange of information processing devices used as a distributed cloud that provides applications over a network,
Each of the information processing devices is an information processing device that constitutes a small data center in a distributed cloud environment that processes data in cooperation with a central data center,
When exchanging the first information processing device constituting the small data center with a second information processing device having the same function as the first information processing device,
Via the network connecting the said second information processing apparatus of the central data center and the first information processing apparatus of the small data center, the files on the storage of the first information processing apparatus Copy to the storage of the second information processing device,
After transporting the second information processing apparatus to the small data center, only a part of the storage difference between the first information processing apparatus and the second information processing apparatus generated during the transport is stored in the first information processing apparatus. Copy to the second information processing device, perform the exchange,
The remaining part of the difference in the storage is autonomously generated in the storage of the replaced second information processing apparatus. In claim 12,
The first information processing apparatus and the second information processing apparatus are:
Controller and comprising at least one of the information processing unit having said storage,
The storage includes a file system having a primary file and a secondary file of a pair in file units or directory units,
The controller has a file system management unit for managing the file system,
The redundancy management table manages the correspondence between primary and secondary in file units, and the first information processing device is written when the primary file of the first information processing device is written during the transportation Write to the secondary file of
The update management table manages whether or not the file is being updated during transportation,
When exchanging from the first information processing device to the second information processing device,
Performing a first copy procedure for copying the primary file and the secondary file of the first information processing apparatus to the second information processing apparatus connected via a network;
After the first copy procedure, with respect to the storage difference, the primary file of the specific file updated during the transportation is transferred to the second information processing apparatus based on the information in the update management table. Perform the second copy procedure to copy,
After the exchange,
Regarding the difference in storage, based on the information in the update management table, execute a third copy procedure for copying in the storage of the second information processing apparatus for files other than the updated primary file. An information processing apparatus replacement support method, wherein the paired primary file and secondary file are constructed. In claim 13,
The update management table is an offline difference management table;
After the first copy procedure, start the function of the offline difference management table,
The offline difference management table of the first information processing apparatus is dirty, and the second copy procedure for copying the primary file of the redundancy management table to the second information processing apparatus is executed.
After the exchange,
Wherein the offline differential management table of the first information processing apparatus is dirty and the secondary file of the redundancy management table, as the second secondary file to said second information processing apparatus from the primary file of the information processing apparatus An information processing apparatus replacement support method, wherein the third copy procedure for copying is executed. In claim 14 ,
The update management table is a priority table;
With the set of off-line exchange trigger to initiate a function of the off-line differential management table,
The offline difference management table of the first information processing apparatus is dirty, and the primary file of the redundancy management table is copied to the second information processing apparatus ,
After the exchange, the remaining part of the storage is autonomously generated with the paired primary file and secondary file in the exchange of the second information processing apparatus. Exchange support method.

Images