WO2007147318A1

WO2007147318A1 - Service health level detection system and method

Info

Publication number: WO2007147318A1
Application number: PCT/CN2007/001588
Authority: WO
Inventors: Yikang Lei
Original assignee: Huawei Technologies Co., Ltd.
Priority date: 2006-06-15
Filing date: 2007-05-16
Publication date: 2007-12-27
Also published as: CN100461719C; CN101043385A

Abstract

A service health level detection system and method are disclosed. The system includes a forwarding device (230) and at least one host (240), wherein the host (240) is used to provide the service to a user and to provide its health level information to the forwarding device (230); the health level information includes the service states and the host states. The forwarding device (230) is used to acquire the health level information of the host (240), and to assign the service to the host (240) according to the acquired health level information. The method comprises the following: a forwarding device (230) sends an inquiring request to a host (240); the host (240) acquires the corresponding health level information according to the type of the inquiring request; the health level information includes the service states and the host states; and the host (240) sends the acquired health level information to the forwarding device (230). According to the system and the method, it is possible to implement the correlation detection of the any service on the host (240) to provide the sufficient basis for the forwarding device (230) to select the service host (240).

Description

Service ^ 13⁄4 Kang degree detection system and method

The present application claims priority to Chinese Patent Application No. 200610061152.8, entitled "Service Health Test System and Method", filed on June 15, 2006, the entire contents of which is hereby incorporated by reference. in.

Technical field

The present invention relates to a communication network service health detection technology, and more particularly to a service health detection system and method.

Background technique

With the development of IP (Internet Protocol) technology, the popularity of broadband, and the exponential growth of network users, network traffic on the Internet has rapidly expanded, and various Internet services have become more and more, and a single server has been unable to meet customer needs. In order to solve the above problem, a set of servers is usually used to provide services to users, and a single interface is provided externally (that is, the group of servers provides only one IP address for service).

Referring to FIG. 1 , FIG. 1 illustrates a solution to the above problem in the prior art, and uses a server load balancing device to detect server health: RS1 to RS7 are a set of servers that ultimately provide services, and server load balancing devices are placed on the server. Between the users being served; the server load balancing device has a set of virtual servers VS1 to VS3, each virtual server corresponding to a service category, as shown in the service classes 1 to 3; after the user service request arrives at the carrier service network, The service request is first transmitted to the server load balancing device. The server load balancing is set to the category of the user service request, and the service request is allocated to the corresponding virtual server, and the virtual server forwards the service request to the corresponding server for processing. In this process, in order to ensure that the service achieves satisfactory results for the user, the server load balancing device needs to check the health of the service to avoid distributing the user request to the device that cannot provide the service.

The so-called health refers to the health status of related application services or host resources on the server, where:

Host resource health status includes: CPU (Central Processing Unit) usage, memory usage, service bandwidth usage, and other critical resources, such as number of processes, number of threads, number of sockets (socket) resources, etc.

The application service health status includes: the number of critical resources that affect user response, concurrent processing capabilities, The number of resources currently consumed, etc., resources include sockets, application available threads, counters, background connections, memory, hard disks, etc. where there may be bottlenecks.

In general, server failure states are manifested as link failures, host failures, and application failures (also known as service failures). Correspondingly, when the server fails to recover, there are three other states, namely the link recovery state, the host recovery state, and the application (service) recovery state. How to detect these states to ensure the reliability of services has become a common concern of communication service operators.

The currently known methods for detecting service health are as follows:

Prior art solution 1 uses Ethernet (Ethernet) OAM technology for service health check.

OAM refers to Operation, Administration, and Maintenance. The functions of this technology include: performance monitoring and generation of maintenance information, and evaluation of network stability based on this information; detection of network failures through periodic query methods, resulting in Various maintenance and alarm information; By scheduling or switching to other entities, bypassing the failed entity to ensure the normal operation of the network; transmitting the fault information to the management entity.

Ethernet OAM provides Layer 2 end-to-end monitoring and detection to detect link status connected through Layer 2 switching devices. However, since this technology can only detect the link status, it cannot detect the state of the server host and has a great dependence on the peripheral devices of the host, which is not conducive to the application.

Prior Art Solution 2: The health check is performed by using the ping (Internet Control Message Protocol Echo Command) method.

Using the ping method for health check is a common server detection method. The implementation method is very simple: On the server load balancing device, the timer is used to periodically send ping packets to the IP addresses of each host, and detect whether there is a response. If there is a response, the host is healthy, otherwise the host has expired.

Although the method has the advantages of simple implementation method and good technical extensibility, it may be shielded by the host. At the same time, frequent pings are easily identified as attack packets and need to start related ping services on the server load balancing device. Configuring an IP address on the router and on the host may conflict with the monitoring network. At the same time, this method can only provide host status detection and cannot provide service status detection.

Prior art scheme 3 uses the arp (address resolution protocol) method for health check.

Using the arp method for health check is a common method, and its implementation is as follows: On the server load balancing device, a timer is used to periodically send an aip message to each host's IP address for address request, and the ARP response packet is detected. If there is a response, the host is healthy. Otherwise, the host is invalid.

The method has the advantages of simple implementation method and low risk of being attacked. However, it can only be applied to Layer 2 network networking. The scalability is not good. It can only provide host status detection and cannot provide service status detection. It may conflict with the monitoring network.

Summary of the invention

In view of this, the main purpose of the embodiments of the present invention is to provide a service health detection method, which is used to implement comprehensive health detection of a server, thereby improving reliability of service processing and increasing user satisfaction.

Another main object of the embodiments of the present invention is to provide a service health detection system for implementing comprehensive health detection of a server to improve reliability of service processing and increase user satisfaction.

To achieve the above object, the present invention is achieved as follows:

Embodiments of the present invention disclose a service health detection system including a front device and at least one host host, wherein:

a host host, configured to provide a service to the user, and the front device provides its health information; the health degree information includes a service status and a host status;

The pre-device is used to obtain the health information of the host host, and the service is allocated to the host host according to the obtained health information.

An embodiment of the present invention further provides a service health detection method, including:

The pre-device sends a health query request to the host host;

The host host obtains corresponding health degree information according to the type of the query request; the health degree information includes a monthly service state and a host state;

The host host sends the acquired health information to the front device.

An embodiment of the present invention further discloses a second service health detection method, including:

The pre-device registers the health registration with the host host;

The host host sends the registered health information to the front device; the health information includes a service status and a host status.

The embodiment of the invention comprehensively checks the health of related services on the host host, including the main Machine status, including service status. The embodiment of the present invention obtains the exact state of the related service by checking the related resource items, provides a sufficient basis for the front-end device to select the background host host and allocates the service, and particularly provides a complete service for the service that only receives the service. The health check method and scheme achieve the effect of improving the reliability of business processing and increasing user satisfaction.

DRAWINGS

FIG. 1 is a schematic diagram of a prior art system for detecting server health by using a server load balancing device;

2 is a schematic diagram of an embodiment of a service health detection system according to the present invention;

3 is a flowchart of an embodiment of a query-response health detection method according to the present invention; - FIG. 4 is a flowchart of an embodiment of a registration-notification health detection method according to the present invention;

FIG. 5 is a flowchart of an embodiment of a registration-query health detection method according to the present invention.

detailed description

The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

The service health detection system provided by the embodiment of the present invention includes a user, an operator service network, and an operator service device. After the service request of the user is transmitted to the operator service device through the service network of the operator, the service device in the service provider The pre-device performs a service health check on the host host and distributes the service to a host host with good health. In this way, the operator service network can provide a better service response to the user. Through the system, the host state and service state of the host host can be detected without any omission, and the pre-device can not allocate the user service request to the host host that cannot provide the service when the service is allocated, and can also be in front of Differentiated service health detection is implemented between the device and the host to improve detection efficiency and save system resources.

The service health detection method provided by the embodiment of the present invention initiates a health query request from the pre-device including the query end to the host host including the host query agent and the service response end, and queries the host host status and service status of the host host, and the host host The pre-device sends a service status and host status response. Through this method, the host state and the service state of the host host can be detected without any omission, and the pre-device can not allocate the user request to the host host that cannot provide the service when the service is allocated, and can also be pre-positioned. Differentiated service health detection between the device and the host host achieves the purpose of improving detection efficiency and saving system resources. The present invention provides three embodiments of service health detection methods: a query-response method, a registration-notification method, and a registration-query method. The host status, that is, the host health status in the foregoing method embodiments and system embodiments includes: the state of the key resources such as the CPU usage rate, the memory usage rate, the service bandwidth occupancy rate, the number of processes, the number of threads, and the number of socket resources; The service health status includes: concurrent processing capability, current consumed resources, socket resources, application available threads, counters, background connections, memory, hard disks, number of sessions, nat (network address translation) addresses, and other application table resources. status.

Referring to FIG. 2, it is a schematic diagram of an embodiment of a service health detection system of the present invention. The system includes a user 100, an operator service network 210, and an operator service device 220. The carrier service device 220 includes a front device 230 and a host host 240. The front device 230 is configured to detect the service health of the host host 240, and the query terminal 231 is disposed thereon;

The query terminal 231 is a general term for the query client 232 and the pre-query proxy 233 on the pre-device, and is used to initiate a query request to the host host.

The query client 232, which may also be referred to as a query request originating end, refers to a specific application on the front device that needs to obtain the state of the server;

The pre-query agent 233, a health check service working on the pre-device 230, is used to transparently query the query request initiated by the client 232 and provide encapsulation and parsing services for the information of the query client.

The host host 240 is configured to provide services to the user 100. The host host 240 is provided with a host query proxy 242, a service responder 243, and a storage module 241. among them:

The host query agent 242, a health check service working on the host host 240, for initiating, responding to, and forwarding the query request from the pre-device 230, including the service status query request and/or the host status query request, and according to The query request of the pre-device 230 sends the host status and/or service status of the host host 240 to the pre-device 230. The query of the iterative method can be implemented by the host query agent 242, that is, the response end of the one-month service health check is also the query end of another service health query, and the host query agent 242 can provide the host host status of the responder; The response end 243 may be an application service that is queried on the host host 240. When receiving the health query of the host query agent 242, the health of the related service, that is, the service status of the related service, is fed back. At the same time, the service responder 243 also provides services to the user 100.

The storage module 241 is configured to cache a service state and/or a host state.

In the system shown in FIG. 2, the user 100 initiates a service request to the operator service network 210. 2007/001588 a 6—

After receiving the service request from the user 100, the service provider network 210 sends the request to the pre-device 230. The pre-device 230 detects the host status and service status of the connected set of host hosts 240 or reads through the query terminal 231. The host state and service state of the local machine stored on the host host 240 are taken, and then the service request is assigned to the host host 240 with good health according to a predetermined rule.

At the same time, it should be noted that the user referred to in the system is not an individual or organization in the general sense, but includes all mechanisms for initiating a service request to the operator, which may be an individual, or an intermediate device or terminal. . The operators referred to in this system are not only the operators in the general sense, but also all the individuals, organizations, intermediate devices and terminals that provide certain applications.

In addition, the query terminal 231 of the pre-device 230 is provided in the system, and includes a query client 232 and a pre-query proxy 233. The host host 240 includes a host query proxy 242 and a service response terminal 243, but the concept of the embodiment is not To be limited thereto, any behavior of recombining or dividing these functions is within the concept of the present embodiment.

One embodiment of the service health detection method disclosed by the present invention is based on a query-response method. The pre-device sends a health query request to the host host, and the host host sends a health query response to the device. This embodiment may include three operations: the pre-device queries the instant service status on the host host, the pre-device queries the cached service status on the host host, and the pre-device queries the host host status. In this way, different frequency service health queries can be implemented between the pre-device and the host host, that is, the query agent on the host side of the query terminal can be a high-frequency query and response, and the host-service is a low-frequency query and response. . Description will be made below with reference to the drawings.

FIG. 3 is a flow chart showing an embodiment of a health-detection method based on a query-response method in the present invention, comprising the following steps:

301. The front device querying end queries the host host to query the agent to initiate a health query request. The query may be an immediate service state or a host state, or may be a service state or a host state cached according to a specific condition on the host host.

302. After receiving the health query request, the host querying agent of the host host determines the type of the request. If the request is for the instant service state, step 303 is performed; if the request is for the cached service state or host state of the host host, , performing step 306;

303. The host query agent of the host host initiates a query request for querying the instant service status to the host response end of the host host; T/CN2007/001588

-7-

304. The service response end of the host host queries the host host to query the agent to feed the instant service status.

305. After receiving the instant service status information of the service responding end in step 304, the host querying agent of the host host forwards the information to the front-end device query end.

If the host query agent does not receive the instant service status information of the service responder under the set condition, it determines that the instant service status is invalid, and notifies the front device query end of the immediate service invalidation status;

306. After receiving the local cached service status and the host status or the instant host status health query request, the host host query agent reads the locally cached service status or the host status from the storage module, or obtains the instant host status.

307. The host host querying agent sends the cached service state or host state obtained in step 306 to the forward device query end.

The health information fed back to the front device by the host host may include a cache mechanism state in addition to the instant service state, the instant host state, the cache service state, and the cache host state. When the host query agent cannot read the locally cached service status under the set condition, it confirms that the cache mechanism status is invalid and notifies the pre-device.

With this embodiment, service status queries of different frequencies can be implemented. For those services or host resources with higher performance requirements, the query can be performed at any time, and for those services or host resources whose performance requirements are not too high, it can be set. Cache its state on the storage module of the host host every certain period of time. When you need to query, just read the information that has been stored. For the cached host state or service state, it can be specified that as long as the obtained cache service state or host state is healthy, it can be determined that the service or host resource is healthy and available.

Another embodiment of the service health detection method disclosed by the present invention is based on a registration-notification method. First, the pre-device registers the health information that needs to be queried on the host host. After the registration is completed, the host host's host query agent periodically sends the registered service health information to the pre-device. Therefore, as long as the host query agent of the host host is healthy, the front device does not need to frequently check the health status of the host host. This method can include steps such as querying health registration, pre-query proxy health query, and host host status query. A detailed description will be given below.

4 is a flow chart of an embodiment of a service health detection method based on a registration-notification method in the present invention -8- Figure, the specific steps include:

401. The front-end device query end completes the local query health registration. The setting is to query the local device health, and the query content may also include the service status and the host status.

402. The front-end device querying end queries the host host for the proxy request to query the health registration;

403. After receiving the foregoing query health registration registration request, the host querying agent of the host host determines the type of query registration required. If the current service status is to be registered, step 404 is performed; if the host host is cached Service status or host status, step 407;

404. The host query agent of the host host requests the service response of the host host to perform an instant service status query registration;

405. After receiving the registration request, the service response end of the host host completes the instant service status query registration, and queries the host host's host to query the instant service status according to the registration condition;

406. The host querying agent of the host host notifies the front-end device query end of the instant service status obtained in step 405;

407. The host querying agent of the host host performs local cached service status and host status query registration, and obtains the cached service status and host status.

408. The host querying agent of the host host notifies the device query end to notify the cache service state or the host state obtained in step 407 according to the registration condition.

In the above process, the following contents may also be included:

After the pre-device requests the query service registration from the host host, the host query agent sends a message to the host host to query whether the query service is successfully registered. If the query service is successfully registered, the front-end device query end does not perform other operations. If the service registration fails, the query service registration request may be initiated again;

The front device query end sends a proxy health query request to the host query agent of the host host to query the working status of the host query agent. If the pre-device receives the query response of the host host's host query agent, the host host is considered to query the agent. It is healthy. Otherwise, it is judged that the host host query agent has expired, and then the host host is judged to be invalid.

It should be noted that, in this embodiment, the host host may be configured to send the registered service state or the host state to the device according to a certain time period as described above, or may set the host host. The state change information is sent to the front device only when the state of the host changes or the registered service state changes. Specifically include:

When the host host changes the host state, the host host queries the proxy device to send a host state change message to the forward device query end. After receiving the status change message, the pre-device query end queries the host host for the status change response sent by the proxy.

When the host host changes the service status, the host host service responds to send a service status change message to the device query end. After receiving the status change message, the front device query end sends a status change response to the host host service response end.

The invention also discloses an embodiment of a service health detection method based on a registration-query method. In the present embodiment, the service registration method involved is completely the same as the registration method in the above registration-notification method. The difference is that, in this embodiment, the pre-device not only passively waits for the host host to send the registered service status or the host host status, but also can set certain conditions. When the set condition is met or the temporary demand occurs, the pre-device Proactively query the host status and service status of the host host. This embodiment will be specifically described below.

FIG. 5 is a flowchart of an embodiment of a service health detection method based on a registration-query method in the present invention, which specifically includes:

501. The front-end device query end completes the local query health registration. The setting is to query the local device health. The query content may also include the service status and the host status.

502. The front-end device querying end queries the host host's host to query the proxy request for querying the health registration;

503. After receiving the foregoing query health registration request, the host querying agent of the host host determines the type of query registration required. If the current service state is to be registered, step 504 is performed; if the host host is cached, Service status or host status, step 507;

504. The host query agent of the host host requests the service response query of the host host to perform instant service status query registration;

505. After receiving the registration request, the service response host of the host host completes the instant service status query registration, and queries the host host's host to query the instant service status according to the registration condition;

506. The host querying agent of the host host notifies the front-end device querying end of the instant service status obtained in step 505; 2007/001588

- 10-

507. The host querying agent of the host host performs local cached service status and host status query registration, and obtains the cached service status and host status.

508. The host querying agent of the host host notifies the device query end to notify the cache service state or the host state obtained in step 507 according to the registration condition.

The above steps are used to obtain the registered service status and the host status under the normal conditions. When the pre-registration device does not receive the notification information of the host host or the specific query request occurs, the embodiment may further include the following. Steps:

509. The querier of the pre-device sends a health query request to the host query agent of the host host. If the host query agent of the host host cannot receive the response after a period of time, the host host is determined to have expired;

510. The host query agent of the host host determines whether the requested query service is already registered. If the service is already registered, step 511 is performed. If the service is not registered and the query request is in the instant service state, step 513 is performed; The service is not registered and the query request is the cached service state or host state on the host host, then step 516 is performed;

511. The host query agent of the host host invokes the registered function to obtain the corresponding service or host status.

512. The host querying agent of the host host sends the service or host status information obtained in step 511 to the forward device query end.

513. The host query agent of the host host requests the host host service response end to perform an instant service status query.

514. The host host service response end queries the host of the host host to query the service status of the agent query in step 513.

515. The host querying agent of the host host sends the monthly service status obtained in step 514 to the query device of the forward device as a query response to the query end of the pre-device;

516. The host querying agent of the host host obtains a locally cached service state or a host state.

517. The host querying agent of the host host sends the service status or the host status obtained in step 516 to the query device of the forward device as a query response to the query end of the pre-device.

As can be seen from the above steps, this embodiment is an improvement of the foregoing query-response method and registration-notification method. Different from the foregoing query-response method, the active query in this embodiment is based on certain Conditions, this condition can be set as desired.

The query device of the pre-device in the method embodiment of the present invention may also be divided into a query client and a query proxy according to requirements, and the query proxy and the service responder on the host host in the method embodiment of the present invention may also be used. Any one of the functional modules that are recombined or divided by the functional modules of the embodiments of the present invention is incorporated in the concept of the embodiments of the present invention.

In summary, the service health detection system provided by the embodiment of the present invention can reliably implement the query of the host status and service status of the host host, and ensure that the front device does not allocate the user service request when the service is allocated. On the host host of the service, it is also possible to implement different service health detections for different state requirements on the pre-device and the host host to which it is connected, so as to improve detection efficiency and save system resources.

The service health detection method provided by the embodiments of the present invention can be applied to any network technology or any host technology, and can be carried in any Layer 2, Layer 3, Layer 4 or application layer protocol, or can be remotely invoked ( The RPC method obtains related services; and the front device can be regarded as a large virtual host for iterative query.

The embodiments described above are merely preferred embodiments of the present invention, and are not intended to limit the scope of the present invention. Any person skilled in the art can easily use the technical scope disclosed in the present invention. All changes or substitutions contemplated are intended to be included within the scope of the invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

Rights request

A service health detection system, comprising: a front device and at least one host host, wherein:

The device is configured to obtain health information of the host host, and allocates services to the host host according to the obtained health information.

2. The system according to claim 1, wherein the pre-device further comprises a query end, configured to initiate a query request to the host host; the query request includes a service status query request and/or a host status Query request.

The system of claim 2, wherein the querying end further comprises a query client and a pre-query proxy, wherein

Query the client, used to initiate a query request;

The pre-query proxy is used for transparently querying the query request initiated by the client and encapsulating and parsing the query request of the query client.

4. The system of claim 2, wherein the host host further comprises a host query agent and a monthly response end, wherein

a host query agent, configured to initiate, respond to, and forward a service status and/or host status query request from the pre-device, and send the host host's host status and/or service status to the pre-device; the service response end is used to The user provides the service and responds to the service status query request initiated by the host query agent.

5. The system of claim 4, wherein the host host further comprises a storage module for storing a host state and a service state of the host host.

6. A service health detection method, comprising:

The pre-device sends a health query request to the host host;

The host host obtains the health information according to the type of the query request; the health information includes a service status and a host status;

The host host sends the acquired health information to the front device.

7. The method according to claim 6, wherein the host host according to the query request The type of obtaining the corresponding health information further includes:

If the query request type is the instant service status, the host query agent of the host host initiates a query for the instant service status request to the service responder, and the service response end feeds back the corresponding service status information to the host query agent;

If the query request type is host status or cached service status, the host host's host query agent queries the host status or reads the locally cached service status to get the corresponding host status or service status information.

The method according to claim 7, wherein the method further comprises: if the host querying agent does not receive the instant service status information of the service responding end under the set condition, determining that the instant service status is invalid.

The method according to claim 8, wherein the health information further includes a cache mechanism state of the host host;

The method further includes confirming that the state of the cache mechanism is invalid if the host query agent cannot read the locally cached service state under the set condition.

10. A service health detection method, comprising:

The pre-device registers the health registration with the host host;

The method according to claim 10, wherein the performing, by the pre-device, the query health registration to the host host further comprises:

The front device query end sends a query registration request to the host query agent of the host host; when the query registration type is an instant service status query, the host host's host query agent initiates an instant service status query registration request to the host host service responder, the service The response end correspondingly performs an instant service status query registration;

When the type of the query registration is the cached service state and the host state, the host host's host query agent completes the cached service state and host state query registration.

The method according to claim 10, wherein the method further comprises: performing a local query health registration by the front device query end.

13. The method of claim 10, wherein the method further comprises: The device detects whether the query service on the host host is successfully registered; if the query service registration fails, the device registers the health registration again.

The method according to claim 13, wherein the method further comprises: after the query service is successfully registered, the pre-device detects the host query agent health of the host host; if not received under the set condition The host host's host queries the agent health response, and the pre-device considers the host host to be invalid.

The method according to any one of claims 10 to 12, wherein the method further comprises: after the pre-device performs the query health registration with the host host:

The pre-device actively initiates a service status query or host status query to the host host.

The method according to claim 15, wherein the pre-device actively initiates a service status query or a host status query to the host host, and further includes:

The pre-device sends a health query request to the host host;

The host host obtains corresponding health information according to the type of the query request, and the health information includes a service status and/or a host status;

The host host sends the acquired health information to the front device.

The method according to claim 16, wherein the obtaining, by the host host, the corresponding health information according to the type of the query request further comprises:

The host query agent of the host host determines whether the query request of the pre-device is registered;

If the query request has been registered, the host host's host query agent callback registration function obtains the service status or host status information;

If the query request is not registered and the request is for the instant service status, the host host's host query agent sends a request to the host host service responder to query the instant service status;

If the query request is not registered and the request is for a cached service state or host state, the host host's host query agent obtains the cached service state or host state.