CN111190790A - Cloud computing cluster monitoring method and system based on peak prediction - Google Patents
Cloud computing cluster monitoring method and system based on peak prediction Download PDFInfo
- Publication number
- CN111190790A CN111190790A CN201911304713.6A CN201911304713A CN111190790A CN 111190790 A CN111190790 A CN 111190790A CN 201911304713 A CN201911304713 A CN 201911304713A CN 111190790 A CN111190790 A CN 111190790A
- Authority
- CN
- China
- Prior art keywords
- data
- submodule
- cluster
- value
- curve
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/3003—Monitoring arrangements specially adapted to the computing system or computing system component being monitored
- G06F11/3006—Monitoring arrangements specially adapted to the computing system or computing system component being monitored where the computing system is distributed, e.g. networked systems, clusters, multiprocessor systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/3051—Monitoring arrangements for monitoring the configuration of the computing system or of the computing system component, e.g. monitoring the presence of processing resources, peripherals, I/O links, software programs
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computing Systems (AREA)
- Physics & Mathematics (AREA)
- Quality & Reliability (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Debugging And Monitoring (AREA)
Abstract
The invention discloses a cloud computing cluster monitoring method and system based on peak value prediction.A data acquisition end is responsible for acquiring system performance information and services of each node server and a Docker container, a Ceph distributed file system and the like, and dynamically acquiring the CPU utilization rate, the memory utilization rate, the disk utilization rate, the service state and logs of the current node server, the resource occupation condition of the Docker container, the Ceph running state, cluster information and the like periodically; the data transmitting and receiving end is responsible for transmitting various information collected by the data acquisition end to the management node, runs on the management node and monitors and receives real-time data transmitted by each node server; and the data storage end adopts whisper as a storage rear end and stores the data received by the management node by using the time sequence database. The invention can accurately carry out prediction analysis, thereby predicting the problems possibly occurring in the cluster, effectively reducing the maintenance cost and improving the stability and reliability of the cluster server.
Description
Technical Field
The invention belongs to the technical field of monitoring, and particularly relates to a cloud computing cluster monitoring method and system based on peak prediction.
Background
As the cloud computing technology is increasingly applied to various fields of the information industry, the demand for monitoring and managing a server cluster in a cloud computing environment is increasing. The cluster server system under cloud computing is mainly composed of a series of server clusters based on a distributed architecture, and a set of effective cloud computing cluster monitoring system is inevitably needed to monitor, regulate and control the cluster server system to manage the cluster server system and ensure high-performance operation of the cluster server system.
In order to facilitate monitoring and management of a user on a server cluster based on embedded hardware and enhance the stability of the whole cluster, an effective and highly reliable cloud resource monitoring system is essential. For a large server cluster, if one node fails, large-area paralysis of the cluster system is likely to be caused, so that in order to avoid the problem of feedback delay of server cluster management, a peak prediction algorithm with high accuracy is required to perform prediction analysis on important indexes in each server, so that an administrator can timely handle the failure.
At present, the commonly used data acquisition modes mostly use some comprehensive and complex acquisition tools, and the tools cannot be completely adapted to the server cluster in the embedded environment due to the defects of complicated codes and excessive interface types.
Disclosure of Invention
The technical problem to be solved by the present invention is to provide a cloud computing cluster monitoring method and system based on peak prediction, which aim to overcome the defects in the prior art, and after the functions of collecting, sending, receiving and storing relevant data indexes in each node in a cluster are realized, a front-end interface chart is used to display the acquired data, the situation of time sequence data is accurately predicted, peak data points are acquired in real time, and an early warning function is finally realized, so that the reliability of the whole cluster is improved.
The invention adopts the following technical scheme:
a cloud computing cluster monitoring system based on peak prediction, comprising:
the data acquisition end is responsible for acquiring system performance information and services of each node server and a Docker container, a Ceph distributed file system and the like, and dynamically acquiring the CPU utilization rate, the memory utilization rate, the disk utilization rate, the service state and logs of the current node server, the resource occupation condition of the Docker container, the Ceph running state, cluster information and the like periodically;
the data transmitting and receiving end is responsible for transmitting various information collected by the data acquisition end to the management node, runs on the management node and monitors and receives real-time data transmitted by each node server;
and the data storage end adopts whisper as a storage rear end and stores the data received by the management node by using the time sequence database.
Specifically, the data acquisition end comprises a CPU collection submodule, a memory collection submodule, an I/O collection submodule, a disk collection submodule, a Ceph submodule, a service submodule and a Docker container submodule.
Furthermore, the CPU collection submodule is used for collecting the real-time state of the node CPU, reading/proc/stat files, wherein the/proc/stat files contain information of all CPU activities, and all values in the files are accumulated from the start of the system to the current moment; the memory collection submodule is used for collecting real-time memory information of the nodes and reading/proc/stat files; the I/O collection submodule is used for collecting the I/O state of the cluster and reading/proc/disks files; the disk collection submodule is used for collecting disk information of the cluster and reading/proc/disks files; the Ceph submodule is used for monitoring the state of the Ceph distributed file system; the service submodule is used for detecting the apache server process in the system; the Docker container submodule is used for monitoring the CPU utilization rate, the memory use condition and the network flow of the Docker container.
Specifically, the data transmitting and receiving end includes:
the sending submodule is programmed based on the socket, the current sending submodule continuously sends the collected data to the receiving module through a TCP/IP protocol, the format of the sent message is < metric path > < metric value > < metric status >, and the sent message is called an index item, and the index item is a measurable quantity which changes continuously along with the time;
the receiving submodule is realized on the basis of a Twisted network library, receives index item data transmitted through various protocols and writes the index item data into a disk; when receiving the index item, the value of the index item is cached in the RAM, and is written into the disk by using a bottom Whisper library according to a specified time interval, and after receiving the message, the format of the message is converted and transmitted to the storage module.
Another technical solution of the present invention is a cloud computing cluster monitoring method based on peak prediction, which utilizes the system, including the steps of:
s1, predicting a time sequence containing both trend and seasonality by using a time sequence prediction algorithm based on a cubic exponential smoothing method, and setting a threshold value of each type index;
s2, approximately representing the curve as a series of points by using a Douglas-Pock algorithm, extracting a peak value and comparing the peak value with a threshold value of each index set in advance;
and S3, predicting whether a node in the cluster is about to have a fault or not by combining the monitored numerical indexes.
Specifically, in step S1, the prediction model is:
wherein the content of the first and second substances,
is a predicted value of T + T period, and a cubic exponential smoothing value of T period, at、btAnd ctFor the prediction model smoothing coefficients, T is the period,
in order to once-exponential-smooth the value,
is a second order exponential smoothing value of the data,
is a cubic exponential smoothing value.
Specifically, in step S2, the specific step of using the douglas-pock algorithm is:
s201, connecting a straight line AB between the head point A and the tail point B of the curve, wherein the straight line is a chord of the curve;
s202, obtaining a point C with the maximum distance from the straight line segment on the curve, and calculating the distance d between the point C and the AB;
s203, comparing the distance with a preset threshold value threshold, and if the distance is smaller than the threshold value threshold, taking the straight line segment as the approximation of a curve, and finishing the processing of the curve segment;
s204, if the distance is larger than the threshold value, dividing the curve into two segments of AC and BC by using C, and respectively carrying out 1-3 processing on the two segments of signals;
and S205, when all the curves are processed, sequentially connecting broken lines formed by the dividing points to be used as the approximation of the curves.
Compared with the prior art, the invention has at least the following beneficial effects:
the cloud computing cluster monitoring system based on peak prediction automatically develops a data acquisition system on each node in a cloud computing cluster environment, is completely suitable for the environment of each development board, does not need to install other services, and has the characteristics of pertinence, accuracy and directness.
Further, the data acquisition end aims at different indexes such as: the acquisition scripts of the CPU, the I/O, the disk utilization rate, various services and the like can monitor important and key performance indexes in a common embedded development board, the running condition of the embedded development board can be mastered in time, the dynamic states of various performance indexes can be monitored for a long time, and the acquisition tool is used to greatly reduce the workload of corresponding operation and maintenance personnel, so that the basic situation of a cluster server is mastered from the source.
Further, in order to store and finally display each item of index data acquired by the data acquisition end, the data transmission and receiving end is required to be responsible for transmitting the data to the receiving end for processing and then storing the processed data in the time sequence database whisper. When data is sent, the format of the data is defined as the format of < metric path > so as to be convenient for the identification of a receiving end, and the receiving end writes the data into a disk quickly and efficiently through the data received by a transmission protocol and converts the format of the message to be transmitted to a storage module. The series of processes of sending and receiving firstly ensure the safety of data transmission, secondly accelerate the data transmission efficiency and improve the running speed of the whole system.
The invention discloses a cloud computing cluster monitoring method based on peak value prediction.
In summary, the triple exponential smoothing method and the Douglas-Puck algorithm are combined with the designed embedded cloud computing cluster monitoring system, compared with a general monitoring system, the triple exponential smoothing method and the Douglas-Puck algorithm can reduce the resource utilization rate more simply and effectively, are suitable for the embedded system, and can carry out prediction analysis accurately, so that problems possibly occurring in a cluster can be estimated, the maintenance cost is reduced effectively, and the stability and the reliability of a cluster server are improved.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
FIG. 1 is a schematic view of a monitoring system according to the present invention;
FIG. 2 is a flow chart of the time series prediction of the present invention;
FIG. 3 is an initial plot of disk utilization using an ordered set of points in the Douglas-Puck algorithm;
FIG. 4 is a schematic diagram of a point C on a curve with a maximum distance from the straight line segment (AB) and a distance d between the point C and the straight line segment (AB) being calculated by setting a distance parameter epsilon > 0;
FIG. 5 is a schematic diagram illustrating the above process recursively performed on two curves that have been obtained;
fig. 6 is a graph in which a plurality of peaks are connected.
Detailed Description
In order to efficiently and conveniently detect various important indexes, a monitoring platform integrating three modules of data acquisition, data transmission, data receiving and data storage and combining an index prediction algorithm needs to be realized, and the reliability of the whole system is further improved.
The data stored by the data storage module is time series data, and the situation of the time series data has stability and regularity, so that the time series can be reasonably delayed. Generally, an exponential smoothing method is used for predicting the data, and the basic idea is as follows: the predicted value is a weighted sum of previous observations, with different weights given to different data, with new data given more weight and old data given less weight.
The exponential smoothing method is mainly classified into three types according to different smoothing times: first order exponential smoothing, second order exponential smoothing, and third order exponential smoothing.
Wherein, the first exponential smoothing method is suitable for the time series without obvious trend change; the quadratic exponential smoothing method is a method for once again exponentially smoothing the value obtained by once exponential smoothing, cannot perform independent prediction, and must be matched with the first exponential smoothing method; the method adopts a cubic exponential smoothing method, is suitable for the trend of time sequence data in the cloud computing cluster environment, because the change of the time sequence can generate a quadratic curve trend, but the cubic exponential smoothing method also carries out primary smoothing on the basis of a quadratic smoothing exponential method so as to carry out data prediction.
In order to guarantee cluster security, judgment of a peak value should be performed after a function of data trend prediction is implemented. And compressing a large number of redundant graphic data points by adopting a Douglas-Pock algorithm to extract necessary data points, comparing the extracted data points with a preset threshold, and triggering a system early warning function once the extracted data points exceed the threshold given by a certain index.
Referring to fig. 1, the cloud computing cluster monitoring system based on peak prediction according to the present invention includes three modules:
(1) data acquisition terminal
And the system is responsible for collecting system performance information and services of each node server, a Docker container, a Ceph distributed file system and the like. The method comprises the steps that dynamic collection is carried out on the CPU utilization rate, the memory utilization rate, the disk utilization rate, the service state and the log of a current node server, the resource occupation condition of a Docker container, the Ceph running state, cluster information and the like periodically;
the system comprises seven sub-modules: the system comprises a CPU collection submodule, a memory collection submodule, an I/O collection submodule, a disk collection submodule, a Ceph submodule, a service submodule and a Docker container submodule, wherein each collection submodule is responsible for realizing general numerical parameters in a collection system.
1) The CPU submodule is responsible for collecting the real-time state of a node CPU, reading/proc/stat files, the files contain information of all CPU activities, including indexes of user, system, idle and the like, some items needed in a data set corresponding to each index are specified, after an acquisition period is determined, data of the current time are continuously calculated in the period, and final data are obtained, and all values in the files are accumulated from the start of a system to the current time;
2) the memory submodule collects real-time memory information of the node, reads/proc/stat files, the files contain all memory information, all values in the files are not accumulated from the start of a system to the current moment, therefore, calculation is not needed, and the current values are directly obtained;
3) I/O submodule
Collecting the I/O state of the cluster, and reading a/proc/disks files;
4) magnetic disk sub-module
Collecting disk information of a cluster, and reading a/proc/disks file;
5) ceph submodule
Monitoring the state of the Ceph distributed file system, and reading a Ceph log file to obtain effective information so as to judge whether the state of the Ceph distributed file system is normal;
6) service submodule
The module detects the progress of an apache server in the system and judges the progress through a state code returned when the apache service is opened;
7) container submodule
The module monitors the CPU utilization rate, the memory use condition and the network flow of the Docker container, and acquires log files of various indexes in the container by connecting corresponding interfaces so as to acquire data.
(2) Data transmitting and receiving end
The data receiving end is operated on the management node to monitor and receive real-time data sent by each node server;
the sending submodule is responsible for sending the performance data collected by the data collecting terminal, and the receiving submodule is responsible for receiving the data.
The sending submodule is programmed based on socket, and the current sending submodule sends collected data to the receiving module continuously through the TCP/IP protocol, where the format of the sent message is < metric path > < metric value > < metric status >, called indicator item, and indicator item (metric) is a measurable quantity that changes continuously with time, for example: number of requests per second, request processing time, CPU usage.
The receiving submodule is implemented based on a Twisted web library, and Twisted is a highly scalable and event-driven Python I/O framework. The method can enhance the interactivity between the data receiving end and a large number of clients and reduce the overhead of processing a large number of data flows. The receiving submodule receives the index item data transmitted through various protocols and writes them to the disk with as high efficiency as possible; upon receiving the pointer entry, the values of the pointer entry are cached in RAM and written to disk at specified intervals using the underlying Whisper library.
And after the receiving submodule receives the message, the message format is converted and transmitted to the storage module, and the receiving submodule opens a 2003 port to receive data.
(3) Data storage terminal
Using a time series database to store data received by a management node
The storage back end is used for solving the storage problem, and the whisper is used as the storage back end; whisper can be used in the module either dependent on other components or independently. When designing Whisper, the design idea similar to RRD is adopted, namely, when storing, the database is stored according to a round-robin mode, a user can define a period, and after the period, the following data will overwrite the previous data. Thus, the database is suitable for use in dynamically generating graphical representations and does not require long-term stored data.
Whisper in the present system is a fixed size and fast and reliable database dedicated to storing time series data in the format of data points (datapoints). Wherein, the data points contain three types of information: index item name, metric value, a particular point in the time series (usually a timestamp).
The Whisper database is similar to the basic operation statement of the common SQL Server, and the most basic operations include create, update, fetch and other commands. In Whisper, a new database file is created by using a create statement, a new data point is written into a specific file by using an update statement, and finally, the data point is retrieved by using a fetch statement.
The invention provides a cloud computing cluster monitoring method based on peak value prediction, which comprises the following steps:
s1, the acquisition tool of the data acquisition end performs corresponding acquisition work on various performance indexes, such as CPU, memory, magnetic disk, etc., on each embedded development board in the cluster server;
s2, the data sending end transmits the collected data to the data receiving end through a TCP/IP protocol, and the data receiving end converts the received data format into a message format which can be identified by the data storage end;
and S3, the data storage end stores the data received by the data receiving end in the whisper time sequence database, provides a corresponding front-end interface, acquires the data transmitted by the interface through writing a script and finally displays the data on the front end.
And predicting the time sequence containing both trend and seasonality by using a cubic exponential smoothing prediction algorithm based on a primary exponential smoothing method and a secondary exponential smoothing method according to the characteristics of the acquired data.
The calculation formula is as follows:
the prediction model is:
taking the index of the utilization rate of the disk as an example, the average value of the first three data in the period is selected as a first smooth value
And the second smoothing value
Since the tendency of the actual data sequence of the disk usage rate varies significantly, the weight coefficient (smoothing coefficient) a is preferably not too small.
Will calculate the initial
And
substituting into a calculation formula to obtain an initial cubic smoothness value
The three calculated smoothed values are substituted into respective calculation expressions (where t is set to 10min) in the prediction model to obtain at、bt、ctFinally, a predicted value is obtained
Other data indexes can be used to obtain the predicted value in the same way, and the flow is shown in fig. 2.
In order to prevent the cluster server from being out of order and pre-process the impending problem of the server to prevent the cluster from crashing, the peak value is continuously extracted in the prediction process and compared with the threshold value of each index set in advance, so as to judge whether a certain node is about to have the problem. To solve this problem, a curve is approximately represented as a series of points using the Douglas-pock algorithm and the number of points is reduced, the classical Douglas-Peucker algorithm is described as follows:
(1) a straight line AB is connected between the head point A and the tail point B of the curve, and the straight line is a chord of the curve;
(2) obtaining a point C with the maximum distance from the straight line segment on the curve, and calculating the distance d between the point C and the AB;
(3) comparing the distance with a preset threshold value threshold, if the distance is smaller than the threshold value threshold, taking the straight line segment as the approximation of a curve, and finishing the processing of the curve segment;
(4) if the distance is larger than the threshold value, dividing the curve into two segments of AC and BC by using C, and respectively carrying out 1-3 processing on the two segments of the credit;
(5) when all the curves are processed, the broken lines formed by all the dividing points are connected in sequence, and the broken lines can be used as the approximation of the curves.
And selecting the utilization rate of the disk to perform an experiment by combining numerical indexes to be monitored in the system, and acquiring a peak value in a specified period by using the algorithm. First, an initial curve of disk utilization for a series of ordered sets of points is shown in FIG. 3.
A distance parameter epsilon is set to be more than 0, and a straight line AB is connected between the head point A and the tail point B of the curve at the beginning. The algorithm automatically stores the head point and the tail point into a result point set. The point C on the curve that is the greatest distance from the straight line segment (AB) is found and its distance d from AB is calculated, as shown in fig. 4.
If the straight line segment AB is used as an approximation of the original curve, point C is clearly the point on the curve that is farthest from AB. Now the distance d is compared with a predetermined threshold epsilon, and if smaller than epsilon, this indicates that any currently unmarked point can be discarded, since the straight line segment that has been obtained is not earlier than epsilon as an approximation of the curve, i.e. the curve segment is processed; if the distance d is greater than the threshold, the curve is divided into two segments AC and BC by C, and the point C is noted, and then the above-mentioned processing is recursively performed on the two segments of the curve that have been obtained, respectively, as shown in FIG. 5.
When all the curves are processed, the broken lines formed by all the dividing points are connected in sequence, and the broken lines can be used as the approximation of the curves. The resulting curve is shown in fig. 6, which is a connection of a plurality of peaks.
The data points processed by the algorithm all belong to peaks that may cause failure.
The above-mentioned contents are only for illustrating the technical idea of the present invention, and the protection scope of the present invention is not limited thereby, and any modification made on the basis of the technical idea of the present invention falls within the protection scope of the claims of the present invention.
Claims (7)
1. A cloud computing cluster monitoring system based on peak prediction, comprising:
the data acquisition end is responsible for acquiring system performance information and services of each node server and a Docker container, a Ceph distributed file system and the like, and dynamically acquiring the CPU utilization rate, the memory utilization rate, the disk utilization rate, the service state and logs of the current node server, the resource occupation condition of the Docker container, the Ceph running state, cluster information and the like periodically;
the data transmitting and receiving end is responsible for transmitting various information collected by the data acquisition end to the management node, runs on the management node and monitors and receives real-time data transmitted by each node server;
and the data storage end adopts whisper as a storage rear end and stores the data received by the management node by using the time sequence database.
2. The system of claim 1, wherein the data collection end comprises a CPU collection submodule, a memory collection submodule, an I/O collection submodule, a disk collection submodule, a Ceph submodule, a service submodule, and a Docker container submodule.
3. The system of claim 2, wherein the CPU collection submodule is configured to collect the real-time status of the node CPU, and read a/proc/stat file, where the/proc/stat file contains information about all CPU activities, and all values in the file are accumulated from the start of the system to the current time; the memory collection submodule is used for collecting real-time memory information of the nodes and reading/proc/stat files; the I/O collection submodule is used for collecting the I/O state of the cluster and reading/proc/disks files; the disk collection submodule is used for collecting disk information of the cluster and reading/proc/disks files; the Ceph submodule is used for monitoring the state of the Ceph distributed file system; the service submodule is used for detecting the apache server process in the system; the Docker container submodule is used for monitoring the CPU utilization rate, the memory use condition and the network flow of the Docker container.
4. The peak prediction-based cloud computing cluster monitoring method and system according to claim 1, wherein the data sending and receiving end comprises:
the sending submodule is programmed based on the socket, the current sending submodule continuously sends the collected data to the receiving module through a TCP/IP protocol, the format of the sent message is < metric path > < metric value > < metric status >, and the sent message is called an index item, and the index item is a measurable quantity which changes continuously along with the time;
the receiving submodule is realized on the basis of a Twisted network library, receives index item data transmitted through various protocols and writes the index item data into a disk; when receiving the index item, the value of the index item is cached in the RAM, and is written into the disk by using a bottom Whisper library according to a specified time interval, and after receiving the message, the format of the message is converted and transmitted to the storage module.
5. A cloud computing cluster monitoring method based on peak prediction, characterized in that, with the system of claim 1, the method comprises the following steps:
s1, predicting a time sequence containing both trend and seasonality by using a time sequence prediction algorithm based on a cubic exponential smoothing method, and setting a threshold value of each type index;
s2, approximately representing the curve as a series of points by using a Douglas-Pock algorithm, extracting a peak value and comparing the peak value with a threshold value of each index set in advance;
and S3, predicting whether a node in the cluster is about to have a fault or not by combining the monitored numerical indexes.
6. The method according to claim 5, wherein in step S1, the prediction model is:
wherein the content of the first and second substances,
is a predicted value of T + T period, and a cubic exponential smoothing value of T period, at、btAnd ctFor the prediction model smoothing coefficients, T is the period,
in order to once-exponential-smooth the value,
is a second order exponential smoothing value of the data,
is a cubic exponential smoothing value.
7. The method according to claim 5, wherein in step S2, the Douglas-Pock algorithm is specifically used:
s201, connecting a straight line AB between the head point A and the tail point B of the curve, wherein the straight line is a chord of the curve;
s202, obtaining a point C with the maximum distance from the straight line segment on the curve, and calculating the distance d between the point C and the AB;
s203, comparing the distance with a preset threshold value threshold, and if the distance is smaller than the threshold value threshold, taking the straight line segment as the approximation of a curve, and finishing the processing of the curve segment;
s204, if the distance is larger than the threshold value, dividing the curve into two segments of AC and BC by using C, and respectively carrying out 1-3 processing on the two segments of signals;
and S205, when all the curves are processed, sequentially connecting broken lines formed by the dividing points to be used as the approximation of the curves.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911304713.6A CN111190790A (en) | 2019-12-17 | 2019-12-17 | Cloud computing cluster monitoring method and system based on peak prediction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911304713.6A CN111190790A (en) | 2019-12-17 | 2019-12-17 | Cloud computing cluster monitoring method and system based on peak prediction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111190790A true CN111190790A (en) | 2020-05-22 |
Family
ID=70707414
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911304713.6A Pending CN111190790A (en) | 2019-12-17 | 2019-12-17 | Cloud computing cluster monitoring method and system based on peak prediction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111190790A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111709623A (en) * | 2020-06-04 | 2020-09-25 | 中国科学院计算机网络信息中心 | High-performance computing environment evaluation method and device, electronic equipment and storage medium |
| CN112162821A (en) * | 2020-09-25 | 2021-01-01 | 中国电力科学研究院有限公司 | Container cluster resource monitoring method, device and system |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103825779A (en) * | 2014-02-21 | 2014-05-28 | 南京邮电大学 | Method for monitoring state of cloud data center |
| US9355357B2 (en) * | 2011-10-21 | 2016-05-31 | Hewlett Packard Enterprise Development Lp | Computing predicted data according to weighted peak preservation and time distance biasing |
| CN105912436A (en) * | 2015-09-17 | 2016-08-31 | 乐视网信息技术（北京）股份有限公司 | Method and device for predicting system computing resource based on exponential smoothing prediction |
| CN108038040A (en) * | 2017-12-08 | 2018-05-15 | 上海市信息网络有限公司 | Computer cluster performance indicator detection method, electronic equipment and storage medium |
| CN108200156A (en) * | 2017-12-29 | 2018-06-22 | 南京邮电大学 | The dynamic load balancing method of distributed file system under a kind of cloud environment |
| CN108920153A (en) * | 2018-05-29 | 2018-11-30 | 华南理工大学 | A kind of Docker container dynamic dispatching method based on load estimation |
| CN109117352A (en) * | 2017-06-23 | 2019-01-01 | 华为技术有限公司 | Server performance prediction technique and device |
| US10191778B1 (en) * | 2015-11-16 | 2019-01-29 | Turbonomic, Inc. | Systems, apparatus and methods for management of software containers |
-
2019
- 2019-12-17 CN CN201911304713.6A patent/CN111190790A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9355357B2 (en) * | 2011-10-21 | 2016-05-31 | Hewlett Packard Enterprise Development Lp | Computing predicted data according to weighted peak preservation and time distance biasing |
| CN103825779A (en) * | 2014-02-21 | 2014-05-28 | 南京邮电大学 | Method for monitoring state of cloud data center |
| CN105912436A (en) * | 2015-09-17 | 2016-08-31 | 乐视网信息技术（北京）股份有限公司 | Method and device for predicting system computing resource based on exponential smoothing prediction |
| US10191778B1 (en) * | 2015-11-16 | 2019-01-29 | Turbonomic, Inc. | Systems, apparatus and methods for management of software containers |
| CN109117352A (en) * | 2017-06-23 | 2019-01-01 | 华为技术有限公司 | Server performance prediction technique and device |
| CN108038040A (en) * | 2017-12-08 | 2018-05-15 | 上海市信息网络有限公司 | Computer cluster performance indicator detection method, electronic equipment and storage medium |
| CN108200156A (en) * | 2017-12-29 | 2018-06-22 | 南京邮电大学 | The dynamic load balancing method of distributed file system under a kind of cloud environment |
| CN108920153A (en) * | 2018-05-29 | 2018-11-30 | 华南理工大学 | A kind of Docker container dynamic dispatching method based on load estimation |
Non-Patent Citations (1)
| Title |
|---|
| 李万涛等: "《运筹学》", 31 March 2018, 中国铁道出版社 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111709623A (en) * | 2020-06-04 | 2020-09-25 | 中国科学院计算机网络信息中心 | High-performance computing environment evaluation method and device, electronic equipment and storage medium |
| CN112162821A (en) * | 2020-09-25 | 2021-01-01 | 中国电力科学研究院有限公司 | Container cluster resource monitoring method, device and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111984499B (en) | Fault detection method and device for big data cluster | |
| CN115186883A (en) | Industrial equipment health state monitoring system and method based on Bian Yun collaborative computing | |
| CN110147470B (en) | Cross-machine-room data comparison system and method | |
| CN108519942A (en) | A kind of Log Analysis System | |
| CN113626198B (en) | Database flow load balancing system and method | |
| CN111190790A (en) | Cloud computing cluster monitoring method and system based on peak prediction | |
| CN112084224A (en) | Data management method, system, device and medium | |
| CN109034580A (en) | A kind of information system holistic health degree appraisal procedure based on big data analysis | |
| CN114154035A (en) | Data processing system for dynamic loop monitoring | |
| CN113487086A (en) | Method and device for predicting remaining service life of equipment, computer equipment and medium | |
| CN106649034B (en) | Visual intelligent operation and maintenance method and platform | |
| CN110532153A (en) | A kind of business level user's operation experience visualization system | |
| CN104021079B (en) | Real-time monitoring and analyzing system and method for host system data | |
| CN115080363B (en) | System capacity evaluation method and device based on service log | |
| CN114625805B (en) | Return test configuration method, device, equipment and medium | |
| CN116069618A (en) | Application scene-oriented domestic system evaluation method | |
| CN110647070A (en) | Power environment monitoring system for super-large-scale data center | |
| CN115776449A (en) | Train Ethernet communication state monitoring method and system | |
| CN114756376A (en) | Cross-domain multi-service resource automatic adaptation maintenance platform | |
| CN109766243B (en) | Multi-core host performance monitoring method based on power function | |
| CN111245632A (en) | Safe log storage and management system based on cloud analysis | |
| CN115033457B (en) | Multi-source data real-time acquisition method and system capable of monitoring and early warning | |
| CN110413431B (en) | Intelligent identification early warning method for large data platform fault | |
| CN114140032A (en) | Facility running state monitoring method, device, equipment and storage medium | |
| CN117762741A (en) | Sliding window-based method for realizing statistics of micro-service application resource index data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200522 |
|
| RJ01 | Rejection of invention patent application after publication |