CN104407910A - Virtualization server performance monitoring method and system - Google Patents
Virtualization server performance monitoring method and system Download PDFInfo
- Publication number
- CN104407910A CN104407910A CN201410598698.1A CN201410598698A CN104407910A CN 104407910 A CN104407910 A CN 104407910A CN 201410598698 A CN201410598698 A CN 201410598698A CN 104407910 A CN104407910 A CN 104407910A
- Authority
- CN
- China
- Prior art keywords
- test
- virtual machine
- performance
- server
- client
- 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
Abstract
The invention discloses a virtualization server performance monitoring method, which comprises the following steps that a virtual machine server side starts socket monitoring in a server thread pool, and a linking request of a client side is monitored; a test machine client side uses TCPs (transmission control protocols), UDP (user datagram protocols) or TCP/UDP hybrid network protocols to carry out socket communication with the virtual machine server side, a test task is executed, a test result is collected, and the test result is sent to a main control machine; a main control machine manager sends a test command to a client side test machine, and distributed test results are collected and counted; the main control machine manager sends a RPC (remote procedure call) request to a virtual machine, and various performance indexes of the virtual machine are obtained and are displayed in a chart form. The virtualization server performance monitoring method and a virtualization server performance monitoring system have the advantages that the network performance test and evaluation is realized in virtualization environment, and in addition, the RPC is provided for remotely monitoring various operation system performance indexes of the virtual machine in real time.
Description
Technical field
The present invention relates to virtual server field, particularly a kind of monitoring method of virtualized server performance and system.
Background technology
In recent years along with the development of computer hardware makes powerful and relatively cheap multiprocessor servers become main flow, but in many cases, the performance of these hardware all can not be fully utilized, and popularizing in recent years along with multi-core processor, also make this insufficiency more serious, this makes industry more and more pay attention to Intel Virtualization Technology.
Virtual (Virtualization) technique functions comes from the IBM large computer system of the sixties in 20th century, come into vogue gradually in System 370 series of the seventies, these machines generate many virtual machine (Virtual Machine) examples that can run SOS software by monitor of virtual machine (Virtual Machine Monitor, VMM) on physical hardware.Virtualized environment provides a software layer, make user can create multiple independently virtual machine in same physical server, although run the expense that multiple virtual machine has some virtual aspects simultaneously, a station server can be made to obtain higher utilization rate.Although the underlying physical resource of these virtual machines is shared, each virtual machine completely independently, and performs an independent operating system and independent application software.The virtual technology as making full use of server performance has been an irreversible trend now, by different Intel Virtualization Technologies, can effectively reduce physical server quantity, and in fact the quantity reducing physical server also correspondingly reduce server cost.[20] Intel Virtualization Technology is utilized to devise automated performance testing framework in [21] document, workload generator is loaded in above the virtual machine of different dispersions by this testing tool, thus effectively improve testing efficiency and improve load pressure, for software performance test brings great convenience, reduce testing cost.Meanwhile, by virtual, client also reaches the effect [6] of the utilization factor improving server resource.And the computing velocity that now computing machine is more and more faster and more powerful Intel Virtualization Technology flexibly also create more chance for virtualized application, but how to weigh server virtualization performance be a difficult problem [2].
Traditional server reference tests the inspection only laid particular emphasis on single workload, and cannot obtain the performance of server when supporting the performance of multiple dummy machine system or support multiple workload on the same server simultaneously.More appropriate virtual machine benchmark test should use and run on reality in multiple operating system and the performance of various workload to server is checked, and this benchmark test should produce a understandable index [2].Standard performance evaluation and test company (standard performance evaluationcorporation, SPEC) specializes in virtualization solution evaluating method [8] a committee was called into being the end of the year 2006.But no matter being in academia or in industry member, is the best method of evaluation and test virtualization solution for which type of method, never reach common understanding [4].
The reason of these differences is determined by virtual technical characteristic own:
First virtual is the term of a broad sense, and different applications represents different implications.In computer science, virtual to represent computational resource abstract, and be not only confined to the concept of virtual machine.In general Intel Virtualization Technology is mainly divided into following large class [1]:
1) platform virtual (Platform Virtualization), virtual for computing machine and operating system;
2) resource virtualizing (Resource Virtualization), virtual for specific system resource, such as internal memory, storage, Internet resources etc.
3) application program virtual (Application Virtualization), comprises emulation, simulation, interpretation technique etc.
Secondly, platform Intel Virtualization Technology can be subdivided into again fully virtualized, partial virtual, hardware and assists virtual etc. [3].
Fully virtualizedly refer to that virtual machine simulates complete bottom hardware, comprise processor, physical memory, clock, peripheral hardware etc., make the operating system for original hardware designs or other system software not make any amendment completely and just can run in virtual machine.Can regard as alternately between operating system and real hardware is undertaken by a prespecified hardware interface, and fully virtualized VMM provides whole interface in the mode of full simulation hardware, and relatively more typical representative has Vmware, VirtualBox etc.
Partial virtualization only simulates part bottom hardware, merely provide the simulation to key computation module or instruction set, therefore Client OS is not made an amendment and cannot be run in virtual machine, other program may also need to modify, popular local virtual VMM has Xen, Hyper-V.
Fully virtualized and partial virtual also has different impacts to the overall performance of virtual machine itself as can be seen here, sees document [5].The people such as Du Chunyang [6] are tested than the performance of right several different Intel Virtualization Technologies in high performance computing system and Clustering by the symmetric multi-processors performance of NPB, in single node, serial and parallel performance is tested based on NPB, and compare with the NPB benchmark of ecosystem, draw and do not had any Intel Virtualization Technology perfectly can reach the conclusion of the performance of primary system and analyze the performance of different VMM in various criterion roughly.But test benchmark derives from the software of fluid dynamics application in literary composition, not there is ubiquity and representativeness.
Therefore, meaning and practical value is got a good eye for the virtual machine method of testing that the different choice of concrete enterprise application is suitable.
Although virtual machine with it at subregion, isolation, and encapsulation on technical advantage be obtained in a lot of field widely apply [7], the operation of virtual machine in physical machine still can consume part physical resource.Virtual machine is that VMM needs, for each virtual machine distributes a set of data structure to manage their states, to comprise a complete set of register of virtual processor, the service condition of physical memory, state of virtual unit etc. to the abstract of true computing environment and simulation.During VMM scheduling virtual machine, its partial status is returned in host computer system.And the state of not all needs to recover, what such as, deposit in host C R3 register is the page table physical address that VMM is arranged, instead of the value that Guest OS is arranged.Host-processor directly runs the machine instruction of Guest OS, because Guest OS operates in low privilege level, when accessing host computer system privileged mode (as writing GDT register), it is abnormal that insufficient permission causes host-processor to produce, and operation power is given back VMM automatically.VMM may need first to write back in status data structure by the current state of this virtual machine, and then the reason that analyzing virtual machine is suspended represents Guest OS and perform corresponding privileged operation.Operation power is returned Guest OS by last VMM, and Guest OS continues to perform from interrupted place last time, or the process virtual interrupt that is absorbed in of VMM and exception.As can be seen here in most cases, VMM needs just can complete shirtsleeve operation originally through complicated flow process, and this can have a huge impact the application system overall performance run in virtual machine.
Enterprise or scientific research institution, when Intel Virtualization Technology of a great variety, respectively to have feature, often compare the performance issue paying close attention to Intel Virtualization Technology.First, whether the performance that virtualized various technology realizes providing exists very large difference; Secondly, application program runs have how many with the difference run directly in physical machine in performance on a virtual machine.
The people such as Lorenzo Martignoni [9] are undertaken carrying out different between physical machine of more different virtual machines based on the fuzz testing of particular protocol and difference analysis by the User space respectively under physical machine and virtual machine environment and kernel mode two kinds of operational modes, but this method of testing center of gravity has been placed on the difference above physical machine and virtual machine function more, the difference shorter mention in performance.
Because dummy machine system uses traditional memory module to cause virtual disk storage aspect to there is poor performance, the people such as Dongsheng Yin [10] store Experimental comparison by carrying out actual situation to VMware vSphere4.0 virtual machine, give a kind of based on SSD and the I/O storage optimization scheme based on the RAID array shared.Memory module just for virtual machine and physical machine in literary composition conducts in-depth research, and solves the poor problem of virtual machine memory property emphatically, therefore can not carry out overall performance analysis to virtual machine and physical machine on the whole from virtual machine itself.
By document [11], we can find out not enough to the research of the complete load testing of virtual machine at present.The people such as Marco A.S.Netto [12] use Loadrunner performance testing tool tentatively to attempt the virtual machine under different CPU pattern and physical machine load testing, and have carried out detailed virtual-real comparison.But this experiment is confined to the virtual-real comparison under Tomcat compact applications and light-load conditions, do not have the contrast to physical machine under high load condition and virtual machine, and experiment condition also only has this set condition of Xen.
Jianhua Che [13] etc. use CPU2006, RAMSPEED, the standard mechanism test procedures such as Bonnie++ are to OpenVZ, virtual based on container (Container-Based) of Xen and KVM these three kinds employings respectively of increasing income, local virtual, the VMM instrument of Full-virtualization strategy has carried out Black-box Testing, its test index includes the CPU of virtual machine, internal memory, disk I/O, the system operation context blocked operation frequency of network and finer microoperation rank, finally by excellent cruel from its Code Design of angle analysis of white-box testing of test result data.Wherein studying angle is more that the angle standing in VMM deviser carrys out test analysis, provides good suggestion to the design improving VMM, but limited to the directive significance under actual application environment.
The people such as Fares N.Almari [14] compared for the performance class expansion of Orcale database under Orcale VM and Vmware ESXi virtualized environment to verify that OrcaleVm that on market, Orcale declares is whether the saying of the several times of Vmware ESXi performance is correct, result draws and selects virtualization product should according to concrete application demand by experiment, and particularly the application of database should carry out the kind of absolute ESXi according to the height of readwrite bandwidth and transaction rate.But this experiment mainly for the performance comparison of this kind of database of Orcale database, has significant limitation, do not relate to the performance evaluation under enterprise-level application servers loading condition.
People such as Deshane [15] [17] is to the overall performance of Xen and KvM, and isolation, expansibility etc. have carried out the comparison of some row, and object is the Optimizing Suggestions in order to provide design to VMM developer, for commercial user provides the suggestion selecting virtual machine.
The benchmark of people's standards such as Walters [16] has carried out detailed contrast to different classes of VMM, include network design, SMP performance, the availability etc. of file system and MPI, but the focus of this test is virtual machine operating characteristic under high-performance (HPC) design conditions compares.
Daniel Schlosser [18] compares OpenVZ in detail, KVM, Xen v4, the performance of this network bandwidth of monitor of virtual machine under commercial hardware environment in 5 of VirtualBox, VMware ESXi.
Intel Virtualization Technology is as the important ingredient of cloud computing, the people such as Dionisio Leite [19] have studied the performance difference of Intel Virtualization Technology under cloud environment, in equally loaded situation, load testing is carried out to host and virtual machine, and maximum quantity when having quantized Xen and KVM parallel running under certain condition.
Document [22] devises micro benchmark test frame to test the isolation of virtual machine.
Document [23] proposes performance test framework vTestKit under a kind of single VM scene and many VM scene to test the overall performance of virtual machine, because framework is using one group of benchmark program as standard, although can dynamically can add and revise testing standard for ceaselessly application, this test frame some characteristics remained for VMM itself carry out assessing.
Thus, we can find out the domestic and international assessment having focused on the performance of itself between different VMM to virtualizing performance research and compare, contrast between virtual machine and physical machine is then confined to some and specifically applies, and the performance comparison problem therefore between virtual machine and physical machine awaits further research.Relation simultaneously owing between physical machine and virtual machine being one-to-many, different application, particularly CPU is intensive, IO is intensive, network-intensive to be applied in the emphasis utilized physical resource under different combination conditions be different, how different virtual robot arm is combined in and ensures that the condition of overall applicability performance improves physical resource utilization factor substantially and has very strong actual application value, and the research of this respect is still in commitment at present.
List of references:
[1]http://en.wikipedia.org/wiki/Virtualization
[2]Makhija V,Herndon B,Smith P,et al.VMmark:A scalable benchmarkfor virtualized systems[J].VMware Inc,CA,Tech.Rep.VMware-TR-2006-002,2006.
[3]Seetharaman S,Krishna Murthy B V S.Test optimization usingsoftware virtualization[J].Software,IEEE,2006,23(5):66-69.
[4] Lan Yuqing, Song Xiaoyu, horse Garrick, etc. system virtualization technical feature evaluation and test [J]. telecommunication science, 2010,1.
[5]http://wenku.baidu.com/link？url＝-v2Lnd5MOOxRqKpGXkkfLdym9IUA_8e3DzcqxaD_mc
Z-Es8prg4gvHTS-sUaPLPh3sfjbPEE0WLGsGX52bnEOaQ2fWEeswE8_mR8SjKyljO
[6] Du Chunyang, Ma Hong, Xia Junbo. based on Intel Virtualization Technology Performance Detection and the analysis [J] of the universal standard. the communication technology, 2009,5:061.
[7]Martignoni L,Paleari R,Fresi Roglia G,et al.Testing systemvirtual machines[C].In Proceedings of the 19th international symposiumon Software testing and analysis.ACM,2010:171-182.
[8]Standard Performance Evaluation Corporation.SpecVirtualization Committee,April 2008
[9]Testing System Virtual Machines,ISSTA’10,July 12–16,2010,Trento,Italy.Copyright 2010ACM 978-1-60558-823-0/10/07
[10]Yin D,Liang J.A research on storage performance optimizationof virtual machine[C].In the 6th International Conference on.IEEE,2011:547-551.
[11]Gaisbauer S,Kirschnick J,Edwards N,et al.Vats:Virtualized-aware automated test service[C].In the Fifth InternationalConference on.IEEE,2008:93-102.
[12]Netto M A S,Menon S,Vieira H V,et al.Evaluating loadgeneration in virtualized environments for software performancetesting[C].Parallel and Distributed Processing Workshops and Phd Forum(IPDPSW),2011IEEE International Symposium on.IEEE,2011:993-1000.
[13]Che J,Yu Y,Shi C,et al.A synthetical performance evaluationof openvz,xen and kvm[C].Services Computing Conference(APSCC),2010IEEE Asia-Pacific.IEEE,2010:587-594.
[14]Almari F N,Zavarsky P,Ruhl R,et al.Performance Analysis ofOracle Database in Virtual Environments[C].Advanced InformationNetworking and Applications Workshops(WAINA),201226th InternationalConference on.IEEE,2012:1238-1245.
[15]Deshane T,Shepherd Z,Matthews J,et al.Quantitativecomparison of Xen and KVM[J].Xen Summit,Boston,MA,USA,2008:1-2.
[16]Chaudhary V,Cha M,Walters J P,et al.A comparison ofvirtualization technologies for HPC[C].Advanced Information Networkingand Applications,2008.AINA 2008.22nd International Conference on.IEEE,2008:861-868.
[17]Tope I E,Zavarsky P,Ruhl R,et al.Performance Evaluation ofOracle VM Server Virtualization Software 64Bit LinuxEnvironment[C].Security Measurements and Metrics(Metrisec),2011ThirdInternational Workshop on.IEEE,2011:51-57.
[18]Schlosser D,Duelli M,Goll S.Performance comparison of hardwarevirtualization platforms[M].NETWORKING 2011.Springer Berlin Heidelberg,2011:393-405.
[19]Ye K,Che J,Jiang X,et al.vtestkit:A performance benchmarkingframework for virtualization environments[C].ChinaGrid Conference(ChinaGrid),2010Fifth Annual.IEEE,2010:130-136.
[20]Kim G H,Kim Y G,Chung K Y.Towards virtualized and automatedsoftware performance test architecture[J].Multimedia Tools andApplications,2013:1-15.
[21]Kim G,Moon H,Song G P,et al.Software performance testingscheme using virtualization technology[C].Ubiquitous InformationTechnologies&Applications,2009.ICUT'09.Proceedings of the 4thInternational Conference on.IEEE,2009:1-5.
[22]Yuan P,Ding C,Cheng L,et al.VITS Test Suit:A Micro-Benchmarkfor Evaluating Performance Isolation of VirtualizationSystems[C].e-Business Engineering(ICEBE),2010IEEE 7th InternationalConference on.IEEE,2010:132-139.
[23]Ye K,Che J,Jiang X,et al.vtestkit:A performance benchmarkingframework for virtualization environments[C].ChinaGrid Conference(ChinaGrid),2010Fifth Annual.IEEE,2010:130-136.
Summary of the invention
The object of the invention is to overcome the shortcoming of prior art and deficiency, a kind of monitoring method of virtualized server performance is provided.
Another object of the present invention is to the monitoring system that a kind of virtualized server performance is provided.
Object of the present invention is realized by following technical scheme:
A monitoring method for virtualized server performance, comprises the step of following order:
S1. virtual machine Server end starts socket monitoring in Server thread pool, monitors the linking request of client;
S2. test machine Client holds to use TCP procotol, UDP procotol or TCP/UDP hybrid network agreement and virtual machine server holds to carry out Socket and communicates, and performs test assignment, collection test result, and to main control computer transmission test result;
S3. main control computer Manager sends test command to client test machine, collects and statistical cloth test test result;
S4. main control computer Manager sends RPC request to virtual machine, obtains virtual machine property indices, and graphically displays.
In step S3, described collection and statistical cloth test test result be specially: main control computer Manager gathers the test result data of a client test machine every 3s, and different image data is averaged, count current Socket transfer rate.
In step S3, described collection distributed testing test result specifically comprises following steps:
The IP address of a, RPC client according to destination server and the default program port one 11 of portmapper, by Socket socket structure udp request, transmit a request to the program end slogan of the portmapper application asks rstatd of target machine;
After b, portmapper application program receives request, the mapping table between inquiry port and application program, and the port numbers of rstatd watchdog routine is returned RPC client;
After c, RPC client obtains the port numbers of rstatd program, socket Socket re-constructs new Udp according to former target ip address and rstatd in port numbers and asks, and sends " request service " to rstatd watchdog routine;
After d, Rstatd watchdog routine receives " request service ", return the performance data that current performance counter is collected.
In step S4, described acquisition virtual machine property indices, and graphically display, be specially:
A, the performance data in test process to be kept in DataTable object, and by data transmission real-time update that the monitoring of Linux remote performance obtains;
B, after test terminates, statistical summaries is carried out to the performance data in DataTable, obtain final performance test data;
C, then utilize XSL to define the pattern of data exhibiting, then according to certain rule, test data is write in XML file;
D, last, by the XslCompiledTransform class under System.Xml.Xsl NameSpace, the data file of XML format is changed into the user-friendly Html page.
Another object of the present invention is realized by following technical scheme:
A kind of monitoring system of virtualized server performance, comprise interconnective virtual machine Server end, test machine Client end, main control computer Manager between two, wherein virtual machine Server end starts socket monitoring in Server thread pool, monitors the linking request of client; Test machine Client holds to use the procotol such as TCP, UDP and virtual machine server holds to carry out Socket and communicates, and performs test assignment, collection test result, and to main control computer transmission test result; Main control computer Manager sends test command to client test machine, collects and statistical test result; Send RPC request to virtual machine, obtain virtual machine property indices, and graphically display.
Compared with prior art, tool has the following advantages and beneficial effect in the present invention:
1, the Intel Virtualization Technology that the present invention studies refers in particular to the performance evaluating under platform Intel Virtualization Technology, by using control program (Control Program, be also referred to as Virtual Machine Monitor or Hypervisor), hide the real physical characteristics of specific calculation platform, for user provides computing environment (being called virtual machine) that is abstract, unified, simulation.The operating system run in virtual machine is called as Client OS (Guest OS), and the operating system running monitor of virtual machine is called as host operating system (Host OS).
2, the present invention to pay close attention under virtualized environment for the exploitation of network performance evaluation and test and realization, devises applied in network performance test framework general under physical machine and virtual machine environment.This software has following features: support Multi net voting host-host protocol, multithreading distributed testing, system performance index remote real-time monitoring.Specifically comprise:
(1) statistics is based on the network transmission speed of VMM
The network transmission protocol that different application uses is different, because herein is provided for the virtual machine network Performance Detection mechanism under different host-host protocols, and the three kinds of Internet Transmission modes supporting TCP, UDP and TCP/UDP to mix at present.
(2) multithreading distributed testing
Test assignment is deployed in different distributed systems, by the unified management of Host main frame.Each test assignment supports multi-threaded parallel operation, real-time statistics virtual machine network transmission performance, and shows in the mode of chart.
(3) system remote performance monitoring
In test assignment implementation, main control computer can carry out real-time system monitoring to tested test-run a machine property indices, in order to auxiliary evaluation virtual machine network performance.
Accompanying drawing explanation
Fig. 1 is the process flow diagram that the system call of RPC client remote obtains Linux performance parameter;
Fig. 2 is the process flow diagram that report data represents processing procedure;
Fig. 3 is the schematic diagram collecting distributed testing test data;
Fig. 4 is the structural representation of the monitoring system of a kind of virtualized server performance of the present invention.
Embodiment
Below in conjunction with embodiment and accompanying drawing, the present invention is described in further detail, but embodiments of the present invention are not limited thereto.
A monitoring method for virtualized server performance, comprises the step of following order:
S1. virtual machine Server end starts socket monitoring in Server thread pool, monitors the linking request of client;
S2. test machine Client holds to use TCP procotol, UDP procotol or TCP/UDP hybrid network agreement and virtual machine server holds to carry out Socket and communicates, and performs test assignment, collection test result, and to main control computer transmission test result;
S3. main control computer Manager sends test command to client test machine, collects and statistical cloth test test result; Described collection and statistical cloth test test result be specially: main control computer Manager gathers the test result data of a client test machine every 3s, and different image data is averaged, and counts current Socket transfer rate;
As Fig. 3, described collection distributed testing test result specifically comprises following steps:
The IP address of a, RPC client according to destination server and the default program port one 11 of portmapper, by Socket socket structure udp request, transmit a request to the program end slogan of the portmapper application asks rstatd of target machine;
After b, portmapper application program receives request, the mapping table between inquiry port and application program, and the port numbers of rstatd watchdog routine is returned RPC client;
After c, RPC client obtains the port numbers of rstatd program, socket Socket re-constructs new Udp according to former target ip address and rstatd in port numbers and asks, and sends " request service " to rstatd watchdog routine;
After d, Rstatd watchdog routine receives " request service ", return the performance data that current performance counter is collected;
A, b, c, d step by step above (1), (2), (3) in Fig. 3, (4) difference correspondence;
S4. main control computer Manager sends RPC request to virtual machine, obtains virtual machine property indices, and graphically displays;
As Fig. 2, described acquisition virtual machine property indices, and graphically display, be specially:
A, the performance data in test process to be kept in DataTable object, and by data transmission real-time update that the monitoring of Linux remote performance obtains;
B, after test terminates, statistical summaries is carried out to the performance data in DataTable, obtain final performance test data;
C, then utilize XSL to define the pattern of data exhibiting, then according to certain rule, test data is write in XML file;
D, last, by the XslCompiledTransform class under System.Xml.Xsl NameSpace, the data file of XML format is changed into the user-friendly Html page.
As Fig. 4, a kind of monitoring system of virtualized server performance, comprise interconnective virtual machine Server end, test machine Client end, main control computer Manager between two, wherein virtual machine Server end starts socket monitoring in Server thread pool, monitors the linking request of client; Test machine Client holds to use the procotol such as TCP, UDP and virtual machine server holds to carry out Socket and communicates, and performs test assignment, collection test result, and to main control computer transmission test result; Main control computer Manager sends test command to client test machine, collects and statistical test result; Send RPC request to virtual machine, obtain virtual machine property indices, and graphically display.
Monitoring system of the present invention passes through the managing distributed test machine of master controller, to client test machine allocation for test task; Client is connected by LAN (Local Area Network) and virtual machine, performs test assignment, and returns real-time testing result to main control computer.Master controller sends RPC request to vmm virtual machine, obtains virtual machine property indices.Below in conjunction with Fig. 4, each ingredient of this system is described in detail:
Virtual machine Server holds
In Server thread pool, start socket monitor, monitor the linking request of client; Starting the operating system, (if do not install rstat port monitoring instrument, first general Linux system default installation, needs the Tar compressed file downloading rstat from official website to rstat Daemon.After decompression tar file, just can compile, configure and install rstat watchdog routine).Monitored by Linux remote system, for network performance, the following network performance index of main monitoring, as shown in table 1:
Table 1
Test machine Client holds
Use the procotol such as TCP, UDP and virtual machine server holds to carry out Socket and communicate, perform test assignment, collection test result, and to main control computer transmission test result.Client supports the test assignment of three kinds of agreements of TCP, UDP and TCP/UDP mixing.Can unrestricted choice at Client.
Main control computer Manager
Send test command to client test machine, collect and statistical test result; Send RPC request to virtual machine, obtain virtual machine property indices, and graphically display.
Collect and statistical cloth test test result:
Main control computer collects test machine can gather test machine test result data every 3s, and different image data is averaged, and counts current Socket transfer rate (M/S).
Linux performance monitoring:
Linux performance monitoring mainly adopts remote procedure call (RPC) technology, remote procedure call (RemoteProcedure Call, RPC) be a kind of mode of interprocess communication, it allows computer program to perform program or the subroutine of another address space (usually another computing machine in shared network), and do not need to understand detailed process realize details.Most important difference is called in remote procedure call and this locality, and remote procedure call can because unpredictable network problem cause malloc failure malloc, and call can not this thing happens in this locality.In order to allow different clients can access services device, create some standardized RPC systems.These systems of great majority all employ interface description language (IDL), and the interface defined by IDL allows the client under different platforms realize far call.RPC communication spans transport layer and application layer [21], and supposes the existence of some host-host protocol, and as TCP (transmission control protocol) or UDP (User Datagram Protoco (UDP)), these agreements are carry data between signal procedure.RPC service is initiated by client, and client sends a request message performs (comprising call parameters) assignment procedure of known remote server.When server processing requests, client is in blocked state, and it can wait for until server complete process is called always.As shown in Figure 1, its built-in function is as follows:
(1) client call client handle.This calls is a local invocation of procedure, and parameter can be transferred to storehouse;
(2) client handle can be encapsulated into parameter in message, and sends message by invoke system call;
(3) native operating sys-tern of client is sent to message remote server from local machine;
(4) server OS receives the message packets arrived, and packet is passed to server handle;
(5) server handle is unpacked to message packets, obtains the parameter of transmission, and the invoking server invocation of procedure;
(6) the result return service device handle of the invocation of procedure;
(7) server handle returns results, and server OS forms message to result dress bag;
(8) server OS passes client back results messages;
(9) client handle is unpacked to results messages, and reduction result Data Concurrent delivers to client;
(10) client receives the data that client handle returns.
Corresponding above-mentioned intraoperation steps (1) ~ (10), label (1) ~ (10) in Fig. 1.
Obtain the real-time parameter of virtual machine performance monitor control index, and graphically show:
Performance data in test process can be kept in DataTable object.The object serialization due to DataTable object implementatio8, the data transmission real-time update obtained by the monitoring of Linux remote performance.After test terminates, after statistical summaries is carried out to the performance data in DataTable, obtain final performance test data.In order to represent test report better, improving the legibility of test report, have employed the method that XML and XSL combines.First utilize XSL to define the pattern of data exhibiting, then according to certain rule, test data is write in XML file; Finally, by the XslCompiledTransform class under System.Xml.Xsl NameSpace, the data file of XML format is changed into the user-friendly Html page.
Software installation and use key step #
Server holds:
1, start the operating system rstat Daemon; Rstat port monitoring instrument is installed, first, needs the Tar compressed file downloading rstat from official website.After decompression tar file, just can compile, configure and install rstat watchdog routine;
2, catalogue is changed to/etc/xinetd.d, and be called the file of rstatd with following content creating;
3, inetd (or xinetd) Daemon is restarted;
Using as issued orders :/etc/rc.d/init.d/xinetd restart, starting Socket multithreading and monitoring client-requested program, click Server end and monitor engineering VirtualNetWork_Server, start multithreading oracle listener;
Client holds: start client UI master routine.
Manager main control computer:
Install and start RPC client, configuring tested virtual machine ip parameter and properties Testing index, collecting performance monitoring index parameter.Collect step and be roughly divided into 4 steps:
1, the IP address of RPC client according to destination server and the default program port one 11 of portmapper, by Socket socket structure udp request, transmit a request to the program end slogan of the portmapper application asks rstatd of target machine; After portmapper application program receives request, the mapping table between inquiry port and application program, and the port numbers of rstatd watchdog routine is returned RPC client;
2, after RPC client obtains the port numbers of rstatd program, socket Socket re-constructs new Udp according to former target ip address and rstatd in port numbers and asks, and sends " request service " to rstatd watchdog routine;
3, after Rstatd watchdog routine receives " request service ", the performance data that current performance counter is collected is returned.
Test assignment is sent to client.
1, start main control computer program, main control computer is deployed in tomcat server;
Access main control computer Manager by http://localhost:8080/manager/host.action and distribute the main interface of test assignment;
2, upload test assignment, and click beginning testing button, test assignment is distributed to Client end;
3, graphically real-time exhibition virtual machine various performance parameters:
Network performance index mainly comprises Bytes Total/sec, Current Bandwidth, Output QueueLenth, Packets Received/s, PacketsReceived Error.
Above-described embodiment is the present invention's preferably embodiment; but embodiments of the present invention are not restricted to the described embodiments; change, the modification done under other any does not deviate from Spirit Essence of the present invention and principle, substitute, combine, simplify; all should be the substitute mode of equivalence, be included within protection scope of the present invention.
Claims (5)
1. a monitoring method for virtualized server performance, is characterized in that, comprises the step of following order:
S1. virtual machine Server end starts socket monitoring in Server thread pool, monitors the linking request of client;
S2. test machine Client holds to use TCP procotol, UDP procotol or TCP/UDP hybrid network agreement and virtual machine server holds to carry out Socket and communicates, and performs test assignment, collection test result, and to main control computer transmission test result;
S3. main control computer Manager sends test command to client test machine, collects and statistical cloth test test result;
S4. main control computer Manager sends RPC request to virtual machine, obtains virtual machine property indices, and graphically displays.
2. the monitoring method of virtualized server performance according to claim 1, it is characterized in that, in step S3, described collection and statistical cloth test test result be specially: main control computer Manager gathers the test result data of a client test machine every 3s, and different image data is averaged, count current Socket transfer rate.
3. the monitoring method of virtualized server performance according to claim 2, is characterized in that, in step S3, described collection distributed testing test result specifically comprises following steps:
The IP address of a, RPC client according to destination server and the default program port one 11 of portmapper, by Socket socket structure udp request, transmit a request to the program end slogan of the portmapper application asks rstatd of target machine;
After b, portmapper application program receives request, the mapping table between inquiry port and application program, and the port numbers of rstatd watchdog routine is returned RPC client;
After c, RPC client obtains the port numbers of rstatd program, socket Socket re-constructs new Udp according to former target ip address and rstatd in port numbers and asks, and sends " request service " to rstatd watchdog routine;
After d, Rstatd watchdog routine receives " request service ", return the performance data that current performance counter is collected.
4. the monitoring method of virtualized server performance according to claim 1, is characterized in that, in step S4, and described acquisition virtual machine property indices, and graphically display, be specially:
A, the performance data in test process to be kept in DataTable object, and by data transmission real-time update that the monitoring of Linux remote performance obtains;
B, after test terminates, statistical summaries is carried out to the performance data in DataTable, obtain final performance test data;
C, then utilize XSL to define the pattern of data exhibiting, then according to certain rule, test data is write in XML file;
D, last, by the XslCompiledTransform class under System.Xml.Xsl NameSpace, the data file of XML format is changed into the user-friendly Html page.
5. the monitoring system of a virtualized server performance, it is characterized in that: comprise interconnective virtual machine Server end, test machine Client end, main control computer Manager between two, wherein virtual machine Server end starts socket monitoring in Server thread pool, monitors the linking request of client; Test machine Client holds to use the procotol such as TCP, UDP and virtual machine server holds to carry out Socket and communicates, and performs test assignment, collection test result, and to main control computer transmission test result; Main control computer Manager sends test command to client test machine, collects and statistical test result; Send RPC request to virtual machine, obtain virtual machine property indices, and graphically display.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410598698.1A CN104407910A (en) | 2014-10-29 | 2014-10-29 | Virtualization server performance monitoring method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410598698.1A CN104407910A (en) | 2014-10-29 | 2014-10-29 | Virtualization server performance monitoring method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104407910A true CN104407910A (en) | 2015-03-11 |
Family
ID=52645543
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410598698.1A Pending CN104407910A (en) | 2014-10-29 | 2014-10-29 | Virtualization server performance monitoring method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104407910A (en) |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104852912A (en) * | 2015-04-29 | 2015-08-19 | 中国互联网络信息中心 | Domain name registration server performance testing system and domain name registration server performance testing method |
| CN104881347A (en) * | 2015-05-26 | 2015-09-02 | 山东蚁巡网络科技有限公司 | General virtualization monitoring display method |
| CN105095084A (en) * | 2015-08-25 | 2015-11-25 | 浪潮(北京)电子信息产业有限公司 | Memory performance test system and method of virtual machine of server virtualization platform |
| CN105610646A (en) * | 2015-12-30 | 2016-05-25 | 山石网科通信技术有限公司 | Testing method, device and system based on distributed virtual network equipment |
| CN105893010A (en) * | 2015-11-17 | 2016-08-24 | 乐视云计算有限公司 | Host scheduling method and system used for container deployment |
| CN106095688A (en) * | 2016-06-23 | 2016-11-09 | 微梦创科网络科技（中国）有限公司 | A kind of software performance testing method and device |
| CN106161130A (en) * | 2015-04-14 | 2016-11-23 | 阿里巴巴集团控股有限公司 | The performance monitoring device of sing on web Socket agreement, system and method |
| CN106201650A (en) * | 2016-06-29 | 2016-12-07 | 联想(北京)有限公司 | A kind of virtual machine control method and system |
| CN106815150A (en) * | 2017-01-19 | 2017-06-09 | 网易（杭州）网络有限公司 | Service end interface test system and method |
| CN107066372A (en) * | 2017-03-03 | 2017-08-18 | 九次方大数据信息集团有限公司 | Backstage resource monitoring method and device in system performance testing |
| CN107248922A (en) * | 2016-12-21 | 2017-10-13 | 常熟市顺网网络技术服务有限公司 | A kind of design method of remote desktop compression transmission |
| CN107370625A (en) * | 2017-07-24 | 2017-11-21 | 北京哈工大计算机网络与信息安全技术研究中心 | A kind of cloud platform running state monitoring method and monitor supervision platform |
| CN107396438A (en) * | 2017-08-25 | 2017-11-24 | 哈尔滨工业大学 | Communication port construction method, method for synchronizing time and structure health monitoring method based on smart mobile phone WLAN |
| CN107528831A (en) * | 2017-08-04 | 2017-12-29 | 无锡天脉聚源传媒科技有限公司 | A kind of data capture method and device |
| CN108509256A (en) * | 2017-02-28 | 2018-09-07 | 华为技术有限公司 | Method, equipment and the running equipment of management and running equipment |
| CN108540439A (en) * | 2018-02-01 | 2018-09-14 | 顺丰科技有限公司 | Data analysis method and system, equipment and storage medium |
| CN108733562A (en) * | 2018-04-26 | 2018-11-02 | 烽火通信科技股份有限公司 | A kind of test method and system of software platform |
| CN109857643A (en) * | 2018-12-29 | 2019-06-07 | 北京奇安信科技有限公司 | A kind of method and device of virtual platform performance automatic test |
| CN109922141A (en) * | 2019-02-21 | 2019-06-21 | 北京宝兰德软件股份有限公司 | The real time acquiring method and device of activity request list in Java application server |
| CN110232013A (en) * | 2019-02-25 | 2019-09-13 | 上海蔚来汽车有限公司 | Test method, device and controller and medium |
| CN110737564A (en) * | 2018-07-19 | 2020-01-31 | 朱国骏 | VmWare-based virtual machine performance monitoring method and system |
| CN111225078A (en) * | 2019-12-31 | 2020-06-02 | 北京安码科技有限公司 | Socket technology-based method for rapidly acquiring KVM virtual machine address |
| CN112331324A (en) * | 2020-10-09 | 2021-02-05 | 创业慧康科技股份有限公司 | Digital household delivery room information management system and method |
| CN114449025A (en) * | 2020-10-16 | 2022-05-06 | 上海汽车集团股份有限公司 | Communication method and system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101303656A (en) * | 2008-06-06 | 2008-11-12 | 中国科学院计算技术研究所 | Java application program remote direct memory access method |
| CN203039708U (en) * | 2013-01-18 | 2013-07-03 | 山东理工大学 | Network storage management software virtualization test system |
-
2014
- 2014-10-29 CN CN201410598698.1A patent/CN104407910A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101303656A (en) * | 2008-06-06 | 2008-11-12 | 中国科学院计算技术研究所 | Java application program remote direct memory access method |
| CN203039708U (en) * | 2013-01-18 | 2013-07-03 | 山东理工大学 | Network storage management software virtualization test system |
Non-Patent Citations (2)
| Title |
|---|
| KIM G H. ETC: "Towards virtualized and automated software performance test architecture", 《MULTIMEDIA TOOLS AND APPLICATIONS》 * |
| 曾波: "基于Web的性能测试平台关键技术研究与实现", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (37)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106161130A (en) * | 2015-04-14 | 2016-11-23 | 阿里巴巴集团控股有限公司 | The performance monitoring device of sing on web Socket agreement, system and method |
| CN104852912A (en) * | 2015-04-29 | 2015-08-19 | 中国互联网络信息中心 | Domain name registration server performance testing system and domain name registration server performance testing method |
| CN104852912B (en) * | 2015-04-29 | 2018-07-06 | 中国互联网络信息中心 | Domain name registration service device Performance Test System and method |
| CN104881347A (en) * | 2015-05-26 | 2015-09-02 | 山东蚁巡网络科技有限公司 | General virtualization monitoring display method |
| CN105095084B (en) * | 2015-08-25 | 2018-02-09 | 浪潮(北京)电子信息产业有限公司 | The virutal machine memory Performance Test System and method of a kind of server virtualization platform |
| CN105095084A (en) * | 2015-08-25 | 2015-11-25 | 浪潮(北京)电子信息产业有限公司 | Memory performance test system and method of virtual machine of server virtualization platform |
| CN105893010A (en) * | 2015-11-17 | 2016-08-24 | 乐视云计算有限公司 | Host scheduling method and system used for container deployment |
| CN105610646A (en) * | 2015-12-30 | 2016-05-25 | 山石网科通信技术有限公司 | Testing method, device and system based on distributed virtual network equipment |
| CN105610646B (en) * | 2015-12-30 | 2019-06-07 | 山石网科通信技术股份有限公司 | Test method, apparatus and system based on distributed virtual network device |
| CN106095688A (en) * | 2016-06-23 | 2016-11-09 | 微梦创科网络科技（中国）有限公司 | A kind of software performance testing method and device |
| CN106201650B (en) * | 2016-06-29 | 2020-08-25 | 联想(北京)有限公司 | Virtual machine control method and system |
| CN106201650A (en) * | 2016-06-29 | 2016-12-07 | 联想(北京)有限公司 | A kind of virtual machine control method and system |
| CN107248922A (en) * | 2016-12-21 | 2017-10-13 | 常熟市顺网网络技术服务有限公司 | A kind of design method of remote desktop compression transmission |
| CN106815150A (en) * | 2017-01-19 | 2017-06-09 | 网易（杭州）网络有限公司 | Service end interface test system and method |
| CN106815150B (en) * | 2017-01-19 | 2020-02-07 | 网易（杭州）网络有限公司 | Server-side interface test system and method |
| CN108509256B (en) * | 2017-02-28 | 2021-01-15 | 华为技术有限公司 | Method and device for scheduling running device and running device |
| CN108509256A (en) * | 2017-02-28 | 2018-09-07 | 华为技术有限公司 | Method, equipment and the running equipment of management and running equipment |
| WO2018157768A1 (en) * | 2017-02-28 | 2018-09-07 | 华为技术有限公司 | Method and device for scheduling running device, and running device |
| CN107066372A (en) * | 2017-03-03 | 2017-08-18 | 九次方大数据信息集团有限公司 | Backstage resource monitoring method and device in system performance testing |
| CN107370625A (en) * | 2017-07-24 | 2017-11-21 | 北京哈工大计算机网络与信息安全技术研究中心 | A kind of cloud platform running state monitoring method and monitor supervision platform |
| CN107528831A (en) * | 2017-08-04 | 2017-12-29 | 无锡天脉聚源传媒科技有限公司 | A kind of data capture method and device |
| CN107396438A (en) * | 2017-08-25 | 2017-11-24 | 哈尔滨工业大学 | Communication port construction method, method for synchronizing time and structure health monitoring method based on smart mobile phone WLAN |
| CN107396438B (en) * | 2017-08-25 | 2020-05-26 | 哈尔滨工业大学 | Channel construction, time synchronization and structure health monitoring method based on smart phone |
| CN108540439A (en) * | 2018-02-01 | 2018-09-14 | 顺丰科技有限公司 | Data analysis method and system, equipment and storage medium |
| CN108540439B (en) * | 2018-02-01 | 2021-10-29 | 顺丰科技有限公司 | Data analysis method, system, device and storage medium |
| CN108733562A (en) * | 2018-04-26 | 2018-11-02 | 烽火通信科技股份有限公司 | A kind of test method and system of software platform |
| CN108733562B (en) * | 2018-04-26 | 2021-07-20 | 烽火通信科技股份有限公司 | Software platform testing method and system |
| CN110737564A (en) * | 2018-07-19 | 2020-01-31 | 朱国骏 | VmWare-based virtual machine performance monitoring method and system |
| CN109857643A (en) * | 2018-12-29 | 2019-06-07 | 北京奇安信科技有限公司 | A kind of method and device of virtual platform performance automatic test |
| CN109922141A (en) * | 2019-02-21 | 2019-06-21 | 北京宝兰德软件股份有限公司 | The real time acquiring method and device of activity request list in Java application server |
| CN110232013B (en) * | 2019-02-25 | 2022-09-06 | 上海蔚来汽车有限公司 | Test method, test device, controller and medium |
| CN110232013A (en) * | 2019-02-25 | 2019-09-13 | 上海蔚来汽车有限公司 | Test method, device and controller and medium |
| CN111225078A (en) * | 2019-12-31 | 2020-06-02 | 北京安码科技有限公司 | Socket technology-based method for rapidly acquiring KVM virtual machine address |
| CN111225078B (en) * | 2019-12-31 | 2023-04-28 | 北京安码科技有限公司 | Socket technology-based method for rapidly acquiring KVM virtual machine address |
| CN112331324A (en) * | 2020-10-09 | 2021-02-05 | 创业慧康科技股份有限公司 | Digital household delivery room information management system and method |
| CN114449025A (en) * | 2020-10-16 | 2022-05-06 | 上海汽车集团股份有限公司 | Communication method and system |
| CN114449025B (en) * | 2020-10-16 | 2024-03-08 | 上海汽车集团股份有限公司 | Communication method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104407910A (en) | Virtualization server performance monitoring method and system | |
| US9756118B2 (en) | Virtual performance monitoring decoupled from hardware performance-monitoring units | |
| Ren et al. | A methodology towards virtualisation-based high performance simulation platform supporting multidisciplinary design of complex products | |
| CN102759979B (en) | A kind of energy consumption of virtual machine method of estimation and device | |
| Varghese et al. | A survey on edge performance benchmarking | |
| US10360050B2 (en) | Simulation of high performance computing (HPC) application environment using virtual nodes | |
| CN103780655A (en) | Message transmission interface task and resource scheduling system and method | |
| US20170093669A1 (en) | Load response performance counters | |
| CN106302628B (en) | Unified management scheduling method for computing resources in ARM architecture network cluster | |
| WO2018196296A1 (en) | Virtual machine scheduling device and method under non-uniform memory access architecture | |
| Kim et al. | Towards virtualized and automated software performance test architecture | |
| US10831539B2 (en) | Hardware thread switching for scheduling policy in a processor | |
| Kjorveziroski et al. | Kubernetes distributions for the edge: serverless performance evaluation | |
| Noronha et al. | Performance evaluation of container based virtualization on embedded microprocessors | |
| RahimiZadeh et al. | Performance modeling and analysis of virtualized multi-tier applications under dynamic workloads | |
| US20210099354A1 (en) | Method and system that determines application topology using network metrics | |
| Goga et al. | Simulation, modeling, and performance evaluation tools for cloud applications | |
| Bose et al. | Benchmarking database performance in a virtual environment | |
| Guo et al. | Decomposing and executing serverless applications as resource graphs | |
| Barone et al. | Gaas: Customized grids in the clouds | |
| CN110737564A (en) | VmWare-based virtual machine performance monitoring method and system | |
| CN113703913B (en) | Equipment testing method and device | |
| Fong et al. | Duality of virtualization: simplification and complexity | |
| US8656375B2 (en) | Cross-logical entity accelerators | |
| US9043814B1 (en) | Dispatcher framework to provide inter-application communication |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150311 |