CN101201751B - Method for planning capacity of multiple machines aiming at OLTP application - Google Patents
Method for planning capacity of multiple machines aiming at OLTP application Download PDFInfo
- Publication number
- CN101201751B CN101201751B CN200610165969XA CN200610165969A CN101201751B CN 101201751 B CN101201751 B CN 101201751B CN 200610165969X A CN200610165969X A CN 200610165969XA CN 200610165969 A CN200610165969 A CN 200610165969A CN 101201751 B CN101201751 B CN 101201751B
- Authority
- CN
- China
- Prior art keywords
- type
- performance
- capacity planning
- test
- oltp
- 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.)
- Active
Links
Images
Landscapes
- Debugging And Monitoring (AREA)
Abstract
The invention discloses a method which aims at implementing capacity planning over different machine types to OLTP applications and includes: executing flexibility evaluation of the OLTP application to determine whether the OLTP application has flexibility problem; if the method determines that the OLTP application does not have flexibility problem, a performance comparison test is executed among different machine types to acquire the running performances of the different machine types; computing performance ratio among the different machine types according to the acquired running performances of the different machine types; acquiring the performance values of various machine types in different series, which is necessarily based on the performance ratios among different machine types acquired from the performance comparison test, and the acquired performance values are input into a parameter database for capacity planning, consequently a new parameter database for capacity planning is formed, and performance comparison and performance prediction can be implemented over different machine types; furthermore, the new parameter database for capacity planning can be used by a modeling method for capacity planning and a capacity planning tool to implement capacity planning over different machine types which aims at the OLTP application.
Description
Technical field
The present invention relates generally to the host capacity planning of IT system planning and design, particularly relate at what concrete OLTP (Transaction Processing) used and stride the type method for planning capacity based on the performance comparison test.
Background technology
Capacity planning is meant in order to ensure system and can working efficiently in the future, the system load coupling for supporting the process of the only server hardware that this load is required, is used to instruct Capacity design and investment, to guarantee the rationality of IT cost.The service level that capacity planning needs the corresponding techniques means to come prognoses system can provide when increasing load or number of users, under the hardware configuration of supposition is provided, and this often sets up the volumetry model by the capacity planning instrument and realizes.
In the process of IT system planning and design, need carry out the capacity measuring and calculating to target computer system, and carry out lectotype selection on this basis.The benchmark test of industrywide standard at present is of a great variety, comprises TPC-C, TPC-H, and SPEC int, SPEC web, SPECjbb, SAPSD or the like, what different benchmark tests reflected is the performance of machine under the different application condition.Though various benchmark test indexs have certain reference value, have following problem:
1. the type of dissimilar (manufacturers) all has than big-difference in host board structure, CPU design, IO design, changing can appear in the performance performance of operation different application, and this just causes the performance gap of two machine often to have bigger variation in different benchmark test indexs;
2. the benchmark test index of various announcements all draws through after the meticulous sufficient performance optimization in former producer, and the system of existing deployment often can not reach this index;
3. in the process of actual capacity measuring and calculating, concrete OLTP uses inconsistent to the definition and the definition in the benchmark test of trade transactions or task, and proportionate relationship is unclear, just by virtue of experience estimate, error is bigger, as, the benchmark affairs in certain trade transactions in user's the OLTP system and the TPC-C test are inconsistent;
4. some manufacturer does not participate in whole benchmark test, has caused very big difficulty for the type selecting reference yet, as, SUN company does not participate in the TPC-C test in recent years;
5. the capacity planning parameter library that provides by manufacturer of the capacity planning instrument of main flow (as TeamQuest Model) can be finished the performance prediction of different class machines of the same type preferably, but be based on and the 1st order the same reason, the performance prediction of striding type accurately can't be provided, need the user to import the performance ratio of dissimilar machines.
Therefore, the host capacity planning field of IT system planning and design press for a kind of at concrete OLTP use based on dissimilar main frame performance comparison tests stride type capacity planning and performance prediction method, carry out lectotype selection more accurately.
Summary of the invention
The objective of the invention is to improve and use the order of accuarcy of striding the model-performance prediction, improve investment validity and utilization factor, make lectotype selection can select suitable machine, avoid performance deficiency or performance surplus at concrete OLTP.
This method is after confirming that the OLTP application software does not have the retractility problem, by performance test at the different type of machines of practical application, obtain the contrast of the runnability of different machines, then according to this ratio, expansion capacity planning parameter library, the modeling method and the capacity planning instrument of utilization capacity planning are finished and are striden type capacity planning and performance prediction accurately at this application.
According to the present invention, provide a kind of and striden the type method for planning capacity at what Transaction Processing (OLTP) was used, this method comprises: adopt the automated performance testing method to carry out OLTP and use the retractility evaluation and test, whether have the retractility problem to determine that OLTP uses; There is not the retractility problem if determine the OLTP application, then to different type of machines execution performance contrast test, to obtain the runnability of described different type of machines; According to the runnability of the described different type of machines that obtains, calculate the performance ratio of different type of machines; Need be based on the performance ratio of the different type of machines that draws in the performance comparison test and the performance number of each type, obtain each model-performance value of interdepartmental row, and each model-performance value of the described interdepartmental row that will obtain is input to the capacity planning parameter library, obtain new capacity planning parameter library, make that can stride type carries out performance comparison and performance prediction; And utilizing described new capacity planning parameter library, the modeling method and the capacity planning instrument of utilization capacity planning are finished at what this OLTP used and are striden the type capacity planning.
The method according to this invention has solved the problem that type carries out capacity planning and performance prediction of striding that existing capacity planning instrument can not finish, employing combines the strict performance contrast test of experimental situation with the capacity planning process mode, the reflection of science concrete OLTP be applied in performance performance difference on the different type of machines platform.
In addition, the method according to this invention can instruct strides the type capacity planning, improve investment validity and utilization factor, make lectotype selection can select suitable machine, avoid performance deficiency or performance surplus, and the method according to this invention serious interference of also having avoided the application software retractility and having brought for lectotype selection, capacity planning in the insufficient performance decline problem that causes of some platform property tuning.
Description of drawings
The following drawings constitutes the part of instructions and provides and further specifies of the present invention, and embodiments of the invention are described.
Fig. 1 illustrates the process flow diagram of using at concrete OLTP according to of the present invention of striding the type method for planning capacity based on the performance comparison test;
Fig. 2 illustrates the ultimate principle of capacity planning process.
Embodiment
Before specifically describing embodiments of the invention, at first clear and definite following term:
Contraction-expansion factor (Stretch Factor): contraction-expansion factor=(queue waiting time+service time)/service time=1/ (1-utilization factor), when contraction-expansion factor less than 2 the time, substantially with the slow increase of the increase approximately linear of load pressure, queue waiting time (Queue Time) is less than service time (Service Time) for cpu busy percentage (Utilization); When contraction-expansion factor=2, cpu busy percentage is 50%; When cpu busy percentage continues to increase, surpass at 70% o'clock, contraction-expansion factor and queue length (Queue Length) (according to queuing network theory, queue length=contraction-expansion factor-1) begin non-linear growth apace.
Perf
A type/B type: the A type is than the performance ratio of B type.
Throughput: the throughput rate that test draws.
CPUUtil: the host CPU utilization factor during test.
Fig. 1 illustrates the process flow diagram of using at concrete OLTP according to of the present invention of striding the type method for planning capacity based on the performance comparison test.
As shown in Figure 1, in step S1, carry out the evaluation and test of OLTP application software retractility, whether have the retractility problem to determine that OLTP uses.Have the retractility problem if determine the OLTP application, then process finishes.There is not the retractility problem if determine the OLTP application, enters step S2 in process.In step S2, formulate the performance comparison testing scheme, the performance comparison testing scheme is used to instruct the enforcement of performance comparison test.Then, process proceeds to step S3.In step S3, implement the performance contrast test, promptly implement test, and logging test results according to the performance comparison testing scheme, wherein Ce Shi index comprises: the cpu busy percentage of concurrent number, each type and IO handle up and response time etc.After finishing test, process proceeds to step S4.In step S4, calculate the performance ratio of different type of machines (i.e. the machine of this type), the calculating of the performance ratio of concrete different type of machines will be described in more detail below.After the calculating of the performance ratio of finishing different type of machines, determine the ratio scope that whether surpasss the expectation.The scope if whether ratio surpasss the expectation, then process finishes.The scope if ratio does not surpass the expectation, then process proceeds to step S5.In step S5, the performance ratio of different vendor's type of drawing in the performance comparison test is input to the capacity planning parameter library, make that striding type can carry out performance comparison and performance prediction.After performance ratio was input to the capacity planning parameter library, process proceeded to step S6.In step S6, carry out the capacity planning of striding type.
Below, be described in detail at each step in the accompanying drawing 1.
The evaluation and test of OLTP application software retractility
In the step S1 that accompanying drawing 1 is described, carry out the retractility evaluation and test that OLTP uses, to confirm that the load pressure that OLTP uses for variation has reliably flexible ability, do not exist because the major defect on internal data structure and the algorithm makes the performance performance sharply descend with the increase of load pressure.The specific implementation process of OLTP application retractility evaluation and test as described below.
Use the retractility evaluation and test according to OLTP according to the present invention and can adopt the automated performance testing method to implement, the choosing of compression tools can use industrial main flow as MercuryLoadRunner, also can adopt the pressure test program of writing voluntarily.
Use the retractility evaluation and test according to OLTP according to the present invention and choose main and business operation use-case that actual pressure is big continues the OLTP application software is exerted pressure in CLOSED enclosed type (promptly with fixing concurrent number) mode, and test can single use-case scene and the scene of comprehensively exerting pressure formation.
Then, to OLTP application implementation pressure test, and reasonably control concurrent number and the gap of test between iteration, make each type cpu busy percentage＜70% (handling up also lower) in principle for the typical OLTP IO of system, and in the queuing system each formation do not exist wait in line phenomenon (as, the request queue that guarantees the OLTP application software is not waited in line), guarantee that generally the whole queuing network of computer system is in the scope of linear extendible, contraction-expansion factor, (queue waiting time+service time)/service time is near 1;
After as above the OLTP application software having been carried out pressure test, use whether there is the retractility problem according to following standard determination OLTP:
A) with Little law check test result: according to the concurrent several N in the case type queueing network of Little law calculating CLOSED, N=R*T (R: response time, T: throughput rate (Throughput)), concurrent several N of proof theory are consistent with the concurrent number of exerting pressure;
B) under the different concurrent pressure of test case, cpu busy percentage is directly proportional with throughput rate;
C) under the blend pressure of many use-cases, the cpu busy percentage under cpu busy percentage and each the single use-case situation meets overlaying relation, as A+B ≈ C, and D+E ≈ F; And throughput rate also is an overlaying relation;
D) if satisfy a), b), c) then there is not the retractility problem, otherwise, there is the retractility problem
Have the retractility problem if this OLTP uses, then be not suitable for carrying out capacity planning and performance prediction, need finish and carry out again after software improves, perhaps the test index of basis on physical device done the foundation as lectotype selection; If the inductility problem can continue.
Formulate the performance comparison testing scheme
Finish in the evaluation and test of OLTP application software retractility, and definite OLTP uses after the inductility problem, as the described formulation performance comparison of the step S2 of accompanying drawing 1 testing scheme.
The determining of performance comparison testing scheme comprises content for example as described below in order to be used to instruct the enforcement of performance comparison test.
1, determines test environment, comprise: the type environment of determining to participate in the performance comparison test, for example: the type of supposing the test of participation performance comparison is the A type and the B type of different manufacturers, but those skilled in the art are to be understood that above-mentioned two kinds of types only are exemplary, the type of participating in the performance comparison test is not limited only to two kinds of types, can be the type of any amount as required; And definite OLTP is applied in the deployment scheme and the deployment parameters (for example comprising that version, OLTP application software memory parameters or the OLTP application software queue parameter etc. of OLTP application software influence the important parameter of system performance) of each type, wherein various the big parameter of performance performance influence is consistent as far as possible or carries out certain optimization at platform.For example: if the OLTP application system has comprised oracle database, the major parameter of Oralce database should be consistent on A type and B type.
2, after having determined test environment, specified data is prepared scheme, for example determine to use data when, how many data volumes have been comprised, the data that relate to during test have what etc., and guaranteed performance contrast carries out under identical data, with data consistent and the statistics of database information unanimity that guarantees each environment.
3, obtain the user's of tested OLTP application software use information, comprise operation commonly used, operation frequency, concurrent user number or the like.
4, determine the load simulation scheme.In the present invention, can adopt the automated performance testing method to implement, the choosing of compression tools can use industrial main flow as MercuryLoadRunner, also can adopt the pressure test program of writing voluntarily.According to the customer service investigation result, frequent and the business operation use-case that actual pressure is big of selection operation continues the OLTP application software is exerted pressure in CLOSED enclosed type (promptly with fixing concurrent number) mode, and test constitutes with a plurality of scenes, and each scene comprises single use-case.Reasonably control the gap between concurrent number and test iteration, make each type cpu busy percentage＜70% (handling up also lower) in principle for the typical OLTP IO of system, and in the queuing system each formation do not exist wait in line phenomenon (as, the request queue that guarantees Tuxedo Server is not waited in line), guarantee that generally the whole queuing network of computer system is in the scope of linear extendible, the contraction-expansion factor contraction-expansion factor, (queue waiting time+service time)/service time is near 1.
After having determined the load simulation scheme, determine performance ratio calculation scheme and evaluation criterion.Can think that owing to passed through the retractility evaluation and test on each type platform, when the computer system queuing network was in the linear extendible scope, each test case can linear extendible; Under the logic that this OLTP uses was stable situation, " the unit affairs consume CPU " also was stable; And can adopt PerfA
Type/ B
Type=(CPUUtil B
Type* ThroughputA
Type)/(CPUUtilA
Type* Throughput B
Type) calculated performance ratio, wherein CPUUtil B
TypeThe cpu busy percentage of expression B type, CPUUtil A
TypeThe cpu busy percentage of expression A type.
In addition, can be according to characteristic (C/C++, the JAVA of this OLTP application, WEB, SAP etc.) determine suitable industrial benchmark test index correlative value, as a reference, and determine certain variation range of bearing (single use-case and comprehensive condition are formulated respectively) according to actual needs.If actual calculated performance ratio exceeds this scope, illustrate that then this OLTP is applied in the performance optimization of not carrying out peer-level on relative another type platform on certain type platform as yet, suggestion is carried out performance optimization or is abandoned adopting this platform at this problem type platform.
Implement the performance contrast test
After the formulation of performance comparison testing scheme is finished, as the described enforcement performance of the step S3 of accompanying drawing 1 contrast test.
This performance comparison test is implemented according to the performance comparison testing scheme, and logging test results, and wherein Ce Shi important index for example comprises: the cpu busy percentage of concurrent number, each type and IO handle up and the response time.
Calculate the performance ratio of different type of machines
After the performance comparison test is finished, as the performance ratio of the described calculating different type of machines of the step S4 of accompanying drawing 1.
For the calculating of the performance ratio of different type of machines, at each single use-case, can be according to Perf
A type/B type=(CPUUtil
The B type* Throughput
The A type)/(CPUUtil
The A type* Throughput
The B type) calculate performance ratio at this use-case.
In addition, can draw the ratio of each use-case in actual loading, carry out the weighted mean of each single use-case PerfA type/B type ratio, draw the combination property ratio of system according to this ratio according to customer service investigation.
In addition, if each single use-case Perf
A type/B typeRatio exceeds corresponding scope or overall system performance ratio predetermined in the scheme and exceeds predetermined corresponding scope, think that then this OLTP uses for this use-case/integral body in the performance optimization of not carrying out peer-level on relative another type platform on certain type platform as yet, can carry out performance optimization or abandon adopting this platform at this type platform.
Input capacity projecting parameter storehouse
After the performance ratio calculation of different type of machines is finished, described with the performance ratio input capacity projecting parameter storehouse of calculating as the step S5 of accompanying drawing 1.
In the capacity planning instrument (as TeamQuest Model) of main flow, the capacity planning parameter library is arranged, for each each type of series has defined performance index, but these performance index only just possess comparability between with each a series of types (as is all HP, the performance index that perhaps are all between each class machine of SUN possess comparability), performance index between the type of different series do not possess any comparability, therefore can't finish the manufacturer's type of striding of capacity model and predict.
Therefore, need be based on for example performance ratio of two producer's type that draws in the performance comparison test, obtain each model-performance value of interdepartmental row (producer), and the performance number that obtains is input to the capacity planning parameter library, obtain new capacity planning parameter library, make that striding type can carry out performance comparison and performance prediction.In the capacity planning storehouse, for example each machine type data is listed by following form:
| Type series | The type model | Performance number |
| The I of producer | Type A | 2.200000e+002 |
| The II of producer | Type B | 8.040000e+002 |
Because two types adhere to different series separately, so performance number do not have a comparability, but through contrast test, Perf as can be known
A type/B typePerformance ratio, if: Perf
A type/B type=2, then can in type series " I of producer ", add type model " type B ", and in type series " II of producer ", add type model " type A " by calculating, formula is as follows:
Performance number=type the A/Perf of type B in the I of producer
A type/B type=2.200000e+002/2=1.100000e+002
Performance number=type the B*Perf of type A in the II of producer
A type/B type=2.200000e+002*2=16.080000e+002
Capacity planning parameter library such as following table after the increase new argument:
| Type series | The type model | Performance number |
| The I of producer | Type A | 2.200000e+002 |
| The I of producer | Type B (II of producer) | 1.100000e+002 |
| The II of producer | Type B | 8.040000e+002 |
| The II of producer | Type A (I of producer) | 16.080000e+002 |
After increasing parameter, in the tabulation of the I of producer of capacity planning instrument (TeamQuest), can see the type B of the II of producer, performance number is: 1.100000e+002; In like manner, in the tabulation of the II of producer, can see the type B of the I of producer, performance number is: 16.080000e+002.These two numerical value can be used for capacity planning.
The capacity planning of type is striden in execution
Behind each model-performance parameter input capacity planning library of interdepartmental row, can utilize in the capacity planning process system queuing's network model (as the model of in TeamQuest Model, setting up) of setting up based on production environment actual loading or simulated environment pressure test, suppose under the business load of each series, each type configuration and variation the performance performance of predicted application system.Fig. 2 shows the ultimate principle of capacity planning process.Method for planning capacity shown in Figure 2 is identical with conventional method, does not repeat them here.
For the present invention, before not importing each model-performance parameter of interdepartmental row, only can be used for predicting the performance performance (as the type C of producer I) of this model on other types of the I of producer at system queuing's network model of setting up on the type A, but after importing each model-performance parameter of interdepartmental row, the performance performance (as the type B of producer II) of available this model prediction on the type of the II of producer realized striding the type capacity planning thus.
Though detailed description of the present invention is at exemplary case, to those skilled in the art, various modification and the replacement form of these embodiment all can be imagined.Therefore, all modification and replacement forms in the clear and definite scope of patent protection of the present invention of claims have been contained in the present invention.
Claims (5)
1. stride the type method for planning capacity at what Transaction Processing was used for one kind, this method comprises:
Adopt the automated performance testing method to carry out Transaction Processing and use the retractility evaluation and test, whether have the retractility problem to determine that Transaction Processing is used;
There is not the retractility problem if determine the Transaction Processing application, then to different type of machines execution performance contrast test, to obtain the runnability of described different type of machines;
According to the runnability of the described different type of machines that obtains, calculate the performance ratio of different type of machines;
Need be based on the performance ratio of the different type of machines that draws in the performance comparison test and the performance number of each type, obtain each model-performance value of interdepartmental row, and each model-performance value of the described interdepartmental row that will obtain is input to the capacity planning parameter library, obtain new capacity planning parameter library, make that can stride type carries out performance comparison and performance prediction; And
Utilize described new capacity planning parameter library, the modeling method and the capacity planning instrument of utilization capacity planning are finished at what this Transaction Processing was used and are striden the type capacity planning.
2. adopt Little law and affairs throughput rate and cpu busy percentage whether to satisfy the pressure test result of linear superposition relation checking sealing mode, to be used to judge whether the Transaction Processing application exists the retractility problem according to the process of claim 1 wherein.
3. use at the Transaction Processing that does not have flexible problem according to the process of claim 1 wherein, adopt " the unit affairs consume CPU " method to calculate the performance ratio of different type of machines, i.e. Perf
A type/B type=(CPUUtil
The B type* Throughput
The A type)/(CPUUtil
The A type* Throughput
The B type), the host CPU utilization factor when CPUUtil represents to test, Throughput represents to test the throughput rate that draws.
4. according to the method for claim 1, also comprise and carry out the whether available evaluation of performance ratio, comprise: determine that according to the characteristic that this Transaction Processing is used suitable industrial benchmark test index correlative value is as the reference benchmark, the definite according to actual needs variation range that can bear, if the performance ratio that calculates exceeds this scope, then abandon this performance ratio.
5. according to the process of claim 1 wherein that the index of described contrast test comprises concurrent number, type cpu busy percentage at least and handles up and the response time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200610165969XA CN101201751B (en) | 2006-12-12 | 2006-12-12 | Method for planning capacity of multiple machines aiming at OLTP application |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200610165969XA CN101201751B (en) | 2006-12-12 | 2006-12-12 | Method for planning capacity of multiple machines aiming at OLTP application |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101201751A CN101201751A (en) | 2008-06-18 |
| CN101201751B true CN101201751B (en) | 2011-04-27 |
Family
ID=39516925
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN200610165969XA Active CN101201751B (en) | 2006-12-12 | 2006-12-12 | Method for planning capacity of multiple machines aiming at OLTP application |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101201751B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104866408B (en) * | 2014-02-20 | 2017-12-22 | 阿里巴巴集团控股有限公司 | The capacity prediction methods and device of a kind of application system |
| CN109656796B (en) * | 2018-12-14 | 2022-05-10 | 武汉斗鱼鱼乐网络科技有限公司 | Task selection method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1547408A (en) * | 2003-12-09 | 2004-11-17 | 中兴通讯股份有限公司 | A method of capacity planning evaluation |
| CN1790397A (en) * | 2005-12-28 | 2006-06-21 | 浙江工业大学 | Third party logistics data processing method based on online analysis |
-
2006
- 2006-12-12 CN CN200610165969XA patent/CN101201751B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1547408A (en) * | 2003-12-09 | 2004-11-17 | 中兴通讯股份有限公司 | A method of capacity planning evaluation |
| CN1790397A (en) * | 2005-12-28 | 2006-06-21 | 浙江工业大学 | Third party logistics data processing method based on online analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101201751A (en) | 2008-06-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kumar Chhabra et al. | A survey of dynamic software metrics | |
| US8788986B2 (en) | System and method for capacity planning for systems with multithreaded multicore multiprocessor resources | |
| Böhme et al. | Scalable critical-path based performance analysis | |
| US20040148152A1 (en) | System performance prediction mechanism and method based on software component performance measurements | |
| US20090055823A1 (en) | System and method for capacity planning for systems with multithreaded multicore multiprocessor resources | |
| US20050027661A1 (en) | Method and computer system for providing a cost estimate for sizing a computer system | |
| CN112433853A (en) | Heterogeneous sensing data partitioning method for parallel application of supercomputer data | |
| US8443073B2 (en) | Automated performance prediction for service-oriented architectures | |
| US20050192781A1 (en) | System and method for modeling LPAR behaviors in a simulation tool | |
| Cope et al. | Robust data placement in urgent computing environments | |
| CN101201751B (en) | Method for planning capacity of multiple machines aiming at OLTP application | |
| Diao et al. | Comparative studies of load balancing with control and optimization techniques | |
| CN112784435B (en) | GPU real-time power modeling method based on performance event counting and temperature | |
| Liu et al. | Predicting the performance of middleware-based applications at the design level | |
| US20130346436A1 (en) | Simulation Techniques for Predicting In-Memory Database Systems Performance | |
| Keong et al. | Toward using software metrics as indicator to measure power consumption of mobile application: A case study | |
| JP2009037369A (en) | Resource assignment method to database server | |
| Sen et al. | Predictive price-performance optimization for serverless query processing | |
| Elsayed et al. | Optimal design of proportional hazards based accelerated life testing plans | |
| EP3826233B1 (en) | Enhanced selection of cloud architecture profiles | |
| Smith et al. | Introduction to Software Performance Engineering. | |
| CN117234711B (en) | Dynamic allocation method, system, equipment and medium for Flink system resources | |
| CN103745092B (en) | Evaluation method and evaluation system for utilization ratio of server | |
| Caniou et al. | Multicriteria scheduling heuristics for Gridrpc systems | |
| Merkuryeva et al. | Simulation-based approach for comparison of (s, Q) and (R, S) replenishment policies utilization efficiency in multi-echelon supply chains |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |