Розроблення алгоритму прогнозування продуктивності хмарних сервісів
Main stages of data center service performance prediction were discussed, specifically data monitoring and gathering, calculation and prediction of key indexes and performance index prediction. It was proposed to build data center service performance prediction algorithm based on an analysis of the...
Gespeichert in:
| Datum: | 2018 |
|---|---|
| Hauptverfasser: | , , , |
| Format: | Artikel |
| Sprache: | Englisch |
| Veröffentlicht: |
The National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute"
2018
|
| Schlagworte: | |
| Online Zugang: | https://journal.iasa.kpi.ua/article/view/139702 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Назва журналу: | System research and information technologies |
| Завантажити файл: |  |
Institution
System research and information technologies| _version_ | 1867334339877928960 |
|---|---|
| author | Zuev, Denis O. Kropachev, Artemii V. Usov, Aleksey Ye. Gorshunov, Roman A. |
| author_facet | Zuev, Denis O. Kropachev, Artemii V. Usov, Aleksey Ye. Gorshunov, Roman A. |
| author_institution_txt_mv | [
{
"author": "Denis O. Zuev",
"institution": null
},
{
"author": "Artemii V. Kropachev",
"institution": "Bell Integrator"
},
{
"author": "Aleksey Ye. Usov",
"institution": "Rosgosstrakh"
},
{
"author": "Roman A. Gorshunov",
"institution": "AT&T, Bratislava"
}
] |
| author_sort | Zuev, Denis O. |
| baseUrl_str | http://journal.iasa.kpi.ua/oai |
| collection | OJS |
| datestamp_date | 2019-01-17T13:29:35Z |
| description | Main stages of data center service performance prediction were discussed, specifically data monitoring and gathering, calculation and prediction of key indexes and performance index prediction. It was proposed to build data center service performance prediction algorithm based on an analysis of the service transactions index, service resource occupancy index and service performance index. Prediction of the indexes is based on chaotic time series analysis that was used to estimate service transactions index time series trend, the radar chart method to calculate the service resource occupancy index value and weighted average method to calculate service performance index. For performance prediction, it is proposed to use a fuzzy judgment matrix with the service transactions index and service resource occupancy index as input values. It was taken into consideration that service transactions index is usually represented by nonlinear time series and thus the index time series parameters had to be predicted by chaos theory and for the calculation of this index, the estimation procedure of Lyapunov exponent value can be used. The radar chart demonstrates service resource occupancy index estimation of shared storage, mobile storage, memory, computational capability and network bandwidth. The prediction technique was based on the fuzzy nearness category that use input values of transactions index and dynamic changes of the service resource occupancy index. |
| doi_str_mv | 10.20535/SRIT.2308-8893.2018.2.01 |
| first_indexed | 2025-07-17T10:24:01Z |
| format | Article |
| fulltext |
© D.O. Zuev, A.V. Kropachev, A.Ye. Usov, R.A. Gorshunov, 2018
Системні дослідження та інформаційні технології, 2018, № 2 7
TIДC
ПРОГРЕСИВНІ ІНФОРМАЦІЙНІ ТЕХНОЛОГІЇ,
ВИСОКОПРОДУКТИВНІ КОМП’ЮТЕРНІ
СИСТЕМИ
УДК 681.3.093:044.3
DOI: 10.20535/SRIT.2308-8893.2018.2.01
DEVELOPMENT OF THE PERFORMANCE PREDICTION
ALGORITHMS FOR CLOUD SERVICES
D.O. ZUEV, A.V. KROPACHEV, A.YE. USOV, R.A. GORSHUNOV
Abstract. Main stages of data center service performance prediction were discussed,
specifically data monitoring and gathering, calculation and prediction of key indexes
and performance index prediction. It was proposed to build data center service per-
formance prediction algorithm based on an analysis of the service transactions in-
dex, service resource occupancy index and service performance index. Prediction of
the indexes is based on chaotic time series analysis that was used to estimate service
transactions index time series trend, the radar chart method to calculate the service
resource occupancy index value and weighted average method to calculate service
performance index. For performance prediction, it is proposed to use a fuzzy judg-
ment matrix with the service transactions index and service resource occupancy in-
dex as input values. It was taken into consideration that service transactions index is
usually represented by nonlinear time series and thus the index time series parame-
ters had to be predicted by chaos theory and for the calculation of this index, the es-
timation procedure of Lyapunov exponent value can be used. The radar chart dem-
onstrates service resource occupancy index estimation of shared storage, mobile
storage, memory, computational capability and network bandwidth. The prediction
technique was based on the fuzzy nearness category that use input values of transac-
tions index and dynamic changes of the service resource occupancy index.
Keywords: data center, service transactions index, service resource occupancy in-
dex, service performance index, fuzzy judgment matrix, Lyapunov exponent, radar
chart.
INTRODUCTION
Nowadays requirements to cloud platform data center services performance have
significantly grown. Thereby it’s important to develop effective and multipurpose
algorithm of estimation of key aspects that refers to the stability of the network
infrastructure work. An efficient strategy should be based on an analysis of the
whole dataset of gathered information of a monitoring platform and to be able to
predict indexes of the data center performance at any moment of time with the
high accuracy.
The assigned task could be solved by mathematical methods of the chaotic
analysis and fuzzy logic, but adaption of them stands a nontrivial task. In order to
identify the main aspects of the problem, an analysis of recent studies and publi-
D.O. Zuev, A.V. Kropachev, A.Ye. Usov, R.A. Gorshunov
ISSN 1681–6048 System Research & Information Technologies, 2018, № 2 8
cations was done. It was analyzed that aspects of data center service performance
that were mentioned to be key ones [1, 2], specifically service transactions index,
service resource occupancy index and service performance index. To solve the
problem of prediction of those, works were studied devoted to chaotic analysis [2-
4], radar chart method [1, 5] and weighted average method [6]. Also, within the
bounds of this study, the fundamental mathematical materials were analyzed
[7–9] related to fuzzy logic in order to use it at cloud platform data center services
performance analysis and prediction. The system analysis shows the possibility to
develop an effective technique based on monitoring and gathering of information
for estimation and accurate prediction of key aspects that refers to the data center
service performance.
SERVICE INDEXES PREDICTION AND CALCULATION PROCEDURE
The data center service performance prediction procedure [1, 2] usually includes
following stages (Fig. 1):
• monitoring and gathering of data center indicators’ data;
• calculation and prediction of key indexes of data center infrastructure
work;
• prediction of the data center service performance index.
Indicators’ data contain recorded by virtual machine (VM) monitoring
plugins information about transaction logs, utilization level of physical resources
(shared storage, computational capability, network bandwidth, etc.) and response
time of each monitoring spot which refers to the system performance. The analy-
sis of gathered data allows defining key indexes of data center infrastructure work
efficiency (Fig. 1):
• service transactions index (STI);
• service resource occupancy index (SROI);
• service performance index (SPI).
STI value refers to the number of data center’s transactions that require
a service to process. This index indicates service’s loads at each moment and
should be recorded as a time series ],,,[: 21 ni xxxx K which corresponds to the
time chart ],,,[: 21 ni tttt K . As it is shown at Figure 1, usually STI time series
have to be modeled as a nonlinear sequence. Thus, STI trends can be predicted by
nonlinear time series forecasting methods based on the artificial neural networks
(ANNs) platform. In other hand, SROI value refers to data center servers’ physi-
cal resources allocated to the service at each moment and SPI value refers to the
data center service’s response time at each moment. It should be mentioned that
SPI directly reflects service performance while this index is the comprehensive
result of the key monitoring points’ analysis.
Prediction of key indexes procedure includes a variety of methods or algo-
rithms that can be used. Within the bounds of this study it is proposed to use (Fig. 1):
• chaotic time series analysis to estimate STI time series trend [2–4];
• radar chart method to calculate SROI value [1, 5];
• weighted average method to calculate SPI value [6].
Development of the performance prediction algorithms for cloud services
Системні дослідження та інформаційні технології, 2018, № 2 9
Performance prediction of the modern data center service work process
should be based on a fuzzy judgment matrix. It uses STI and SROI values (Figure
1) according to the definition of the fuzzy closeness degree, and estimate the best
matching value of STI and SROI at the predicted moment by a similarity match-
ing algorithm. Thereby, SPI which corresponds to the obtained value represents
prediction result data center service performance to be compared with the value
that was obtained experimentally.
STI TIME SERIES PREDICTION ALGORITHM
It was mentioned above that the modern data center service based on cloud para-
digm is usually has to be represented by a nonlinear system. It could be added that
Fig. 1. Data center service performance prediction scheme
D.O. Zuev, A.V. Kropachev, A.Ye. Usov, R.A. Gorshunov
ISSN 1681–6048 System Research & Information Technologies, 2018, № 2 10
STI time series would be nonlinear time series on a cloud platform. Thereby, STI
time series parameters have to be predicted by the chaos theory.
For reconstruction of STI time series the delay embedding theorem should
be used (Takens’ theorem). Let us suppose that time series ],,,[: 21 ni xxxx K
which corresponds to the time ],,,[: 21 ni tttt K have power system dimension d
and thus the system must be considered form d -dimensional state vector )(txi
that evolves according to an unknown but continuous and deterministic dynamics.
For simplified form of Takens’ theorem [1, 7–9] adapted to the time series predic-
tion it can be said that observable result xF is a smooth function of ix dataset.
)(tFx has to be supplemented by observations made within certain time lag τ
multiplied by values mk ,,1K= :
)](,),(,),2(),(),([:),( τ−τ−τ−τ− mtFktFtFtFtFktF xxxxxx KK .
It’s obvious that for increasing number of lags m it will lead motion in the
lagged space to become more predictable, and for ∞→m system will tend to
become deterministic and equivalent to original state space. Takens’ theorem [1]
demonstrates that lagged vectors become deterministic at a finite dimension of
12 +≥ dm . Thereby STI time series prediction’s target function )( ix tF of
m -dimensional phase space with N phase points could be defined in every point
in space phase as:
⎩
⎨
⎧
−τ−==+≥
−τ+τ+τ+=
.)1(;,,2,1;12
))]1((,),2(),(),([)(
mnNNidm
mtxtxtxtxtF iiiiix
K
K
It has to be noticed that 12 +≥ dm is not a necessary but sufficient condi-
tion of determination of system dynamics.
STI time series’ calculation could be done not only by a qualitative analysis
but also by a quantitative algorithm. It’s based on calculating some chaotic quan-
tities. The most effective way is to estimate Lyapunov exponent value. Lyapunov
exponent of a dynamical system is a quantity that characterizes the rate of separa-
tion of infinitesimally close trajectories []. Two trajectories in phase space with
initial separation 0Zδ diverge as:
⎟
⎟
⎟
⎟
⎟
⎠
⎞
⎜
⎜
⎜
⎜
⎜
⎝
⎛
⎟
⎟
⎟
⎟
⎟
⎠
⎞
⎜
⎜
⎜
⎜
⎜
⎝
⎛
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛
δ
δ
≥δ≈δ
→δ∞→τ t
Z
tZ
ZtZ
Z
||
|)(|ln
limlim)( 0
00
0
,
where λ is the Lyapunov exponent and 00 →δZ criteria ensure the validity of
the linear approximation at each moment of time. Thereby, the biggest obtained
value of Lyapunov exponent (MLE: maximal Lyapunov exponent) is a parameter
which could be used for estimation whether a system is a chaotic one ( 0>λ ) or
not ( 0≤λ ). It should be noticed that the initial separation vector usually contain
some component in the direction associated with the MLE, and thus an effect of
the other exponents can be neglected.
Development of the performance prediction algorithms for cloud services
Системні дослідження та інформаційні технології, 2018, № 2 11
For analysis of STI time series the proposed mathematical model could be
slightly simplified. Let us suppose that we need to predict knx + for dataset of
],,,[: 21 ni xxxx K . We have to choose a point iX for the prediction center in a
phase space of the system. iX is defined as:
])1(,,)1(,)1([: 1 −τ−−τ−−τ− ++ mxmxmxX knnni K .
The next step is to define nearest point },,,{: 121 −ij XXXX K . While dis-
tance between iX and jX is d , then d could be defined as ji XXd −= .
Therefore, MLE could be estimated by comparison of 1+−= ii XXd and
1+− jj XX differences.
⎟
⎟
⎠
⎞
⎜
⎜
⎝
⎛
−
−
=λ≥−=−
+
+
+
λ
+
1
1
111 ln1
jj
ii
jjii XX
XX
XXeXX .
While 1λ is obtained MLE for time series ],,,[: 21 ni xxxx K , it predicts
1+nx . To predict knx + , k -step prediction should be done.
SROI AND SPI VALUE PREDICTION ALGORITHM
As it was mentioned above, SROI value refers to the data center service physical
resources utilization level. Physical resources are distributed on different servers
and VMs so estimation of SROI value is a nontrivial task. The most efficient
method of SROI analysis is development of radar char, a graphical method of dis-
playing multivariate data more than two quantitative variables [1, 5].
The radar chart area R(t) for SROI evaluation and prediction can be gotten as
follows:
∑
⎟
⎠
⎞
⎜
⎝
⎛ π
=
ji
ji yytR
,
][
2
3
2sin
)( .
Fig. 2 shows the demonstrated radar chart that can be used for SROI analysis
for five resources:
• shared storage;
• mobile storage;
• memory (RAM and cash-memory);
• computational capability (CPU);
• network bandwidth.
There are several methods of effective SPI prediction but all of them based
on estimation of monitoring points response time dataset ][: nii TTT K . Thereby,
basic equation for SPI at any moment of time could be defined as:
∑
=
=
n
i
ii T
n
P
1
1 .
D.O. Zuev, A.V. Kropachev, A.Ye. Usov, R.A. Gorshunov
ISSN 1681–6048 System Research & Information Technologies, 2018, № 2 12
Prediction technique is based on the fuzzy nearness category that use input
values of STI and SROI values dynamic changes (as a real time process). A fuzzy
matching algorithm estimates the nearness degree of STI and SROI of the predic-
tion time. The nearness level of STI values’ dataset ( nX ) and SROI values’ data-
set ( nY ) for the n time moments to be predicted should be estimated to iX and
iY that the closest ones to nX and nY , respectively. To calculate SPI values’
dataset as a set of predicted performance values at predicted time moments,
iX and iY values have to be used (Fig. 3).
Estimation of SPI values’ dataset is impossible without getting the nearness
degree of STI and SROI which is based on calculating of the membership degree
(Figure 3). The membership degree is a value of membership function
%]100;0[∈F that refers to the correlations between an element and some charac-
teristic [1, 9]. The calculation of a membership function is based on eigenvalue
matrix of iX and iY datasets:
⎪
⎪
⎩
⎪⎪
⎨
⎧
−
−
=
−
−
=
.
)(min)(max
)(min)(
;
)(min)(max
)(min)(
nn
ni
i
nn
ni
i
YY
YYYF
XX
XXXF
It allows obtaining fuzzy matrix of )( iXF and c )( iYF datasets (Fig. 3). To-
gether with )( nXF and )( nYF datasets it should be used to obtain the nearness
degree:
∨−∧−∧∨= )))(1())(1((()))()(((),( NiNini XFXFXFXFAAND
Рис. 2. Data center service resources occupancy radar chart
Development of the performance prediction algorithms for cloud services
Системні дослідження та інформаційні технології, 2018, № 2 13
))))(1())(1(( Ni YFYF −∧−∨ ,
where iA represents the matrix in moment i (an estimated moment of time), and
nA represents the matrix in moment n (a predicted moment of time).
CONCLUSIONS
Main stages of data center service performance prediction, such as indicators’
data monitoring and gathering, calculation and prediction of key indexes of data
center infrastructure work and performance index prediction were discussed. It
was proposed to build the data center service performance prediction algorithm
based on an analysis of the service transactions index, service resource occupancy
index and service performance index. The prediction of the indexes was based on
the chaotic time series analysis that was used to estimate the service transactions
index time series trend, radar chart method to calculate the service resource occu-
pancy index value and weighted average method to calculate the service perform-
ance index.
For performance prediction, it was proposed to use the fuzzy judgment ma-
trix with the service transactions index and service resource occupancy index as
input values. Next stages include the definition of a fuzzy closeness degree and an
estimation of the best matching value of the indexes at the predicted moment by
the similarity matching algorithm. It was taken into consideration that service
Fig. 3. Evaluation of nearness degree of STI and SROI.
D.O. Zuev, A.V. Kropachev, A.Ye. Usov, R.A. Gorshunov
ISSN 1681–6048 System Research & Information Technologies, 2018, № 2 14
transactions index is usually represented by nonlinear time series. It was noticed
that the index time series parameters have to be predicted by the chaos theory and
thereby for the calculation of this index the estimation procedure of Lyapunov
exponent value was used. Radar chart that was used for service resource occu-
pancy index estimation was built for five main resources of the cloud platform
service: shared storage, mobile storage, memory, computational capability and
network bandwidth. For calculation of service performance index values’ dataset
it is necessary to find the nearness degree of the service transactions index and
service resource occupancy index; it was proposed to estimate the first member-
ship degree. Therefore, the prediction technique was based on the fuzzy nearness
category that used input values of service transactions index and service resource
occupancy index dynamic changes which was considered as a real time process.
REFERENCES
1. Wu C. Software Monitoring in Data Centers. Handbook on Data Centers / C. Wu,
J. Guo. — 2015. — P. 1209–1253.
2. Newcombe L. Data Center Financial Analysis, ROI and TCO. Data Center Hand-
book / L. Newcombe. — 2014. — P. 103–137.
3. Román-Flores H. Chaos on Set-Valued Dynamics and Control Sets. Chaos Theory /
H. Román-Flores, V. Ayala. — 2018.
4. Tang R. Metaheuristics and Chaos Theory. Chaos Theory / R. Tang, S. Fong,
N. Dey. — 2018.
5. Hongliang L. A Fuzzy Comprehensive Evaluation Method of Maintenance Quality
Based on Improved Radar Chart / L. Hongliang, L. Anxin, Z. Bin et al. // 2008
ISECS International Colloquium on Computing, Communication, Control, and
Management.
6. Shi J. The domain decomposition method based on weighted average / J. Shi, Y. Liu,
W. Zhou // 2011 IEEE International Conference on Computer Science and Au-
tomation Engineering.
7. Harris J. An introduction to fuzzy logic applications / J. Harris. — Dordrecht: Klu-
wer Academic.
8. Anderson M. Fuzzy logic / M. Anderson. — Parkdale, OR: Black Opal Books, 2015.
9. Dimitrov V. Fuzzy logic: A framework for the new millennium / V. Dimitrov,
V. Korotkich. — Heidelberg: Physica-Verlag, 2011.
Received 10.05.2018
From the Editorial Board: the article corresponds completely to submitted manuscript.
|
| id | journaliasakpiua-article-139702 |
| institution | System research and information technologies |
| keywords_txt_mv | keywords |
| language | English |
| last_indexed | 2025-07-17T10:24:01Z |
| publishDate | 2018 |
| publisher | The National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute" |
| record_format | ojs |
| resource_txt_mv | journaliasakpiua/0f/10fb3e82394f67eaef063354281d960f.pdf |
| spelling | journaliasakpiua-article-1397022019-01-17T13:29:35Z Development of the performance prediction algorithms for cloud services Разработка алгоритма прогнозирования производительности облачных сервисов Розроблення алгоритму прогнозування продуктивності хмарних сервісів Zuev, Denis O. Kropachev, Artemii V. Usov, Aleksey Ye. Gorshunov, Roman A. data center service transactions index service resource occupancy index service performance index fuzzy judgment matrix Lyapunov exponent radar chart центр обработки данных коэффициент транзакции коэффициент использования машинных ресурсов коэффициент производительности сервиса матрица нечетких суждений экспонента Ляпунова радарная диаграмма центр обробки даних коефіцієнт транзакції коефіцієнт використання машинних ресурсів коефіцієнт продуктивності сервісу матриця нечітких суджень експонента Ляпунова радарна діаграма Main stages of data center service performance prediction were discussed, specifically data monitoring and gathering, calculation and prediction of key indexes and performance index prediction. It was proposed to build data center service performance prediction algorithm based on an analysis of the service transactions index, service resource occupancy index and service performance index. Prediction of the indexes is based on chaotic time series analysis that was used to estimate service transactions index time series trend, the radar chart method to calculate the service resource occupancy index value and weighted average method to calculate service performance index. For performance prediction, it is proposed to use a fuzzy judgment matrix with the service transactions index and service resource occupancy index as input values. It was taken into consideration that service transactions index is usually represented by nonlinear time series and thus the index time series parameters had to be predicted by chaos theory and for the calculation of this index, the estimation procedure of Lyapunov exponent value can be used. The radar chart demonstrates service resource occupancy index estimation of shared storage, mobile storage, memory, computational capability and network bandwidth. The prediction technique was based on the fuzzy nearness category that use input values of transactions index and dynamic changes of the service resource occupancy index. Проведен анализ основных этапов прогнозирования эффективности обслуживания центров обработки данных, в частности мониторинга и сбора данных, расчета и прогнозирования ключевых аспектов и прогнозирование коэффициентов производительности. Предложено построение алгоритма прогнозирования эффективности обслуживания центра обработки данных на основе анализа коэффициента транзакции, коэффициента использования машинных ресурсов и коэффициента производительности сервиса. Прогнозирование коэффициентов основано на анализе временных рядов, который использовался для оценки временных рядов коэффициента транзакций, метода радар-диаграммы для расчета значения коэффициента использования машинных ресурсов и средневзвешенного метода оценки для расчета коэффициента производительности сервиса. Для прогнозирования производительности предлагается использовать матрицу нечетких суждений с коэффициентом транзакций и коэффициентом занятости ресурса службы в качестве входных значений. Указано, что коэффициент служебных операций обычно представлен нелинейными временными рядами, и, следовательно, параметры временного ряда коэффициента должны быть предсказаны теорией хаоса, а значит для расчета этого коэффициента может быть использована процедура расчета экспоненты Ляпунова. Радарная диаграмма демонстрирует оценку коэффициента использования машинных ресурсов для общего хранилища данных, мобильных хранилищ, памяти, вычислительных возможностей и пропускной способности сети. Метод прогнозирования основывался на категории нечетких приближений, которые используют входные значения коэффициента транзакций и динамические изменения коэффициента использования машинных ресурсов. Виконано аналіз основних етапів прогнозування ефективності обслуговування центрів оброблення даних, зокрема моніторингу і збирання даних, розрахунку і прогнозу ключових аспектів та прогнозування коефіцієнтів продуктивності. Запропоновано побудову алгоритму прогнозування ефективності обслуговування центру оброблення даних на основі аналізу коефіцієнта транзакції, коефіцієнта використання машинних ресурсів і коефіцієнта продуктивності сервісу. Прогнозування коефіцієнтів ґрунтується на аналізі часових рядів, що використовувався для оцінювання часових рядів коефіцієнта транзакцій, методу радар-діаграми для розрахунку значення коефіцієнта застосування машинних ресурсів і середньозваженого методу оцінювання для розрахунку коефіцієнта продуктивності сервісу. Для прогнозування продуктивності запропоновано використати матрицю нечітких суджень з коефіцієнтом транзакцій і коефіцієнтом зайнятості ресурсу служби як вхідних значень. Указано, що коефіцієнт службових операцій подається нелінійними часовими рядами, і, отже, параметри часового ряду коефіцієнта мають бути передбачені теорією хаосу, а тому для розрахунку цього коефіцієнта можна застосувати процедуру розрахунку експоненти Ляпунова. Радарна діаграма демонструє оцінку коефіцієнта використання машинних ресурсів для загального сховища даних, мобільних сховищ, пам’яті обчислювальних можливостей і пропускної здатності мережі. Метод прогнозування ґрунтується на категорії нечітких наближень з використанням вхідних значень коефіцієнта транзакцій і динамічних змін коефіцієнта застосування машинних ресурсів. The National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute" 2018-06-20 Article Article application/pdf https://journal.iasa.kpi.ua/article/view/139702 10.20535/SRIT.2308-8893.2018.2.01 System research and information technologies; No. 2 (2018); 7-14 Системные исследования и информационные технологии; № 2 (2018); 7-14 Системні дослідження та інформаційні технології; № 2 (2018); 7-14 2308-8893 1681-6048 en https://journal.iasa.kpi.ua/article/view/139702/136682 Copyright (c) 2021 System research and information technologies |
| spellingShingle | центр обробки даних коефіцієнт транзакції коефіцієнт використання машинних ресурсів коефіцієнт продуктивності сервісу матриця нечітких суджень експонента Ляпунова радарна діаграма Zuev, Denis O. Kropachev, Artemii V. Usov, Aleksey Ye. Gorshunov, Roman A. Розроблення алгоритму прогнозування продуктивності хмарних сервісів |
| title | Розроблення алгоритму прогнозування продуктивності хмарних сервісів |
| title_alt | Development of the performance prediction algorithms for cloud services Разработка алгоритма прогнозирования производительности облачных сервисов |
| title_full | Розроблення алгоритму прогнозування продуктивності хмарних сервісів |
| title_fullStr | Розроблення алгоритму прогнозування продуктивності хмарних сервісів |
| title_full_unstemmed | Розроблення алгоритму прогнозування продуктивності хмарних сервісів |
| title_short | Розроблення алгоритму прогнозування продуктивності хмарних сервісів |
| title_sort | розроблення алгоритму прогнозування продуктивності хмарних сервісів |
| topic | центр обробки даних коефіцієнт транзакції коефіцієнт використання машинних ресурсів коефіцієнт продуктивності сервісу матриця нечітких суджень експонента Ляпунова радарна діаграма |
| topic_facet | data center service transactions index service resource occupancy index service performance index fuzzy judgment matrix Lyapunov exponent radar chart центр обработки данных коэффициент транзакции коэффициент использования машинных ресурсов коэффициент производительности сервиса матрица нечетких суждений экспонента Ляпунова радарная диаграмма центр обробки даних коефіцієнт транзакції коефіцієнт використання машинних ресурсів коефіцієнт продуктивності сервісу матриця нечітких суджень експонента Ляпунова радарна діаграма |
| url | https://journal.iasa.kpi.ua/article/view/139702 |
| work_keys_str_mv | AT zuevdeniso developmentoftheperformancepredictionalgorithmsforcloudservices AT kropachevartemiiv developmentoftheperformancepredictionalgorithmsforcloudservices AT usovalekseyye developmentoftheperformancepredictionalgorithmsforcloudservices AT gorshunovromana developmentoftheperformancepredictionalgorithmsforcloudservices AT zuevdeniso razrabotkaalgoritmaprognozirovaniâproizvoditelʹnostioblačnyhservisov AT kropachevartemiiv razrabotkaalgoritmaprognozirovaniâproizvoditelʹnostioblačnyhservisov AT usovalekseyye razrabotkaalgoritmaprognozirovaniâproizvoditelʹnostioblačnyhservisov AT gorshunovromana razrabotkaalgoritmaprognozirovaniâproizvoditelʹnostioblačnyhservisov AT zuevdeniso rozroblennâalgoritmuprognozuvannâproduktivnostíhmarnihservísív AT kropachevartemiiv rozroblennâalgoritmuprognozuvannâproduktivnostíhmarnihservísív AT usovalekseyye rozroblennâalgoritmuprognozuvannâproduktivnostíhmarnihservísív AT gorshunovromana rozroblennâalgoritmuprognozuvannâproduktivnostíhmarnihservísív |