EvoGWP: Predicting Long-Term Changes in Cloud Workloads Using Deep Graph-Evolution Learning
Workload prediction plays a crucial role in resource management of large scale cloud datacenters. Although quite a number of methods/algorithms have been proposed, long-term changes have not been explicitly identified and considered. Due to shifty user demands, workload re-locations, or other reason...
Gespeichert in:
Veröffentlicht in: | IEEE transactions on parallel and distributed systems 2024-03, Vol.35 (3), p.499-516 |
---|---|
Hauptverfasser: | , , , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
container_end_page | 516 |
---|---|
container_issue | 3 |
container_start_page | 499 |
container_title | IEEE transactions on parallel and distributed systems |
container_volume | 35 |
creator | Li, Jialun Yao, Jieqian Xiao, Danyang Yang, Diying Wu, Weigang |
description | Workload prediction plays a crucial role in resource management of large scale cloud datacenters. Although quite a number of methods/algorithms have been proposed, long-term changes have not been explicitly identified and considered. Due to shifty user demands, workload re-locations, or other reasons, the "resource usage pattern" of a workload, which is usually quite stable in a short-term view, may change dynamically in a long-term range. Such long-term dynamic changes may cause significant accuracy degradation for prediction algorithms. How to handle such long-term dynamic changes is an open and challenging issue. In this article, we propose Evolution Graph for Workload Prediction (EvoGWP), a novel method that can predict long-term dynamic changes using a delicately designed graph-based evolution learning algorithm. EvoGWP automatically extracts shapelets to explicitly identify resource usage patterns of workloads in a fine-grained level, and predicts workload changes by considering factors in both temporal and spatial dimensions. We design a two-level importance based shapelet extraction mechanism to mine new usage pattern changes in temporal dimension, and design a novel evolution graph model to fuse the interference among resource usage patterns of different workloads in spatial dimension. By combining temporal extraction of shapelets from each single workload and spatial interference of shapelets among different workloads, we then design a spatio-temporal GNN-based encoder-decoder model to predict the long-term dynamic changes of workloads. Experiments using real trace data from Alibaba, Tencent and Google show that EvoGWP improves the prediction accuracy by up to 58.6% over the state-of-the-art prediction methods. Moreover, EvoGWP can outperform the state-of-the-art prediction methods in terms of model convergence. To the best of our knowledge, this is the first work that explicitly identifies fine-grained workload resource usage patterns to accurately predict long-term dynamic changes of workloads. |
doi_str_mv | 10.1109/TPDS.2024.3357715 |
format | Article |
fullrecord | <record><control><sourceid>proquest_RIE</sourceid><recordid>TN_cdi_proquest_journals_2924040238</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>10412676</ieee_id><sourcerecordid>2924040238</sourcerecordid><originalsourceid>FETCH-LOGICAL-c246t-b50adb787aa9757e18b311612c537475224e436a4b1022ba48e585460895ebf23</originalsourceid><addsrcrecordid>eNpNkE1Lw0AQhhdRsFZ_gOBhwXPqfmY33qTWKgQs2NKDh2WTTNvUNBt3E8F_b0J78DQD87zvwIPQLSUTSknysFw8f0wYYWLCuVSKyjM0olLqiFHNz_udCBkljCaX6CqEPSFUSCJG6HP24-brxSNeeCjKvC3rLU5dvY2W4A94urP1FgIuazytXFfgtfNflbNFwKswoM8ADZ572-yivqjq2tLVOAXr6_56jS42tgpwc5pjtHqZLaevUfo-f5s-pVHORNxGmSS2yJRW1iZKKqA645TGlOWSK6EkYwIEj63IKGEss0KD1FLERCcSsg3jY3R_7G28--4gtGbvOl_3Lw1LmCCCMK57ih6p3LsQPGxM48uD9b-GEjM4NINDMzg0J4d95u6YKQHgHy8oi1XM_wAZy2ts</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2924040238</pqid></control><display><type>article</type><title>EvoGWP: Predicting Long-Term Changes in Cloud Workloads Using Deep Graph-Evolution Learning</title><source>IEEE Electronic Library (IEL)</source><creator>Li, Jialun ; Yao, Jieqian ; Xiao, Danyang ; Yang, Diying ; Wu, Weigang</creator><creatorcontrib>Li, Jialun ; Yao, Jieqian ; Xiao, Danyang ; Yang, Diying ; Wu, Weigang</creatorcontrib><description>Workload prediction plays a crucial role in resource management of large scale cloud datacenters. Although quite a number of methods/algorithms have been proposed, long-term changes have not been explicitly identified and considered. Due to shifty user demands, workload re-locations, or other reasons, the "resource usage pattern" of a workload, which is usually quite stable in a short-term view, may change dynamically in a long-term range. Such long-term dynamic changes may cause significant accuracy degradation for prediction algorithms. How to handle such long-term dynamic changes is an open and challenging issue. In this article, we propose Evolution Graph for Workload Prediction (EvoGWP), a novel method that can predict long-term dynamic changes using a delicately designed graph-based evolution learning algorithm. EvoGWP automatically extracts shapelets to explicitly identify resource usage patterns of workloads in a fine-grained level, and predicts workload changes by considering factors in both temporal and spatial dimensions. We design a two-level importance based shapelet extraction mechanism to mine new usage pattern changes in temporal dimension, and design a novel evolution graph model to fuse the interference among resource usage patterns of different workloads in spatial dimension. By combining temporal extraction of shapelets from each single workload and spatial interference of shapelets among different workloads, we then design a spatio-temporal GNN-based encoder-decoder model to predict the long-term dynamic changes of workloads. Experiments using real trace data from Alibaba, Tencent and Google show that EvoGWP improves the prediction accuracy by up to 58.6% over the state-of-the-art prediction methods. Moreover, EvoGWP can outperform the state-of-the-art prediction methods in terms of model convergence. To the best of our knowledge, this is the first work that explicitly identifies fine-grained workload resource usage patterns to accurately predict long-term dynamic changes of workloads.</description><identifier>ISSN: 1045-9219</identifier><identifier>EISSN: 1558-2183</identifier><identifier>DOI: 10.1109/TPDS.2024.3357715</identifier><identifier>CODEN: ITDSEO</identifier><language>eng</language><publisher>New York: IEEE</publisher><subject>Algorithms ; cloud computing ; Correlation ; datacenter ; Dynamic scheduling ; Encoders-Decoders ; Evolutionary algorithms ; graph neural network ; Graph neural networks ; Heuristic algorithms ; Interference ; Machine learning ; Prediction algorithms ; Predictions ; Predictive models ; Resource management ; State of the art ; Use statistics ; Workload ; Workload prediction ; Workloads</subject><ispartof>IEEE transactions on parallel and distributed systems, 2024-03, Vol.35 (3), p.499-516</ispartof><rights>Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2024</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><cites>FETCH-LOGICAL-c246t-b50adb787aa9757e18b311612c537475224e436a4b1022ba48e585460895ebf23</cites><orcidid>0000-0002-4714-7021 ; 0009-0006-8640-3530 ; 0000-0001-6314-3621 ; 0000-0003-0941-9820 ; 0000-0001-6798-9683</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/10412676$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>314,780,784,796,27924,27925,54758</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/10412676$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Li, Jialun</creatorcontrib><creatorcontrib>Yao, Jieqian</creatorcontrib><creatorcontrib>Xiao, Danyang</creatorcontrib><creatorcontrib>Yang, Diying</creatorcontrib><creatorcontrib>Wu, Weigang</creatorcontrib><title>EvoGWP: Predicting Long-Term Changes in Cloud Workloads Using Deep Graph-Evolution Learning</title><title>IEEE transactions on parallel and distributed systems</title><addtitle>TPDS</addtitle><description>Workload prediction plays a crucial role in resource management of large scale cloud datacenters. Although quite a number of methods/algorithms have been proposed, long-term changes have not been explicitly identified and considered. Due to shifty user demands, workload re-locations, or other reasons, the "resource usage pattern" of a workload, which is usually quite stable in a short-term view, may change dynamically in a long-term range. Such long-term dynamic changes may cause significant accuracy degradation for prediction algorithms. How to handle such long-term dynamic changes is an open and challenging issue. In this article, we propose Evolution Graph for Workload Prediction (EvoGWP), a novel method that can predict long-term dynamic changes using a delicately designed graph-based evolution learning algorithm. EvoGWP automatically extracts shapelets to explicitly identify resource usage patterns of workloads in a fine-grained level, and predicts workload changes by considering factors in both temporal and spatial dimensions. We design a two-level importance based shapelet extraction mechanism to mine new usage pattern changes in temporal dimension, and design a novel evolution graph model to fuse the interference among resource usage patterns of different workloads in spatial dimension. By combining temporal extraction of shapelets from each single workload and spatial interference of shapelets among different workloads, we then design a spatio-temporal GNN-based encoder-decoder model to predict the long-term dynamic changes of workloads. Experiments using real trace data from Alibaba, Tencent and Google show that EvoGWP improves the prediction accuracy by up to 58.6% over the state-of-the-art prediction methods. Moreover, EvoGWP can outperform the state-of-the-art prediction methods in terms of model convergence. To the best of our knowledge, this is the first work that explicitly identifies fine-grained workload resource usage patterns to accurately predict long-term dynamic changes of workloads.</description><subject>Algorithms</subject><subject>cloud computing</subject><subject>Correlation</subject><subject>datacenter</subject><subject>Dynamic scheduling</subject><subject>Encoders-Decoders</subject><subject>Evolutionary algorithms</subject><subject>graph neural network</subject><subject>Graph neural networks</subject><subject>Heuristic algorithms</subject><subject>Interference</subject><subject>Machine learning</subject><subject>Prediction algorithms</subject><subject>Predictions</subject><subject>Predictive models</subject><subject>Resource management</subject><subject>State of the art</subject><subject>Use statistics</subject><subject>Workload</subject><subject>Workload prediction</subject><subject>Workloads</subject><issn>1045-9219</issn><issn>1558-2183</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>RIE</sourceid><recordid>eNpNkE1Lw0AQhhdRsFZ_gOBhwXPqfmY33qTWKgQs2NKDh2WTTNvUNBt3E8F_b0J78DQD87zvwIPQLSUTSknysFw8f0wYYWLCuVSKyjM0olLqiFHNz_udCBkljCaX6CqEPSFUSCJG6HP24-brxSNeeCjKvC3rLU5dvY2W4A94urP1FgIuazytXFfgtfNflbNFwKswoM8ADZ572-yivqjq2tLVOAXr6_56jS42tgpwc5pjtHqZLaevUfo-f5s-pVHORNxGmSS2yJRW1iZKKqA645TGlOWSK6EkYwIEj63IKGEss0KD1FLERCcSsg3jY3R_7G28--4gtGbvOl_3Lw1LmCCCMK57ih6p3LsQPGxM48uD9b-GEjM4NINDMzg0J4d95u6YKQHgHy8oi1XM_wAZy2ts</recordid><startdate>20240301</startdate><enddate>20240301</enddate><creator>Li, Jialun</creator><creator>Yao, Jieqian</creator><creator>Xiao, Danyang</creator><creator>Yang, Diying</creator><creator>Wu, Weigang</creator><general>IEEE</general><general>The Institute of Electrical and Electronics Engineers, Inc. (IEEE)</general><scope>97E</scope><scope>RIA</scope><scope>RIE</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7SC</scope><scope>7SP</scope><scope>8FD</scope><scope>JQ2</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope><orcidid>https://orcid.org/0000-0002-4714-7021</orcidid><orcidid>https://orcid.org/0009-0006-8640-3530</orcidid><orcidid>https://orcid.org/0000-0001-6314-3621</orcidid><orcidid>https://orcid.org/0000-0003-0941-9820</orcidid><orcidid>https://orcid.org/0000-0001-6798-9683</orcidid></search><sort><creationdate>20240301</creationdate><title>EvoGWP: Predicting Long-Term Changes in Cloud Workloads Using Deep Graph-Evolution Learning</title><author>Li, Jialun ; Yao, Jieqian ; Xiao, Danyang ; Yang, Diying ; Wu, Weigang</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c246t-b50adb787aa9757e18b311612c537475224e436a4b1022ba48e585460895ebf23</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Algorithms</topic><topic>cloud computing</topic><topic>Correlation</topic><topic>datacenter</topic><topic>Dynamic scheduling</topic><topic>Encoders-Decoders</topic><topic>Evolutionary algorithms</topic><topic>graph neural network</topic><topic>Graph neural networks</topic><topic>Heuristic algorithms</topic><topic>Interference</topic><topic>Machine learning</topic><topic>Prediction algorithms</topic><topic>Predictions</topic><topic>Predictive models</topic><topic>Resource management</topic><topic>State of the art</topic><topic>Use statistics</topic><topic>Workload</topic><topic>Workload prediction</topic><topic>Workloads</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Li, Jialun</creatorcontrib><creatorcontrib>Yao, Jieqian</creatorcontrib><creatorcontrib>Xiao, Danyang</creatorcontrib><creatorcontrib>Yang, Diying</creatorcontrib><creatorcontrib>Wu, Weigang</creatorcontrib><collection>IEEE All-Society Periodicals Package (ASPP) 2005-present</collection><collection>IEEE All-Society Periodicals Package (ASPP) 1998-Present</collection><collection>IEEE Electronic Library (IEL)</collection><collection>CrossRef</collection><collection>Computer and Information Systems Abstracts</collection><collection>Electronics & Communications Abstracts</collection><collection>Technology Research Database</collection><collection>ProQuest Computer Science Collection</collection><collection>Advanced Technologies Database with Aerospace</collection><collection>Computer and Information Systems Abstracts Academic</collection><collection>Computer and Information Systems Abstracts Professional</collection><jtitle>IEEE transactions on parallel and distributed systems</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Li, Jialun</au><au>Yao, Jieqian</au><au>Xiao, Danyang</au><au>Yang, Diying</au><au>Wu, Weigang</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>EvoGWP: Predicting Long-Term Changes in Cloud Workloads Using Deep Graph-Evolution Learning</atitle><jtitle>IEEE transactions on parallel and distributed systems</jtitle><stitle>TPDS</stitle><date>2024-03-01</date><risdate>2024</risdate><volume>35</volume><issue>3</issue><spage>499</spage><epage>516</epage><pages>499-516</pages><issn>1045-9219</issn><eissn>1558-2183</eissn><coden>ITDSEO</coden><abstract>Workload prediction plays a crucial role in resource management of large scale cloud datacenters. Although quite a number of methods/algorithms have been proposed, long-term changes have not been explicitly identified and considered. Due to shifty user demands, workload re-locations, or other reasons, the "resource usage pattern" of a workload, which is usually quite stable in a short-term view, may change dynamically in a long-term range. Such long-term dynamic changes may cause significant accuracy degradation for prediction algorithms. How to handle such long-term dynamic changes is an open and challenging issue. In this article, we propose Evolution Graph for Workload Prediction (EvoGWP), a novel method that can predict long-term dynamic changes using a delicately designed graph-based evolution learning algorithm. EvoGWP automatically extracts shapelets to explicitly identify resource usage patterns of workloads in a fine-grained level, and predicts workload changes by considering factors in both temporal and spatial dimensions. We design a two-level importance based shapelet extraction mechanism to mine new usage pattern changes in temporal dimension, and design a novel evolution graph model to fuse the interference among resource usage patterns of different workloads in spatial dimension. By combining temporal extraction of shapelets from each single workload and spatial interference of shapelets among different workloads, we then design a spatio-temporal GNN-based encoder-decoder model to predict the long-term dynamic changes of workloads. Experiments using real trace data from Alibaba, Tencent and Google show that EvoGWP improves the prediction accuracy by up to 58.6% over the state-of-the-art prediction methods. Moreover, EvoGWP can outperform the state-of-the-art prediction methods in terms of model convergence. To the best of our knowledge, this is the first work that explicitly identifies fine-grained workload resource usage patterns to accurately predict long-term dynamic changes of workloads.</abstract><cop>New York</cop><pub>IEEE</pub><doi>10.1109/TPDS.2024.3357715</doi><tpages>18</tpages><orcidid>https://orcid.org/0000-0002-4714-7021</orcidid><orcidid>https://orcid.org/0009-0006-8640-3530</orcidid><orcidid>https://orcid.org/0000-0001-6314-3621</orcidid><orcidid>https://orcid.org/0000-0003-0941-9820</orcidid><orcidid>https://orcid.org/0000-0001-6798-9683</orcidid></addata></record> |
fulltext | fulltext_linktorsrc |
identifier | ISSN: 1045-9219 |
ispartof | IEEE transactions on parallel and distributed systems, 2024-03, Vol.35 (3), p.499-516 |
issn | 1045-9219 1558-2183 |
language | eng |
recordid | cdi_proquest_journals_2924040238 |
source | IEEE Electronic Library (IEL) |
subjects | Algorithms cloud computing Correlation datacenter Dynamic scheduling Encoders-Decoders Evolutionary algorithms graph neural network Graph neural networks Heuristic algorithms Interference Machine learning Prediction algorithms Predictions Predictive models Resource management State of the art Use statistics Workload Workload prediction Workloads |
title | EvoGWP: Predicting Long-Term Changes in Cloud Workloads Using Deep Graph-Evolution Learning |
url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-25T18%3A04%3A51IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_RIE&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=EvoGWP:%20Predicting%20Long-Term%20Changes%20in%20Cloud%20Workloads%20Using%20Deep%20Graph-Evolution%20Learning&rft.jtitle=IEEE%20transactions%20on%20parallel%20and%20distributed%20systems&rft.au=Li,%20Jialun&rft.date=2024-03-01&rft.volume=35&rft.issue=3&rft.spage=499&rft.epage=516&rft.pages=499-516&rft.issn=1045-9219&rft.eissn=1558-2183&rft.coden=ITDSEO&rft_id=info:doi/10.1109/TPDS.2024.3357715&rft_dat=%3Cproquest_RIE%3E2924040238%3C/proquest_RIE%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2924040238&rft_id=info:pmid/&rft_ieee_id=10412676&rfr_iscdi=true |