Pub Date : 2018-06-01DOI: 10.23919/TMA.2018.8506572
Sarah Wassermann, P. Casas
Internet path changes are frequently linked to path inflation and performance degradation; therefore, predicting their occurrence is highly relevant for performance monitoring and dynamic traffic engineering. In this paper we showcase Dis-NETPerf and NETPerfTrace, two different and complementary tools for distributed Internet paths performance analysis, using machine learning models.
{"title":"Distributed Internet Paths Performance Analysis Through Machine Learning","authors":"Sarah Wassermann, P. Casas","doi":"10.23919/TMA.2018.8506572","DOIUrl":"https://doi.org/10.23919/TMA.2018.8506572","url":null,"abstract":"Internet path changes are frequently linked to path inflation and performance degradation; therefore, predicting their occurrence is highly relevant for performance monitoring and dynamic traffic engineering. In this paper we showcase Dis-NETPerf and NETPerfTrace, two different and complementary tools for distributed Internet paths performance analysis, using machine learning models.","PeriodicalId":6607,"journal":{"name":"2018 Network Traffic Measurement and Analysis Conference (TMA)","volume":" 1","pages":"1-2"},"PeriodicalIF":0.0,"publicationDate":"2018-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"91414144","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-06-01DOI: 10.23919/TMA.2018.8506493
Diana Andreea Popescu, A. Moore
Increased network latency and packets losses can affect substantially application performance. Due to the scale of data centers, custom network monitoring tools have been developed to measure network latency and packet loss. In our previous work, we used the Precision Time Protocol (PTP) to measure one-way delay and to quantify packet loss ratios, and we proposed PTPmesh as a cloud network monitoring tool. In this work, we provide a better understanding on how to exploit the measurement data offered by PTPmesh and present a detailed analysis of PTPmesh measurements collected in ten data centers from three cloud providers. Our findings reveal different latency, latency variance and packet loss characteristics across data centers. Through our analysis, we showcase the strengths and limitations of PTPmesh as a cloud network monitoring tool. To foster further research in this area, we make our dataset available.
{"title":"A First Look at Data Center Network Condition Through The Eyes of PTPmesh","authors":"Diana Andreea Popescu, A. Moore","doi":"10.23919/TMA.2018.8506493","DOIUrl":"https://doi.org/10.23919/TMA.2018.8506493","url":null,"abstract":"Increased network latency and packets losses can affect substantially application performance. Due to the scale of data centers, custom network monitoring tools have been developed to measure network latency and packet loss. In our previous work, we used the Precision Time Protocol (PTP) to measure one-way delay and to quantify packet loss ratios, and we proposed PTPmesh as a cloud network monitoring tool. In this work, we provide a better understanding on how to exploit the measurement data offered by PTPmesh and present a detailed analysis of PTPmesh measurements collected in ten data centers from three cloud providers. Our findings reveal different latency, latency variance and packet loss characteristics across data centers. Through our analysis, we showcase the strengths and limitations of PTPmesh as a cloud network monitoring tool. To foster further research in this area, we make our dataset available.","PeriodicalId":6607,"journal":{"name":"2018 Network Traffic Measurement and Analysis Conference (TMA)","volume":"27 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2018-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82047651","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-06-01DOI: 10.23919/TMA.2018.8506488
Michael Seufert, Bernd Zeidler, Florian Wamser, Theodoros Karagkioules, D. Tsilimantos, Frank Loh, P. Tran-Gia, S. Valentin
YouTube is one of the most popular and demanding services in the Internet today. Thereby, a large portion of this traffic is generated by YouTube's mobile app. While past studies have shown how to monitor browser-based streaming on desktop PCs (e.g., YoMo) or mobile devices (e.g., YoMoApp), streaming in the native app has not been monitored yet. This paper presents an automated framework for monitoring the streaming in YouTube's native app for Android. The concept is based on a wrapper application and the Android Debug Bridge (adb), and can be also extended to automatic measurements with other apps. For YouTube, it allows to collect application-layer streaming data, such as current playtime, buffered playtime, video encoding, and quality switches. These data can be complemented with network measurements on the mobile access link to obtain a holistic view on mobile YouTube streaming on Android devices. In addition to describing the software design and testbed setup, this paper discusses an experimental measurement. This study analyzes the streaming in the native YouTube app and compares it to the streaming from the mobile YouTube website via YoMoApp.
{"title":"A Wrapper for Automatic Measurements with YouTube's Native Android App","authors":"Michael Seufert, Bernd Zeidler, Florian Wamser, Theodoros Karagkioules, D. Tsilimantos, Frank Loh, P. Tran-Gia, S. Valentin","doi":"10.23919/TMA.2018.8506488","DOIUrl":"https://doi.org/10.23919/TMA.2018.8506488","url":null,"abstract":"YouTube is one of the most popular and demanding services in the Internet today. Thereby, a large portion of this traffic is generated by YouTube's mobile app. While past studies have shown how to monitor browser-based streaming on desktop PCs (e.g., YoMo) or mobile devices (e.g., YoMoApp), streaming in the native app has not been monitored yet. This paper presents an automated framework for monitoring the streaming in YouTube's native app for Android. The concept is based on a wrapper application and the Android Debug Bridge (adb), and can be also extended to automatic measurements with other apps. For YouTube, it allows to collect application-layer streaming data, such as current playtime, buffered playtime, video encoding, and quality switches. These data can be complemented with network measurements on the mobile access link to obtain a holistic view on mobile YouTube streaming on Android devices. In addition to describing the software design and testbed setup, this paper discusses an experimental measurement. This study analyzes the streaming in the native YouTube app and compares it to the streaming from the mobile YouTube website via YoMoApp.","PeriodicalId":6607,"journal":{"name":"2018 Network Traffic Measurement and Analysis Conference (TMA)","volume":"39 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2018-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81296540","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-06-01DOI: 10.23919/TMA.2018.8506532
M. Kühlewind, Michael Walter, Iain R. Learmonth, B. Trammell
Investigating Internet Path Transparency means measuring if a network path between two endhosts is impaired by in-network functions on the path. A path is considered transparent if it provides connectivity and the same performance independent of the protocol or protocol stack that is used for the transmission. Unfortunately this is not always the case. Simple firewalls that block e.g. UDP, are an example. Of course such in-network functions are often valuable, like firewalls. However, these middleboxes also, sometimes unintentionally, make assumptions about the traffic passing through them that restricts innovation in the Internet on the higher layers, e.g. the deployment of new UDP-based protocols such as QUIC, to stick with the previous example. PATHspider is an active measurement tool to test for Path Transparency. In this paper we present a new feature of PATH-spider that integrates tracebox-based functionality and analysis to not only detect in-transparency but also further locate the origin of the impairment observed. As an example study we show updated and extended measurements on ECN support and connectivity. By using our enhanced ECN PATHspider plugin to test network support of the ECN IP codepoint and additional path tracing that is correlated with DSCP testing, we show that most in-network ECN IP codepoint zeroing is due to use of the deprecated definition of the IP ToS field for domain-internal service differentiation, while pure resetting of the ECN IP field is more likely an active inference in border networks.
{"title":"Tracing Internet Path Transparency","authors":"M. Kühlewind, Michael Walter, Iain R. Learmonth, B. Trammell","doi":"10.23919/TMA.2018.8506532","DOIUrl":"https://doi.org/10.23919/TMA.2018.8506532","url":null,"abstract":"Investigating Internet Path Transparency means measuring if a network path between two endhosts is impaired by in-network functions on the path. A path is considered transparent if it provides connectivity and the same performance independent of the protocol or protocol stack that is used for the transmission. Unfortunately this is not always the case. Simple firewalls that block e.g. UDP, are an example. Of course such in-network functions are often valuable, like firewalls. However, these middleboxes also, sometimes unintentionally, make assumptions about the traffic passing through them that restricts innovation in the Internet on the higher layers, e.g. the deployment of new UDP-based protocols such as QUIC, to stick with the previous example. PATHspider is an active measurement tool to test for Path Transparency. In this paper we present a new feature of PATH-spider that integrates tracebox-based functionality and analysis to not only detect in-transparency but also further locate the origin of the impairment observed. As an example study we show updated and extended measurements on ECN support and connectivity. By using our enhanced ECN PATHspider plugin to test network support of the ECN IP codepoint and additional path tracing that is correlated with DSCP testing, we show that most in-network ECN IP codepoint zeroing is due to use of the deprecated definition of the IP ToS field for domain-internal service differentiation, while pure resetting of the ECN IP field is more likely an active inference in border networks.","PeriodicalId":6607,"journal":{"name":"2018 Network Traffic Measurement and Analysis Conference (TMA)","volume":"30 1","pages":"1-7"},"PeriodicalIF":0.0,"publicationDate":"2018-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83577332","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-06-01DOI: 10.23919/TMA.2018.8506498
Milan Cermák, Tomás Jirsík, P. Velan, Jana Komárková, Stanislav Špaček, Martin Drasar, Tomáš Plesník
Research in network traffic measurement and analysis is a long-lasting field with growing interest from both scientists and the industry. However, even after so many years, results replication, criticism, and review are still rare. We face not only a lack of research standards, but also inaccessibility of appropriate datasets that can be used for methods development and evaluation. Therefore, a lot of potentially high-quality research cannot be verified and is not adopted by the industry or the community. The aim of this paper is to overcome this controversy with a unique solution based on a combination of distinct approaches proposed by other research works. Unlike these studies, we focus on the whole issue covering all areas of data anonymization, authenticity, recency, publicity, and their usage for research provability. We believe that these challenges can be solved by utilization of semi-labeled datasets composed of real-world network traffic and annotated units with interest-related packet traces only. In this paper, we outline the basic ideas of the methodology from unit trace collection and semi-labeled dataset creation to its usage for research evaluation. We strive for this proposal to start a discussion of the approach and help to overcome some of the challenges the research faces today.
{"title":"Towards Provable Network Traffic Measurement and Analysis via Semi-Labeled Trace Datasets","authors":"Milan Cermák, Tomás Jirsík, P. Velan, Jana Komárková, Stanislav Špaček, Martin Drasar, Tomáš Plesník","doi":"10.23919/TMA.2018.8506498","DOIUrl":"https://doi.org/10.23919/TMA.2018.8506498","url":null,"abstract":"Research in network traffic measurement and analysis is a long-lasting field with growing interest from both scientists and the industry. However, even after so many years, results replication, criticism, and review are still rare. We face not only a lack of research standards, but also inaccessibility of appropriate datasets that can be used for methods development and evaluation. Therefore, a lot of potentially high-quality research cannot be verified and is not adopted by the industry or the community. The aim of this paper is to overcome this controversy with a unique solution based on a combination of distinct approaches proposed by other research works. Unlike these studies, we focus on the whole issue covering all areas of data anonymization, authenticity, recency, publicity, and their usage for research provability. We believe that these challenges can be solved by utilization of semi-labeled datasets composed of real-world network traffic and annotated units with interest-related packet traces only. In this paper, we outline the basic ideas of the methodology from unit trace collection and semi-labeled dataset creation to its usage for research evaluation. We strive for this proposal to start a discussion of the approach and help to overcome some of the challenges the research faces today.","PeriodicalId":6607,"journal":{"name":"2018 Network Traffic Measurement and Analysis Conference (TMA)","volume":"48 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2018-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80252954","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-06-01DOI: 10.23919/TMA.2018.8506563
M. Naldi
Service quality commitments in cloud service provisioning are typically described in Service Level Agreements (SLA). Service availability is always a major parameter to be included in such SLAs. and the cloud provider is bounded to guarantee a minimum availability value, for which current cloud monitoring systems employ a naive estimator. In this paper a new estimation method is proposed for service availability, which is based on the bootstrap technique and employs a non-parametric statistical hypothesis test. Through Monte Carlo simulation, the method is shown to be much more accurate than the naive one under three stochastic models for the durations of operating and outage periods, exhibiting a Type I error probability lower than 1 % in most cases, while the naive estimator yields error probabilities around 40%.
{"title":"Non-Parametric Bootstrap Detection of Availability Service Level Objective Violations in Cloud Storage","authors":"M. Naldi","doi":"10.23919/TMA.2018.8506563","DOIUrl":"https://doi.org/10.23919/TMA.2018.8506563","url":null,"abstract":"Service quality commitments in cloud service provisioning are typically described in Service Level Agreements (SLA). Service availability is always a major parameter to be included in such SLAs. and the cloud provider is bounded to guarantee a minimum availability value, for which current cloud monitoring systems employ a naive estimator. In this paper a new estimation method is proposed for service availability, which is based on the bootstrap technique and employs a non-parametric statistical hypothesis test. Through Monte Carlo simulation, the method is shown to be much more accurate than the naive one under three stochastic models for the durations of operating and outage periods, exhibiting a Type I error probability lower than 1 % in most cases, while the naive estimator yields error probabilities around 40%.","PeriodicalId":6607,"journal":{"name":"2018 Network Traffic Measurement and Analysis Conference (TMA)","volume":"6 1","pages":"1-7"},"PeriodicalIF":0.0,"publicationDate":"2018-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86691937","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-06-01DOI: 10.23919/TMA.2018.8506549
Jan Rüth, O. Hohlfeld
Driven by their quest to improve web performance, Content Delivery Networks (CDNs) are known adaptors of performance optimizations. In this regard, TCP congestion control and particularly its initial congestion window (IW) size is one long-debated topic that can influence CDN performance. Its size is, however, assumed to be static by IETF recommendations-despite being network- and application-dependent-and only infrequently changed in its history. To understand if the standardization and research perspective still meets Internet reality, we study the IW configurations of major CDNs. Our study uses a globally distributed infrastructure of VPNs giving access to residential access links that enable to shed light on network-dependent configurations. We observe that most CDNs are well aware of the IW's impact and find a high amount of customization that is beyond current Internet standards. Further, we find CDNs that utilize different IWs for different customers and content while others resort to fixed values. We find various initial window configurations, most below 50 segments yet with exceptions of up to 100 segments—the tenfold of current standards. Our study highlights that Internet reality drifted away from recommended and standardized practices.
{"title":"Demystifying TCP Initial Window Configurations of Content Distribution Networks","authors":"Jan Rüth, O. Hohlfeld","doi":"10.23919/TMA.2018.8506549","DOIUrl":"https://doi.org/10.23919/TMA.2018.8506549","url":null,"abstract":"Driven by their quest to improve web performance, Content Delivery Networks (CDNs) are known adaptors of performance optimizations. In this regard, TCP congestion control and particularly its initial congestion window (IW) size is one long-debated topic that can influence CDN performance. Its size is, however, assumed to be static by IETF recommendations-despite being network- and application-dependent-and only infrequently changed in its history. To understand if the standardization and research perspective still meets Internet reality, we study the IW configurations of major CDNs. Our study uses a globally distributed infrastructure of VPNs giving access to residential access links that enable to shed light on network-dependent configurations. We observe that most CDNs are well aware of the IW's impact and find a high amount of customization that is beyond current Internet standards. Further, we find CDNs that utilize different IWs for different customers and content while others resort to fixed values. We find various initial window configurations, most below 50 segments yet with exceptions of up to 100 segments—the tenfold of current standards. Our study highlights that Internet reality drifted away from recommended and standardized practices.","PeriodicalId":6607,"journal":{"name":"2018 Network Traffic Measurement and Analysis Conference (TMA)","volume":"7 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2018-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88947552","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-06-01DOI: 10.23919/TMA.2018.8506486
P. Casas
The application of Machine Learning (ML) models to the analysis of network measurement problems has largely increased in the last decade; however, there is still no clear best-practice or silver bullet approach to address these problems in a general context, and only adhoc and very tailored approaches have been evaluated so far. While deep-learning models have provided a major breakthrough in highly-dimensional problems such as image processing, it is difficult to say today which is the best model or most fitted category of models to address the analysis of large volumes of highly-dimensional data collected in operational networks. In this paper we evaluate and benchmark different ML models applied to the analysis of three different and assorted network measurement problems, including detection of network attacks, detection of smartphone-apps anomalies and QoE prediction in cellular networks. We consider an extensive battery of ML models, including both supervised and semi-supervised techniques, as well as ML ensembles such as bagging, boosting and stacking. Proposed models are evaluated using real network measurements coming from operational networks. Results suggest that both neural networks and decision-tree-based models provide in general better results in terms of accuracy and prediction, with a much smaller computation overhead for decision trees as compared to models based on neural networks or support vector machines. In addition, collaborative models taking advantage of multiple machine learning algorithms, and in particular stacking models, are more robust and perform better than single ML models, pointing out the benefits of a crowd as compared to individual models.
{"title":"On the Analysis of Network Measurements Through Machine Learning: The Power of the Crowd","authors":"P. Casas","doi":"10.23919/TMA.2018.8506486","DOIUrl":"https://doi.org/10.23919/TMA.2018.8506486","url":null,"abstract":"The application of Machine Learning (ML) models to the analysis of network measurement problems has largely increased in the last decade; however, there is still no clear best-practice or silver bullet approach to address these problems in a general context, and only adhoc and very tailored approaches have been evaluated so far. While deep-learning models have provided a major breakthrough in highly-dimensional problems such as image processing, it is difficult to say today which is the best model or most fitted category of models to address the analysis of large volumes of highly-dimensional data collected in operational networks. In this paper we evaluate and benchmark different ML models applied to the analysis of three different and assorted network measurement problems, including detection of network attacks, detection of smartphone-apps anomalies and QoE prediction in cellular networks. We consider an extensive battery of ML models, including both supervised and semi-supervised techniques, as well as ML ensembles such as bagging, boosting and stacking. Proposed models are evaluated using real network measurements coming from operational networks. Results suggest that both neural networks and decision-tree-based models provide in general better results in terms of accuracy and prediction, with a much smaller computation overhead for decision trees as compared to models based on neural networks or support vector machines. In addition, collaborative models taking advantage of multiple machine learning algorithms, and in particular stacking models, are more robust and perform better than single ML models, pointing out the benefits of a crowd as compared to individual models.","PeriodicalId":6607,"journal":{"name":"2018 Network Traffic Measurement and Analysis Conference (TMA)","volume":"24 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2018-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73480614","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-06-01DOI: 10.23919/TMA.2018.8506552
Naz Albayrak, E. Zeydan
In this demo, we are proposing an intelligent heatmap service that can be utilized by Mobile Network Operators (MNOs). The demo platform consists of big data enabled ecosystem that can perform analytics over MNO's data which can be serviced to third parties as location based service. As a use case scenario, we are investigating how does the density of mobile subscriber change during the day of the week in Istanbul's historic peninsula. Our results interestingly indicate that based on the heatmap densities built by number of signals received from mobile subscribers, weekdays can be more crowded than weekends in historical touristic locations of Istanbul.
{"title":"An Intelligent Data Visualization Service Platform for Mobile Network Operators","authors":"Naz Albayrak, E. Zeydan","doi":"10.23919/TMA.2018.8506552","DOIUrl":"https://doi.org/10.23919/TMA.2018.8506552","url":null,"abstract":"In this demo, we are proposing an intelligent heatmap service that can be utilized by Mobile Network Operators (MNOs). The demo platform consists of big data enabled ecosystem that can perform analytics over MNO's data which can be serviced to third parties as location based service. As a use case scenario, we are investigating how does the density of mobile subscriber change during the day of the week in Istanbul's historic peninsula. Our results interestingly indicate that based on the heatmap densities built by number of signals received from mobile subscribers, weekdays can be more crowded than weekends in historical touristic locations of Istanbul.","PeriodicalId":6607,"journal":{"name":"2018 Network Traffic Measurement and Analysis Conference (TMA)","volume":"255 1","pages":"1-2"},"PeriodicalIF":0.0,"publicationDate":"2018-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77536012","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-06-01DOI: 10.23919/TMA.2018.8506565
Tianzhu Zhang, Leonardo Linguaglossa, Massimo Gallo, P. Giaccone, D. Rossi
Testing experimental network devices requires deep performance analysis, which is usually performed with expensive, not flexible, hardware equipment. With the advent of highspeed packet I/O frameworks, general purpose equipments have narrowed the performance gap in respect of dedicated hardware and a variety of software-based solutions have emerged for handling traffic at very high speed. While the literature abounds with software traffic generators, existing monitoring solutions do not target worst-case scenarios (i.e., 64B packets at line rate) that are particularly relevant for stress-testing high-speed network functions, or occupy too many resources. In this paper we first analyse the design space for high-speed traffic monitoring that leads us to specific choices characterizing FlowMon-DPDK, a DPDK-based software traffic monitor that we release as an open source project. In a nutshell, FlowMon-DPDK provides tunable fine-grained statistics at both packet and flow levels. Experimental results demonstrate that our traffic monitor is able to provide per-flow statistics with 5-nines precision at high-speed (14.88 Mpps) using an exiguous amount of resources. Finally, we showcase FlowMon-DPDK usage by testing two open source prototypes for stateful flow-level end-host and in-network packet processing.
{"title":"FlowMon-DPDK: Parsimonious Per-Flow Software Monitoring at Line Rate","authors":"Tianzhu Zhang, Leonardo Linguaglossa, Massimo Gallo, P. Giaccone, D. Rossi","doi":"10.23919/TMA.2018.8506565","DOIUrl":"https://doi.org/10.23919/TMA.2018.8506565","url":null,"abstract":"Testing experimental network devices requires deep performance analysis, which is usually performed with expensive, not flexible, hardware equipment. With the advent of highspeed packet I/O frameworks, general purpose equipments have narrowed the performance gap in respect of dedicated hardware and a variety of software-based solutions have emerged for handling traffic at very high speed. While the literature abounds with software traffic generators, existing monitoring solutions do not target worst-case scenarios (i.e., 64B packets at line rate) that are particularly relevant for stress-testing high-speed network functions, or occupy too many resources. In this paper we first analyse the design space for high-speed traffic monitoring that leads us to specific choices characterizing FlowMon-DPDK, a DPDK-based software traffic monitor that we release as an open source project. In a nutshell, FlowMon-DPDK provides tunable fine-grained statistics at both packet and flow levels. Experimental results demonstrate that our traffic monitor is able to provide per-flow statistics with 5-nines precision at high-speed (14.88 Mpps) using an exiguous amount of resources. Finally, we showcase FlowMon-DPDK usage by testing two open source prototypes for stateful flow-level end-host and in-network packet processing.","PeriodicalId":6607,"journal":{"name":"2018 Network Traffic Measurement and Analysis Conference (TMA)","volume":"159 s1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2018-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"91422758","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：