Ambari's 3.0 focuses and lays the ground work for Multi-everything architecture (multi-mpacks, multi-service instances, multi-component instances etc).
- Management Packs (MPacks) enables Ambari to remove the tight coupling with existing Hortonworks Data Platform (HDP) and Hortonworks Data Flow (HDF) stacks and enables bringing-in and manage any service into Ambari.
- Ambari 3.0 goal is to enable deployments of multiple MPack instances, where a MPack can have a dependency on another MPack and a given service can have service dependency on another service instance from within or outside the Mpack.
ambari 3.0, apache, hortonworks, hortonworks data flow (hdf, hortonworks data platform (hdp), management packs, mpacks, multi everything architecture, multiple component instance, multiple mpacks architecture, multiple service instance, stacks
The document discusses BGP zombie routes, which occur when an active routing table entry remains for a prefix that has been withdrawn by its origin network. This can cause issues like partial outages or routing loops. The document provides examples of real-world incidents involving zombie routes, including outages at Telia, Interoute, and CenturyLink/Level3. It also covers ways to detect and debug zombie routes, such as looking for routing loops or inconsistencies using traceroute. Mitigation strategies are also discussed.
This document summarizes a study on using Multipath TCP (MPTCP) to tolerate packet reordering and path heterogeneity in wireless networks. The study evaluated the performance of different MPTCP congestion controllers combined with various packet reordering recovery algorithms. The results showed that MPTCP with D-SACK or TCP-DOOR performed best in terms of throughput by increasing path utilization. D-SACK required less memory and was best for asymmetric paths, while TCP-DOOR was best for symmetric paths. In general, packet reordering solutions improved MPTCP performance significantly.
Application Behavior-Aware Flow Control in Network-on-ChipIvonne Liu
This document discusses application behavior-aware flow control in network-on-chip (NoC) architectures. The author proposes a proactive approach to congestion detection that predicts changes in global, end-to-end network traffic patterns of running applications. Traffic prediction is based on a table-driven predictor that uses application communication patterns. The flow control algorithm schedules packet injection to avoid predicted network congestion and improve throughput. Simulation results on SPLASH-2 benchmarks and synthetic traffic show significant performance improvements with negligible overhead.
Slides supporting the "Computer Networking: Principles, Protocols and Practice" ebook. The slides can be freely reused to teach an undergraduate computer networking class using the open-source ebook.
This document introduces GridSAM, a job management system that allows users to submit, monitor, and terminate jobs across different computational resource management systems. It provides an overview of GridSAM's architecture and standards compliance, describes how it can be installed and deployed, and outlines some future development plans. Instructions are given for accessing a demo instance of GridSAM for hands-on use in an integrated practical session.
Slides supporting the "Computer Networking: Principles, Protocols and Practice" ebook. The slides can be freely reused to teach an undergraduate computer networking class using the open-source ebook.
Ambari's 3.0 focuses and lays the ground work for Multi-everything architecture (multi-mpacks, multi-service instances, multi-component instances etc).
- Management Packs (MPacks) enables Ambari to remove the tight coupling with existing Hortonworks Data Platform (HDP) and Hortonworks Data Flow (HDF) stacks and enables bringing-in and manage any service into Ambari.
- Ambari 3.0 goal is to enable deployments of multiple MPack instances, where a MPack can have a dependency on another MPack and a given service can have service dependency on another service instance from within or outside the Mpack.
ambari 3.0, apache, hortonworks, hortonworks data flow (hdf, hortonworks data platform (hdp), management packs, mpacks, multi everything architecture, multiple component instance, multiple mpacks architecture, multiple service instance, stacks
The document discusses BGP zombie routes, which occur when an active routing table entry remains for a prefix that has been withdrawn by its origin network. This can cause issues like partial outages or routing loops. The document provides examples of real-world incidents involving zombie routes, including outages at Telia, Interoute, and CenturyLink/Level3. It also covers ways to detect and debug zombie routes, such as looking for routing loops or inconsistencies using traceroute. Mitigation strategies are also discussed.
This document summarizes a study on using Multipath TCP (MPTCP) to tolerate packet reordering and path heterogeneity in wireless networks. The study evaluated the performance of different MPTCP congestion controllers combined with various packet reordering recovery algorithms. The results showed that MPTCP with D-SACK or TCP-DOOR performed best in terms of throughput by increasing path utilization. D-SACK required less memory and was best for asymmetric paths, while TCP-DOOR was best for symmetric paths. In general, packet reordering solutions improved MPTCP performance significantly.
Application Behavior-Aware Flow Control in Network-on-ChipIvonne Liu
This document discusses application behavior-aware flow control in network-on-chip (NoC) architectures. The author proposes a proactive approach to congestion detection that predicts changes in global, end-to-end network traffic patterns of running applications. Traffic prediction is based on a table-driven predictor that uses application communication patterns. The flow control algorithm schedules packet injection to avoid predicted network congestion and improve throughput. Simulation results on SPLASH-2 benchmarks and synthetic traffic show significant performance improvements with negligible overhead.
Slides supporting the "Computer Networking: Principles, Protocols and Practice" ebook. The slides can be freely reused to teach an undergraduate computer networking class using the open-source ebook.
This document introduces GridSAM, a job management system that allows users to submit, monitor, and terminate jobs across different computational resource management systems. It provides an overview of GridSAM's architecture and standards compliance, describes how it can be installed and deployed, and outlines some future development plans. Instructions are given for accessing a demo instance of GridSAM for hands-on use in an integrated practical session.
Slides supporting the "Computer Networking: Principles, Protocols and Practice" ebook. The slides can be freely reused to teach an undergraduate computer networking class using the open-source ebook.
Using open source tools for network device dataplane testing.
Our experiences from redGuardian DDoS mitigation scrubber testing.
Presented at PLNOG 20 (2018).
Optimize Single Particle Orbital (SPO) Evaluations Based on B-splinesIntel® Software
Orbital representations that are based on B-splines are widely used in quantum Monte Carlo (QMC) simulations of solids, which historically take as much as 50 percent of the total runtime. Random access to a large four-dimensional array make it challenging to efficiently use caches and wide vector units in modern CPUs. So, we present node-level optimizations of B-spline evaluations on multicore and manycore shared memory processors.
To increase single instruction multiple data (SIMD) efficiency and bandwidth utilization, we first apply data layout transformation from an array of structures (AoS) to a structure of arrays (SoA). Then, by blocking SoA objects, we optimize cache reuse and get sustained throughput for a range of problem sizes. We implement efficient nested threading in B-spline orbital evaluation kernels, paving the way towards enabling strong scaling of QMC simulations, resulting with performance enhancements. Finally, we employ roofline performance analysis to model the impacts of our optimizations.
Cache "Less for More" in Information-centric Networks (CL4M)wkchai
W. K. Chai, D. He, I. Psaras and G. Pavlou "Cache "Less for More" in Information-Centric Networks (Extended Version)", Elsevier Computer Communications, Special Issue on Information-Centric Networking, vol. 36, no. 7, pp. 758-770, 1 April 2013, (DOI) 10.1016/j.comcom.2013.01.007.
W. K. Chai, D. He, I. Psaras, G. Pavlou, "Cache "Less for More" in Information-centric Networks", Proceedings of the 11th IFIP Networking, Prague, Czech Republic, 21-25 May 2012. BEST PAPER AWARD
LAWIN: a Latency-AWare InterNet Architecture for Latency Support on Best-Effo...Katsushi Kobayashi
Abstract—While strict latency restrictions are imposed on network applications, current best-effort Internet architecture entirely lacks this support. In this paper, we propose a “Latency AWare InterNet” (LAWIN) architecture that supports various la- tency requirements while retaining the best-effort service model. In the LAWIN architecture, applications specify their desired network latency limits, or deadlines, into all packets, and routers schedule these packets according to their deadlines. To this end, we propose two earliest-deadline-first (EDF)-based packet schedulers. The first imposes the same packet loss rate on all applications regardless of the latency specified by each, and provides rough flow-rate fairness. The second scheduler imposes a biased packet loss probability. The biased scheduler also provides an efficient latency and bandwidth trading mechanism for application settings, which motivates applications to set optimal latencies in order to improve efficiency.
This document discusses using Multipath TCP (MPTCP) to utilize multiple network paths simultaneously for increased bandwidth and reliability. It proposes using shadowsocks and redsocks to enable MPTCP for end users. Shadowsocks would encrypt traffic and redsocks would redirect connections through multiple paths, including the user's ADSL and fiber connections, via a gateway box running Ubuntu with MPTCP and Webmin for configuration. The gateway would split traffic across up to 6 possible paths to the destination server.
ES-HAS: An Edge- and SDN-Assisted Framework for HTTP Adaptive Video StreamingAlpen-Adria-Universität
Recently, HTTP Adaptive Streaming (HAS) has become the dominant video delivery technology over the Internet. In HAS, clients have full control over the media streaming and adaptation processes. Lack of coordination among the clients and lack of awareness of the network conditions may lead to sub-optimal user experience, and resource utilization in a pure client-based HAS adaptation scheme. Software-Defined Networking (SDN) has recently been considered to enhance the video streaming process. In this paper, we leverage the capability of SDN and Network Function Virtualization (NFV) to introduce an edge- and SDN-assisted video streaming framework called ES-HAS. We employ virtualized edge components to collect HAS clients’ requests and retrieve networking information in a time-slotted manner. These components then perform an optimization model in a time-slotted manner to efficiently serve clients’ requests by selecting an optimal cache server (with the shortest fetch time). In case of a cache miss, a client’s request is served (i) by an optimal replacement quality (only better quality levels with minimum deviation) from a cache server, or (ii) by the originally requested quality level from the origin server. This approach is validated through experiments on a large-scale testbed, and the performance of our framework is compared to pure client-based strategies and the SABR system [11]. Although SABR and ES-HAS show (almost) identical performance in the number of quality switches, ES-HAS outperforms SABR in terms of playback bitrate and the number of stalls by at least 70% and 40%, respectively.
NERSC is the production high-performance computing (HPC) center for the United States Department of Energy (DOE) Office of Science. The center supports over 6,000 users in 600 projects, using a variety of applications in materials science, chemistry, biology, astrophysics, high energy physics, climate science, fusion science, and more.
NERSC deployed the Cori system on over 9,000 Intel® Xeon Phi™ processors. This session describes the optimization strategy for porting codes that target traditional manycore architectures to the processors. We also discuss highlights and lessons learned from the optimization process on 20 applications associated with the NERSC Exascale Science Application Program (NESAP).
EPIQ'21: Days of Future Past: An Optimization-based Adaptive Bitrate Algorith...Minh Nguyen
HTTP Adaptive Streaming (HAS) has become a predominant technique for delivering videos in the Internet. Due to its adaptive behavior according to changing network conditions, it may result in video quality variations that negatively impact the Quality of Experience (QoE) of the user. In this paper, we propose Days of Future Past, an optimization- based Adaptive Bitrate (ABR) algorithm over HTTP/3. Days of Future Past takes advantage of an optimization model and HTTP/3 features, including (i) stream multiplexing and (ii) request cancellation. We design a Mixed Integer Linear Programming (MILP) model that determines the optimal video qualities of both the next segment to be requested and the segments currently located in the buffer. If better qualities for buffered segments are found, the client will send corresponding HTTP GET requests to retrieve them. Multiple segments (i.e., retransmitted segments) might be downloaded simultaneously to upgrade some buffered but not yet played segments to avoid quality decreases using the stream multiplexing feature of QUIC. HTTP/3’s request cancellation will be used in case retransmitted segments will arrive at the client after their playout time. The experimental results shows that our proposed method is able to improve the QoE by up to 33.9%.
IRJET- Modeling a New Startup Algorithm for TCP New RenoIRJET Journal
This document presents a new startup algorithm for TCP called the TCP SYN Loss (TSL) Startup Algorithm. The algorithm aims to make TCP more responsive for short-lived transactions by being more robust against packet losses during connection establishment. Specifically, the TSL algorithm reduces the congestion window and slow-start threshold by a maximum of 2 if a SYN or SYN-ACK packet is dropped, rather than resetting them to the standard TCP values of 1 and 2 maximum segment sizes. The document develops a stochastic model to analyze how the TSL algorithm impacts latency for short flows as a function of initial congestion window size, slow-start threshold, and link bandwidth-delay product.
Slides supporting the "Computer Networking: Principles, Protocols and Practice" ebook. The slides can be freely reused to teach an undergraduate computer networking class using the open-source ebook.
ContikiMAC : Radio Duty Cycling ProtocolSalah Amean
ContikiMAC is a radio duty cycling protocol that uses periodic wake-ups to listen for packet transmissions from neighbors. It keeps the radio on only when needed to receive packets, sending acknowledgments and retransmitting packets as necessary. The timing of transmissions and wake-ups is precise to enable efficient communication while minimizing energy usage. Phase awareness allows senders to transmit packets just before expected receiver wake-ups. ContikiMAC is implemented using real-time timers in Contiki to ensure stable and accurate timing.
Quality impact of scalable video coding tunneling for media aware content del...Alpen-Adria-Universität
The document discusses a study on using Scalable Video Coding (SVC) tunneling for media-aware content delivery. SVC can save bandwidth for multicast content delivery but many devices do not support SVC natively. The study proposes using SVC tunneling only for delivery, with transcoding at ingress and egress points. The study tests transcoding MPEG-2 video to SVC and back to MPEG-2, finding a peak signal-to-noise ratio drop of 2.1 dB and 43% lower bandwidth requirements compared to MPEG-2 simulcast. Future work could analyze different video formats and implementations of SVC.
HBaseCon 2013: Scalable Network Designs for Apache HBaseCloudera, Inc.
This document discusses scalable network designs and how modern networks can help applications. It begins with a brief history of network software and describes how switches now run Linux. Typical network designs are presented starting small and scaling up through multiple racks and core switches. The benefits of layer 3 designs, jumbo frames, and deep buffers to prevent packet loss are covered. Finally, it discusses how the network can help applications by detecting server failures, redirecting traffic, and enabling fast failover through features only possible by the switch running Linux.
This document summarizes key topics related to IPv6 and routing in IP networks. It discusses IPv6 addressing architecture, including unicast addresses, link-local addresses, and multicast addresses. It also covers IPv6 packet format, extension headers, fragmentation, and ICMPv6. The document then discusses routing within IP networks, including IPv6 subnets, routing organization with autonomous systems, and interdomain routing protocols.
This document evaluates and compares the performance of seven high-speed TCP congestion control protocols: Bic TCP, Cubic TCP, Hamilton TCP, HighSpeed TCP, Illinois TCP, Scalable TCP and YeAH TCP. It first provides background on the need for high-speed congestion control as internet speeds have increased. It then summarizes the algorithms and mechanisms of each protocol. The document aims to simulate and compare the performance of these protocols with multiple flows, in order to determine the best approach for high-speed networks.
This document discusses network emulation using tc. It begins with an agenda that covers why emulation is useful, what aspects of a network can be emulated, how tc works, how to do emulation with tc, a comparison of tc to Nistnet and WANem, and references for further information. It then goes into detail on each agenda item, providing explanations of concepts like qdiscs, classes, filters, and examples of using tc for tasks like bandwidth emulation, delay/jitter emulation, and loss emulation. The key advantages and limitations of Nistnet and WANem are also outlined.
Labmeeting - 20150831 - Overhead and Performance of Low Latency Live Streamin...Syuan Wang
This document summarizes research into reducing latency for live video streaming using MPEG-DASH. It introduces MPEG-DASH and how using HTTP chunked transfer encoding and Gradual Decoding Refresh encoding can help lower latency compared to basic DASH. The paper describes experiments conducted to generate and distribute live content using these techniques and evaluate latency, finding they were able to achieve latency as low as 240ms.
This is a presentation by Prof. Anne Elster at the International Workshop on Open Source Supercomputing held in conjunction with the 2017 ISC High Performance Computing Conference.
This document summarizes the CyCLaDEs approach for caching triple pattern fragments (TPF) in a decentralized manner across Linked Data clients. It proposes using two overlay networks - a Random Peer Sampling network to ensure connectivity between all clients, and a Cluster Overlay Network to connect similar clients based on query histories. Each client builds a profile based on recently used query predicates and finds similar neighbors to cache TPF locally for future queries before querying the central LDF server.
Using open source tools for network device dataplane testing.
Our experiences from redGuardian DDoS mitigation scrubber testing.
Presented at PLNOG 20 (2018).
Optimize Single Particle Orbital (SPO) Evaluations Based on B-splinesIntel® Software
Orbital representations that are based on B-splines are widely used in quantum Monte Carlo (QMC) simulations of solids, which historically take as much as 50 percent of the total runtime. Random access to a large four-dimensional array make it challenging to efficiently use caches and wide vector units in modern CPUs. So, we present node-level optimizations of B-spline evaluations on multicore and manycore shared memory processors.
To increase single instruction multiple data (SIMD) efficiency and bandwidth utilization, we first apply data layout transformation from an array of structures (AoS) to a structure of arrays (SoA). Then, by blocking SoA objects, we optimize cache reuse and get sustained throughput for a range of problem sizes. We implement efficient nested threading in B-spline orbital evaluation kernels, paving the way towards enabling strong scaling of QMC simulations, resulting with performance enhancements. Finally, we employ roofline performance analysis to model the impacts of our optimizations.
Cache "Less for More" in Information-centric Networks (CL4M)wkchai
W. K. Chai, D. He, I. Psaras and G. Pavlou "Cache "Less for More" in Information-Centric Networks (Extended Version)", Elsevier Computer Communications, Special Issue on Information-Centric Networking, vol. 36, no. 7, pp. 758-770, 1 April 2013, (DOI) 10.1016/j.comcom.2013.01.007.
W. K. Chai, D. He, I. Psaras, G. Pavlou, "Cache "Less for More" in Information-centric Networks", Proceedings of the 11th IFIP Networking, Prague, Czech Republic, 21-25 May 2012. BEST PAPER AWARD
LAWIN: a Latency-AWare InterNet Architecture for Latency Support on Best-Effo...Katsushi Kobayashi
Abstract—While strict latency restrictions are imposed on network applications, current best-effort Internet architecture entirely lacks this support. In this paper, we propose a “Latency AWare InterNet” (LAWIN) architecture that supports various la- tency requirements while retaining the best-effort service model. In the LAWIN architecture, applications specify their desired network latency limits, or deadlines, into all packets, and routers schedule these packets according to their deadlines. To this end, we propose two earliest-deadline-first (EDF)-based packet schedulers. The first imposes the same packet loss rate on all applications regardless of the latency specified by each, and provides rough flow-rate fairness. The second scheduler imposes a biased packet loss probability. The biased scheduler also provides an efficient latency and bandwidth trading mechanism for application settings, which motivates applications to set optimal latencies in order to improve efficiency.
This document discusses using Multipath TCP (MPTCP) to utilize multiple network paths simultaneously for increased bandwidth and reliability. It proposes using shadowsocks and redsocks to enable MPTCP for end users. Shadowsocks would encrypt traffic and redsocks would redirect connections through multiple paths, including the user's ADSL and fiber connections, via a gateway box running Ubuntu with MPTCP and Webmin for configuration. The gateway would split traffic across up to 6 possible paths to the destination server.
ES-HAS: An Edge- and SDN-Assisted Framework for HTTP Adaptive Video StreamingAlpen-Adria-Universität
Recently, HTTP Adaptive Streaming (HAS) has become the dominant video delivery technology over the Internet. In HAS, clients have full control over the media streaming and adaptation processes. Lack of coordination among the clients and lack of awareness of the network conditions may lead to sub-optimal user experience, and resource utilization in a pure client-based HAS adaptation scheme. Software-Defined Networking (SDN) has recently been considered to enhance the video streaming process. In this paper, we leverage the capability of SDN and Network Function Virtualization (NFV) to introduce an edge- and SDN-assisted video streaming framework called ES-HAS. We employ virtualized edge components to collect HAS clients’ requests and retrieve networking information in a time-slotted manner. These components then perform an optimization model in a time-slotted manner to efficiently serve clients’ requests by selecting an optimal cache server (with the shortest fetch time). In case of a cache miss, a client’s request is served (i) by an optimal replacement quality (only better quality levels with minimum deviation) from a cache server, or (ii) by the originally requested quality level from the origin server. This approach is validated through experiments on a large-scale testbed, and the performance of our framework is compared to pure client-based strategies and the SABR system [11]. Although SABR and ES-HAS show (almost) identical performance in the number of quality switches, ES-HAS outperforms SABR in terms of playback bitrate and the number of stalls by at least 70% and 40%, respectively.
NERSC is the production high-performance computing (HPC) center for the United States Department of Energy (DOE) Office of Science. The center supports over 6,000 users in 600 projects, using a variety of applications in materials science, chemistry, biology, astrophysics, high energy physics, climate science, fusion science, and more.
NERSC deployed the Cori system on over 9,000 Intel® Xeon Phi™ processors. This session describes the optimization strategy for porting codes that target traditional manycore architectures to the processors. We also discuss highlights and lessons learned from the optimization process on 20 applications associated with the NERSC Exascale Science Application Program (NESAP).
EPIQ'21: Days of Future Past: An Optimization-based Adaptive Bitrate Algorith...Minh Nguyen
HTTP Adaptive Streaming (HAS) has become a predominant technique for delivering videos in the Internet. Due to its adaptive behavior according to changing network conditions, it may result in video quality variations that negatively impact the Quality of Experience (QoE) of the user. In this paper, we propose Days of Future Past, an optimization- based Adaptive Bitrate (ABR) algorithm over HTTP/3. Days of Future Past takes advantage of an optimization model and HTTP/3 features, including (i) stream multiplexing and (ii) request cancellation. We design a Mixed Integer Linear Programming (MILP) model that determines the optimal video qualities of both the next segment to be requested and the segments currently located in the buffer. If better qualities for buffered segments are found, the client will send corresponding HTTP GET requests to retrieve them. Multiple segments (i.e., retransmitted segments) might be downloaded simultaneously to upgrade some buffered but not yet played segments to avoid quality decreases using the stream multiplexing feature of QUIC. HTTP/3’s request cancellation will be used in case retransmitted segments will arrive at the client after their playout time. The experimental results shows that our proposed method is able to improve the QoE by up to 33.9%.
IRJET- Modeling a New Startup Algorithm for TCP New RenoIRJET Journal
This document presents a new startup algorithm for TCP called the TCP SYN Loss (TSL) Startup Algorithm. The algorithm aims to make TCP more responsive for short-lived transactions by being more robust against packet losses during connection establishment. Specifically, the TSL algorithm reduces the congestion window and slow-start threshold by a maximum of 2 if a SYN or SYN-ACK packet is dropped, rather than resetting them to the standard TCP values of 1 and 2 maximum segment sizes. The document develops a stochastic model to analyze how the TSL algorithm impacts latency for short flows as a function of initial congestion window size, slow-start threshold, and link bandwidth-delay product.
Slides supporting the "Computer Networking: Principles, Protocols and Practice" ebook. The slides can be freely reused to teach an undergraduate computer networking class using the open-source ebook.
ContikiMAC : Radio Duty Cycling ProtocolSalah Amean
ContikiMAC is a radio duty cycling protocol that uses periodic wake-ups to listen for packet transmissions from neighbors. It keeps the radio on only when needed to receive packets, sending acknowledgments and retransmitting packets as necessary. The timing of transmissions and wake-ups is precise to enable efficient communication while minimizing energy usage. Phase awareness allows senders to transmit packets just before expected receiver wake-ups. ContikiMAC is implemented using real-time timers in Contiki to ensure stable and accurate timing.
Quality impact of scalable video coding tunneling for media aware content del...Alpen-Adria-Universität
The document discusses a study on using Scalable Video Coding (SVC) tunneling for media-aware content delivery. SVC can save bandwidth for multicast content delivery but many devices do not support SVC natively. The study proposes using SVC tunneling only for delivery, with transcoding at ingress and egress points. The study tests transcoding MPEG-2 video to SVC and back to MPEG-2, finding a peak signal-to-noise ratio drop of 2.1 dB and 43% lower bandwidth requirements compared to MPEG-2 simulcast. Future work could analyze different video formats and implementations of SVC.
HBaseCon 2013: Scalable Network Designs for Apache HBaseCloudera, Inc.
This document discusses scalable network designs and how modern networks can help applications. It begins with a brief history of network software and describes how switches now run Linux. Typical network designs are presented starting small and scaling up through multiple racks and core switches. The benefits of layer 3 designs, jumbo frames, and deep buffers to prevent packet loss are covered. Finally, it discusses how the network can help applications by detecting server failures, redirecting traffic, and enabling fast failover through features only possible by the switch running Linux.
This document summarizes key topics related to IPv6 and routing in IP networks. It discusses IPv6 addressing architecture, including unicast addresses, link-local addresses, and multicast addresses. It also covers IPv6 packet format, extension headers, fragmentation, and ICMPv6. The document then discusses routing within IP networks, including IPv6 subnets, routing organization with autonomous systems, and interdomain routing protocols.
This document evaluates and compares the performance of seven high-speed TCP congestion control protocols: Bic TCP, Cubic TCP, Hamilton TCP, HighSpeed TCP, Illinois TCP, Scalable TCP and YeAH TCP. It first provides background on the need for high-speed congestion control as internet speeds have increased. It then summarizes the algorithms and mechanisms of each protocol. The document aims to simulate and compare the performance of these protocols with multiple flows, in order to determine the best approach for high-speed networks.
This document discusses network emulation using tc. It begins with an agenda that covers why emulation is useful, what aspects of a network can be emulated, how tc works, how to do emulation with tc, a comparison of tc to Nistnet and WANem, and references for further information. It then goes into detail on each agenda item, providing explanations of concepts like qdiscs, classes, filters, and examples of using tc for tasks like bandwidth emulation, delay/jitter emulation, and loss emulation. The key advantages and limitations of Nistnet and WANem are also outlined.
Labmeeting - 20150831 - Overhead and Performance of Low Latency Live Streamin...Syuan Wang
This document summarizes research into reducing latency for live video streaming using MPEG-DASH. It introduces MPEG-DASH and how using HTTP chunked transfer encoding and Gradual Decoding Refresh encoding can help lower latency compared to basic DASH. The paper describes experiments conducted to generate and distribute live content using these techniques and evaluate latency, finding they were able to achieve latency as low as 240ms.
This is a presentation by Prof. Anne Elster at the International Workshop on Open Source Supercomputing held in conjunction with the 2017 ISC High Performance Computing Conference.
This document summarizes the CyCLaDEs approach for caching triple pattern fragments (TPF) in a decentralized manner across Linked Data clients. It proposes using two overlay networks - a Random Peer Sampling network to ensure connectivity between all clients, and a Cluster Overlay Network to connect similar clients based on query histories. Each client builds a profile based on recently used query predicates and finds similar neighbors to cache TPF locally for future queries before querying the central LDF server.
Packet Optical SDN Field Trial for Multi-Layer Network OptimizationADVA
Jim Theodoras’ presentation, broadcast live from 2016 Internet2 Global Summit in Chicago, covered the work that ADVA Optical Networking and Juniper Networks have accomplished towards interoperability of SDN controllers and outlined the advantages of joint SDN management of router and transport resources.
ANSYS RedHawk-CPA: New Paradigm for Faster Chip-Package ConvergenceAnsys
Due to the increasing size of SoCs and the variation in the switching current and parasitic profile across the chip, the individual connections between the SoC and the package at the C4 bump level need to be as granular as possible to provide resolution to the power analysis. To see the benefit from changes made to the chip and/or package in a timely manner requires that both layouts can be modified and modeled in an integrated manner. This presentation introduces RedHawk-CPA, a new feature which allows the inclusion of both chip and package layouts for a unified DC, transient and AC power integrity analysis. It will demonstrate how RedHawk-CPA can improve the level of accuracy as well as reduce the time to power closure. Learn more on our website: https://bit.ly/1ssSGM0
Paolo Lucente, author of the software pmacct.
Traffic matrices can greatly benefit key IP Service Provider activities like capacity planning, traffic engineering, better understand their traffic patterns and take meaningful peering decisions. This talk wants to present a way to build traffic matrices using telemetry data and BGP, leveraging along the way some case-studies with a technical cut. pmacct (http://www.pmacct.net) is a commonly used, free, open-source IPv4/IPv6a ccounting package which integrates a NetFlow/sFlow and a multi-RIB BGP collector in a single piece of software and is authored by the presenter.
A Grid Computing Platform where Communication Function is in Balance with Computation and Storage.
Lambda Data Grid Service architecture interacts with Cyber-infrastructure, and overcomes data limitations efficiently & effectively by:
treating the “network” as a primary resource just like “storage” and “computation”
treating the “network” as a “scheduled resource”
relying upon a massive, dynamic transport infrastructure: Dynamic Optical Network
User-friendly Metaworkflows in Quantum ChemistrySandra Gesing
The document describes user-friendly metaworkflows for quantum chemistry simulations developed by the Molecular Simulation Grid (MoSGrid) project. The MoSGrid portal provides an intuitive interface to enable non-experts to easily access high-performance computing resources and run quantum chemistry workflows and applications. The portal features quantum chemistry portlets that allow users to generate input files, submit simulation jobs to grids, and analyze output through visualizations. Examples of supported workflows include potential energy surface scans, high-throughput processing of crystal structure files, and locating transition states. The goal is to make advanced molecular simulations accessible to scientists without extensive computing expertise.
Recently, massively parallel processing relational database systems (MPPDBs) have gained much momentum in the big data analytic market. With the advent of hosted cloud computing, we envision that the offering of MPPDB-as-a-Service (MPPDBaaS) will become attractive for companies having analytical tasks on only hundreds gigabytes to some ten terabytes of data because they can enjoy high-end parallel analytics at a cheap cost. This paper presents Thrifty, a prototype implementation of MPPDB-as-a-service. The
major research issue is how to achieve a lower total cost of ownership by consolidating thousands of MPPDB tenants on to a shared hardware infrastructure, with a performance SLA that guarantees the tenants can obtain the query results as if they are executing their queries on dedicated machines. Thrifty achieves the goal by using a tenant-driven design that includes (1) a cluster design that carefully arranges the nodes in the cluster into groups and creates an MPPDB for each group of nodes, (2) a tenant placement that assigns each tenant to several MPPDBs (for high availability service through replication), and (3) a query routing algorithm that routes
a tenant’s query to the proper MPPDB at run-time. Experiments show that in a MPPDBaaS with 5000 tenants, where each tenant requests 2 to 32 nodes MPPDB to query against 200GB to 3.2TB of data, Thrifty can serve all the tenants with a 99.9% performance SLA guarantee and a high availability replication factor of 3, using only 18.7% of the nodes requested by the tenants.
My talk at the Winter School on Big Data in Tarragona, Spain.
Abstract: We have made much progress over the past decade toward harnessing the collective power of IT resources distributed across the globe. In high-energy physics, astronomy, and climate, thousands work daily within virtual computing systems with global scope. But we now face a far greater challenge: Exploding data volumes and powerful simulation tools mean that many more--ultimately most?--researchers will soon require capabilities not so different from those used by such big-science teams. How are we to meet these needs? Must every lab be filled with computers and every researcher become an IT specialist? Perhaps the solution is rather to move research IT out of the lab entirely: to leverage the “cloud” (whether private or public) to achieve economies of scale and reduce cognitive load. I explore the past, current, and potential future of large-scale outsourcing and automation for science, and suggest opportunities and challenges for today’s researchers.
Software Defined Optical Networks - Mayur ChannegowdaCPqD
This document discusses software defined optical networks using SDN. Key points include:
- SDN and OpenFlow can decouple the data and control planes in optical networks for automated provisioning and unified control.
- There are challenges in applying SDN to optical networks including switching constraints, physical impairments, multi-domain/multi-technology operation, and network virtualization.
- OpenFlow extensions are needed to abstract optical network elements and account for characteristics like flexible grid networks, impairment awareness, and multi-dimensional resource allocation.
- Proof-of-concept demonstrations have shown the potential for media-aware SDN, packet and optical convergence, and virtualization across multiple domains.
Software Defined Optical Networks - Mayur ChannegowdaCPqD
This document discusses software defined optical networks using SDN. Key points include:
- SDN and OpenFlow can decouple the data and control planes in optical networks for automated provisioning and unified control.
- There are challenges in applying SDN to optical networks including switching constraints, physical impairments, multi-domain/multi-technology operation, and network virtualization.
- OpenFlow extensions are needed to abstract optical network elements and account for characteristics like flexible grid networks, impairment awareness, and multi-dimensional resource allocation.
- Proof-of-concept demonstrations have shown the potential for media-aware SDN, packet and optical convergence, and virtualization across multiple domains.
The document discusses how application architects traditionally focused on solving IO bottlenecks in servers by offloading processing to intelligent network interface cards. With modern distributed applications spanning thousands of servers, application architects now must consider network topology, segmentation, and control plane protocols to optimize latency and bandwidth. The rise of virtualization and cloud computing has changed traffic patterns in datacenters from north-south traffic to dominant east-west traffic between servers. This requires new datacenter fabric designs beyond the traditional three-tiered topology.
Stream Data Processing at Big Data Landscape by Oleksandr Fedirko GlobalLogic Ukraine
This document provides an overview of stream data processing and common stream processing tools. It discusses streaming basics like stateful and stateless operations. It also covers microbatch vs realtime streaming and compositional vs declarative stream processing engines. Typical stream processing architectures and use cases are presented. Main considerations for projects using stream processing are outlined. An overview of popular stream processing tools like Apache Spark, Storm, Flink, and cloud services from AWS, GCP and Azure is provided. The document concludes with a case study example and questions for discussion.
The Challenges of SDN/OpenFlow in an Operational and Large-scale NetworkOpen Networking Summits
Jun Bi
Professor & Director
Tsinghua University
Outline
• Intra-AS (campus level) IPv6 source address validation using OpenFlow (with extension)
– Good for introducing new IP services to network
• Planning next step if we run SDN as a common infrastructure for new services and architectures
– Some personal viewpoints and thoughts on design challenges
– Forwarding abstraction for Post-IP architectures
– Control abstraction for scalable NOS and programmable/manageable virtualization platform
– Inter-AS policies negotiation abstraction
ONS2015: http://bit.ly/ons2015sd
ONS Inspire! Webinars: http://bit.ly/oiw-sd
Watch the talk (video) on ONS Content Archives: http://bit.ly/ons-archives-sd
Many thanks to Nick McKeown (Stanford), Jennifer Rexford (Princeton), Scott Shenker (Berkeley), Nick Feamster (Princeton), Li Erran Li (Columbia), Yashar Ganjali (Toronto)
The document describes the architecture of the Wemlin Hub, a system that provides public transport information. It has a modular design with components that plug into a pipeline. The pipeline uses immutable objects and memoization to improve performance. It also employs a blue-green deployment strategy to enable zero downtime upgrades. The system currently processes data for over 17 agencies in Switzerland and serves 4 customers.
A Platform for Large-Scale Grid Data Service on Dynamic High-Performance Netw...Tal Lavian Ph.D.
Dynamic High-Performance Networks :
Support data-intensive Grid applications
Gives adequate and uncontested bandwidth to an application’s burst
Employs circuit-switching of large flows of data to avoid overheads in breaking flows into small packets and delays routing
Is capable of automatic end-to-end path provisioning
Is capable of automatic wavelength switching
Provides a set of protocols for managing dynamically provisioned wavelengths
DWDM-RAM :
Encapsulates “optical network resources” into a service framework to support dynamically provisioned and advanced data-intensive transport services
Offers network resources as Grid services for Grid computing
Allows cooperation of distributed resources
Provides a generalized framework for high performance applications over next generation networks, not necessary optical end-to-end
Yields good overall utilization of network resources
The document describes GARUDA, a protocol for achieving reliable downstream communication in wireless sensor networks. GARUDA uses a two-stage negative acknowledgment approach and constructs a virtual infrastructure of core nodes to minimize retransmissions and ensure reliable message delivery with small message sizes. Core nodes are selected using an instantaneous construction based on the first transmitted packet to provide efficient and fair core selection.
The document provides an overview of routing protocols in wireless sensor networks. It discusses several categories of routing protocols including data-centric, hierarchical, and location-based. For hierarchical routing protocols, it summarizes LEACH, PEGASIS, HEED, P-LEACH, H-LEACH, and other variants that aim to improve energy efficiency. It provides brief descriptions of how each protocol operates and highlights drawbacks. The document also summarizes several data-centric routing protocols including Directed Diffusion, Rumor Routing, and their limitations.
Similar to Consistent Resource Scheduling and QoS management (20)
Магистерская программа «Распределённые системы и компьютерные сети»ARCCN
Доклад Смелянского Руслана Леонидовича, чл.-корр. РАН, профессор, д.ф.-м-.н., МГУ им. М.В.Ломоносова, факультет ВМК, Кафедра АСВК, Лаборатория Вычислительных Комплексов, на ежегодном заседании участников Консорциума университетов по SDN технологиям, май 2017
Основные направления развития ФГБОУ ВО «РГРТУ» в области программно-конфигури...ARCCN
Доклад Перепелкина Дмитрия, к.т.н., доцент кафедры САПР ВС Рязанский государственный радиотехнический университет, Центр инновационных сетевых и облачных технологий, на ежегодном заседании Участников Косорциума университетов по SDN технологиям, май 2017 года
Методика стратегического управления развитием SDN&NFV-сети оператора связи и ...ARCCN
Доклад Александра Герасимова и Евгения Лашманова, Московский государственный технический университет радиотехники, электроники и автоматики (МГТУ МИРЭА), на ежегодном собрании участников Консорциума по SDN технологиям, мая 2017 года
The document summarizes the MC2E (MetaCloud Computing Environment) project, which aims to create a unified cloud computing platform that interconnects different data centers and heterogeneous computing resources. It describes the main components of the MC2E platform, including a meta-orchestrator, orchestrators, service managers, and a web portal. It also outlines the service lifecycle management process and discusses cooperation with international partners to further develop the MC2E platform and its capabilities.
Возможности импортозамещения коммутационного оборудования в сетях нового пок...ARCCN
Доклад Вячеслава Васина, директор департамента сетевых решений ЦПИКС, на конференции "Локализация производства и импортозамещение коммуникационного и радиоэлектронного оборудования, приборов и устройств для ИКТ отрасли России", 15 февраля 2017 года.
A Perspective on the Future of Computer ArchitectureARCCN
The document discusses the challenges with today's superscalar computer architectures and proposes ideas for future computer architectures. It summarizes the author's experience building real computers in the past, including some of the first implementations of out-of-order execution and VLIW. It then outlines drawbacks of the superscalar paradigm and discusses moving towards an algorithmically oriented post-superscalar era. The author proposes designing an unconstrained "best possible" new architecture first before adding constraints for compatibility.
The document describes several radical steps in computer architecture that were implemented by the author's team before the industry, including carry save arithmetic in 1954 and the Elbrus computer architecture in 1978. It discusses drawbacks of current superscalar architectures and outlines the principles of a proposed "best possible computer system" with support for high-level programming, fine-grained parallelism, dynamic data types, and full security through capability features. Key aspects are a new universal programming language, a compiler with full algorithm and hardware information, and hardware designed for optimization rather than artificial constraints.
Доклад Виталия Антоненко (ЦПИКС) на семинаре Консорциума университетов по изучению и развитию передовых технологий в сфере компьютерных сетей. 20 октября 2016 года
Типовые сервисы региональной сети передачи данныхARCCN
Доклад Вячеслава Васина (ЦПИКС) на семинаре Консорциума университетов по изучению и развитию передовых технологий в сфере компьютерных сетей. 20 октября 2016 года
Разработка OpenFlow-коммутатора на базе сетевого процессора EZchipARCCN
Доклад Васина Вячеслава (ЦПИКС) на семинаре Консорциума университетов по изучению и развитию передовых технологий в сфере компьютерных сетей. 20 октября 2016 года
Исследования SDN в Оренбургском государственном университете: сетевая безопас...ARCCN
Доклад Шухмана А.Е. (ОГУ) на семинаре Консорциума университетов по изучению и развитию передовых технологий в сфере компьютерных сетей. 20 октября 2016 года
Цели и задачи МИЭТ, как участника Консорциума на примере кафедры "Телекоммуни...ARCCN
Доклад Бахтина А.А. (МИЭТ) на семинаре Консорциума университетов по изучению и развитию передовых технологий в сфере компьютерных сетей. 20 октября 2016 года
Доклад Садова О.Л. (ИТМО) на семинаре Консорциума университетов по изучению и развитию передовых технологий в сфере компьютерных сетей. 20 октября 2016 года
Доклад Смелянского Р.Л. на семинаре Консорциума университетов по изучению и развитию передовых технологий в сфере компьютерных сетей. 20 октября 2016 года
Gen Z and the marketplaces - let's translate their needsLaura Szabó
The product workshop focused on exploring the requirements of Generation Z in relation to marketplace dynamics. We delved into their specific needs, examined the specifics in their shopping preferences, and analyzed their preferred methods for accessing information and making purchases within a marketplace. Through the study of real-life cases , we tried to gain valuable insights into enhancing the marketplace experience for Generation Z.
The workshop was held on the DMA Conference in Vienna June 2024.
Ready to Unlock the Power of Blockchain!Toptal Tech
Imagine a world where data flows freely, yet remains secure. A world where trust is built into the fabric of every transaction. This is the promise of blockchain, a revolutionary technology poised to reshape our digital landscape.
Toptal Tech is at the forefront of this innovation, connecting you with the brightest minds in blockchain development. Together, we can unlock the potential of this transformative technology, building a future of transparency, security, and endless possibilities.
Discover the benefits of outsourcing SEO to Indiadavidjhones387
"Discover the benefits of outsourcing SEO to India! From cost-effective services and expert professionals to round-the-clock work advantages, learn how your business can achieve digital success with Indian SEO solutions.
1. Applied Research Center
for Computer Networks
Moscow State University
Ruslan Smelyanskiy
ARCCN Director, Professor at Moscow State University
Consistent Resource Scheduling and
QoS management
2. Content
Brief ARCCN introduction and main
research direction overview
Two research problems examples:
The Consistent Resource Scheduling in DC
FDM-TCP: Flow DeMultiplexing TCP for QoS
management
Huawei Shenzhen R.Smelyanskiy 226.05.2015
3. ARCCN
Research
Ideas
Education
Adoption
Cooperation
Development
Integration
Applied Research Center for Computer Networks (ARCCN) is a Russian
non-profit organization initially funded by Skolkovo intended to:
world-class competency center for Computer Networking in Russia;
collaboration between Russian and international research, scientific,
educational and commercial experts and institutions;
demonstrate innovations in Networking for National Industry;
promote commercialization of a new networking technologies and
services.
ARCCN is the only R&D competency center for SDN&NFV
in Russia today
Partners: Skolkovo, The Ministry of Education and Science of Russia,
Rostelecom, Rostech, Sberbank, EMC, Intel, etc.
Huawei Shenzhen R.Smelyanskiy 326.05.2015
5. Key R&D directions:
Distributed SDN Controller and Applications for it;
Safety and Security in SDN;
QoS management in SDN;
NFV platform and VN functions;
Architecture for Open Flow-switch;
WAN operation prototyping and analysis;
Self Organizing Cloud for Carriers (Cloud Conductor);
Data Center Resource Management;
Forwarding policy validation and troubleshooting automation;
Software for integration with legacy transport on the Internet.
Huawei Shenzhen R.Smelyanskiy 526.05.2015
8. Data center resources model
Physical resource model: ,where
• Р - set of computational nodes:
– vh(p) - number of CPU cores on node p
– qh(p) - amount of RAM on node p
• М - set of data storages:
– uh(m) – size of data storage m
– type(m) - m data storage type
• К - set of network switches:
– - bandwidth of the switch k
• L - set of network channels:
– rh(l) – bandwidth of the channel l
),( LKMPH
( )h k
Huawei Shenzhen R.Smelyanskiy 826.05.2015
9. Tenant model
Tenant model:
• W - set of virtual machines:
– v(w) – requested number of cores for VM w
– q(w) – requested amount of RAM for VM w
• S - set of storages:
– u(s) – size of storage s
– type(s) – s storage type
• E - set of virtual channels:
– r(e) – bandwidth of the virtual channel e
Tenant types:
• Loosely coupled:
• Tightly coupled:
),( ESWG
)]),}{,}[({ 11 ESW K
jj
N
ii
)}]}{,}{{(),}{,}[({ 1
'
1
'
11
K
jj
N
ii
K
jj
N
ii SSWWESW
Huawei Shenzhen R.Smelyanskiy 926.05.2015
10. Tenant Mapping on DC Resources
where
Mapping A is correct if the following constraints are fulfilled:
a)
b)
c)
d)
)()(),()( pqhwqpvhwv
pp WwWw
: { , , { , }},A G H W P S M E K L
)()( lrher
lEe
)()( kher
kEe
),()( muhsu
mSs
)()(: mtypestypeSs m
, , ; , , , W S E G P M K L H
Huawei Shenzhen R.Smelyanskiy 1026.05.2015
11. Replication and migration
Replication:
• Duplicates storage m to m’
• Channel to support consistency between m and m’:
• VM connects to m’
Migration:
• Moves virtual machine or storage to another physical resource
'
: ii AAR
( ', , , , , , ); ; ; ',
1 1 1
m l k k l m k K l L m m M
n n i i
Huawei Shenzhen R.Smelyanskiy 11
'
: ii AAT
26.05.2015
12. Tenant mapping problem
The problem:
• For every tenant Gi from set of tenants Z = {Gi} and
given DC resources Hres define correct mapping Ai
that set A = {Ai} covers the maximum number of
tenants from Z
Huawei Shenzhen R.Smelyanskiy 1226.05.2015
13. Residual graph
Renew Z – 𝒁 → 𝒁′
Residual graph Hres:
• The following functions are recalculated: vh(p), qh(p), rh(l), τh(k), uh(m)
pWw
res wvpvhpvh )()()( ( ) ( ) ( )
p
res
w W
qh p qh p q w
( ) ( ) ( )
l
res
e E
rh l rh l r e
( ) ( ) ( )
k
res
e E
h k h k r e
( ) ( ) ( )
m
res
s S
uh m uh m u s
Huawei Shenzhen R.Smelyanskiy 1326.05.2015
14. Scheduling Round
The scheduling algorithm runs at the begging of the
scheduling rounds
On a scheduling round:
– Add new tenants to Z
– Delete the tenants with TTL=0
– Recalculate Residual graph as a new H
Huawei Shenzhen R.Smelyanskiy 1426.05.2015
15. Scheduling algorithms
• A1: greedy and limited exhaustive search strategies
– maps tenants as a whole (minimal common subgraph
isomorphism algorithm)
– suitable when critical resource is a data center physical
network
• A2: greedy and limited exhaustive search strategies
– constructs mappings element by element (bin-packing
algorithm)
– suitable when critical resources are computational nodes or
data storages
• A3: Ant colony algorithm:
– universal, but greater computational complexity
Huawei Shenzhen R.Smelyanskiy 1526.05.2015
16. Initial data for test
Test # DC model (graph H) Tenants models (graph G)
Test 1
450 computational nodes with
21000MB RAM and 16 cores
per node
450 data storages with
21000GB disk space per
storage
1350 virtual machines and 1350
storages total in 135 tenants.
Test 1 – loosely coupled tenants
Test 2 – tightly coupled tenants,
network load = 70%
Test 2
Test 3
500 computational nodes with
10000MB RAM and 16 cores
per node
500 data storages with
10000GB disk space per
storage
600 virtual machines and 600
storages per 100 tenants.
Test 3 – loosely coupled tenants
Test 4 – tightly coupled tenants,
network load = 70%
Test 4
Fattree topology with 5 TOR switches, 10 aggregation switches, 40 edge
switches
Huawei Shenzhen R.Smelyanskiy
1626.05.2015
17. Comparison with OpenStack
scheduling algorithms
Test #
Open stack CRM
FF RF
Test 1 50% 87.5% 100%
Test 2 0% 1% 100%
Test 3 100% 70% 100%
Test 4 0% 0% 100%
- FF – First Fit
- RF – Random Fit
- CRM – Consistent Resource Mapping
Huawei Shenzhen R.Smelyanskiy 1726.05.2015
18. Conclusions
Open Stack algorithms are:
– Ineffective for some cases of loosely coupled tenants
– Not applicable for tightly coupled tenants
Consistent Resource Mapping algorithms:
– Effective as for loosely as for toughly coupled tenants
– Suitable for IaaS with SLAs
Huawei Shenzhen R.Smelyanskiy 1926.05.2015
19. FDM-SDN: SDN with Flow
DeМultiplexing TCP
is a new way to manage the quality of service
20. Hosts do affect connection quality!
• TCP congestion avoidance algorithms:
– Goal: get connection with maximal bandwidth
– Strategy: cut-and-try to detect the maximum currently
available amount of resources to utilize all of them
– Primary heuristic: AIMD (pessimistic)
– Recovering action: decries CWND
– Parameter: congestion window size (CWND), timeout
– Primary modes: slow start and congestion avoidance
– Triggering criteria: duplicate ACKs&timeouts, threshold
• There is no way to control routes intersections in
traditional networks
Huawei Shenzhen R.Smelyanskiy 2126.05.2015
21. FDM-SDN provides a new leverage
• Flow demultiplexing along several non-xing routes:
– Goal: get connection with required bandwidth
– Strategy: use cut-and-try to detect the amount of provided
resources and utilize all of them on several routes
– Primary heuristic: decries the number of routes
(optimistic)
– Parameter: cumulative bandwidth of all routes
– Primary mode: keep busy all routes
– Triggering criteria: bandwidth deficiency
– Recovering action: open new non-xing route
Huawei Shenzhen R.Smelyanskiy 2226.05.2015
23. Routing FDMP flows with SDN
Host A Host B
SDN Controller
Has MP_CAPABLE option?
Install new FDMP connection!
SYN
MP CAP
Key A
Set up new FDMP connection
Huawei Shenzhen R.Smelyanskiy 2526.05.2015
24. Host A Host B
SDN Controller
Has MP_CAPABLE option?
Complete partial FDMP connection!
SYN, ACK
MP CAP
Key B
Routing FDMP flows with SDN
Set up new FDMP connection
Huawei Shenzhen R.Smelyanskiy 2626.05.2015
25. Host A Host B
SDN Controller
Has MP_JOIN option?
Install new FDMP subflow for a
known connection!
SYN
MP JOIN
Token B
Routing FDMP flows with SDN
Set up new FDMP subflow
Huawei Shenzhen R.Smelyanskiy 2726.05.2015
26. Host A Host B
SDN Controller
Subflow is not active any more!
Remove the path!
Routing FDMP flows with SDN
FDMP subflow manipulation
Actually, we store metadata and
allow some subflows to resume.
We use flow eviction to
remove this data.
Huawei Shenzhen R.Smelyanskiy 2926.05.2015
27. Actually, we store metadata and
allow some subflows to resume.
We use flow eviction to
remove this data.
Host A Host B
SDN Controller
Get FDMP packet of a expired subflow!
Either reroute the remembered subflow,
or force hosts to close it
RSTRST
Close violet subflow!
Reschedule the packet to
red subflow!
Close violet subflow!
Routing FDMP flows with SDN
FDMP subflow manipulation
Huawei Shenzhen R.Smelyanskiy 3026.05.2015
30. Conclusion
1. In context of bandwidth, the party of FDMP & our
routing application outperform both single-flow TCP
and MP TCP without regards to ECMP
2. FDMP improves efficiency of infrastructure:
It can increase utilization under the same load and
allows network to process more traffic
3. Under a heavy load FDMP will support faster
communication and/or transmit more data within
the same period of time
Huawei Shenzhen R.Smelyanskiy 3326.05.2015
31. Challenges we are working on
Use TCP-FDM to reduce delay and packet loss:
– Which packet scheduling algorithm to use?
– What routing algorithms are appropriate?
Enforce resource allocation & QoS policies:
– How to avoid flow competition on congestion?
– Do we need to control request frequency?
– Is it efficient enough to rate-limit subflows?
– How to combine TCP-FDM and Diff Serv model?
Develop FDMP as VNF within network:
– How to integrate TCP-FDM routing application?
– How to build an efficient TCP to TCP-FDM proxy?
Huawei Shenzhen R.Smelyanskiy 3426.05.2015