Pub Date : 2002-11-07DOI: 10.1109/HPSR.2002.1024244
Michael Stübert Berger
The basic concept of packet bundling is to group smaller packets into larger packets based on, e.g., quality of service or destination within the packet switch. This paper presents novel applications of bundling in packet switching. The larger packets created by bundling are utilized to extend switching capacity by use of parallel switch planes. During the bundling operation, packets will experience a delay that depends on the actual implementation of the bundling and scheduling scheme. Analytical results for delay bounds and buffer size requirements are presented for a specific scheduling algorithm, and compared to simulation results.
{"title":"Multipath packet switch using packet bundling","authors":"Michael Stübert Berger","doi":"10.1109/HPSR.2002.1024244","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024244","url":null,"abstract":"The basic concept of packet bundling is to group smaller packets into larger packets based on, e.g., quality of service or destination within the packet switch. This paper presents novel applications of bundling in packet switching. The larger packets created by bundling are utilized to extend switching capacity by use of parallel switch planes. During the bundling operation, packets will experience a delay that depends on the actual implementation of the bundling and scheduling scheme. Analytical results for delay bounds and buffer size requirements are presented for a specific scheduling algorithm, and compared to simulation results.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114276913","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 : 2002-11-07DOI: 10.1109/HPSR.2002.1024236
K. Yamakoshi, E. Oki, N. Yamanaka
A dynamic deficit round-robin (DDRR) scheduling scheme for a 5-Tb/s switch is proposed. DDRR is a DRR-based packet scheduler and it can satisfy the max-min fair share. However, DRR cannot satisfy both high throughput and delay requirements. DDRR resolves the problem by changing the granularity for deficit counters according to the packet lengths at queue heads. Simulation results showing the efficiency of DDRR are presented and an implementation of DDRR for the switch is also described.
{"title":"Dynamic deficit round-robin scheduler for 5-Tb/s switch using wavelength routing","authors":"K. Yamakoshi, E. Oki, N. Yamanaka","doi":"10.1109/HPSR.2002.1024236","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024236","url":null,"abstract":"A dynamic deficit round-robin (DDRR) scheduling scheme for a 5-Tb/s switch is proposed. DDRR is a DRR-based packet scheduler and it can satisfy the max-min fair share. However, DRR cannot satisfy both high throughput and delay requirements. DDRR resolves the problem by changing the granularity for deficit counters according to the packet lengths at queue heads. Simulation results showing the efficiency of DDRR are presented and an implementation of DDRR for the switch is also described.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124253468","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}
In optical burst-switched (OBS) networks, the most important problem is the collision problem, which occurs when burst packets contend for the same outgoing interface. TCP-based retransmission is a poor solution since it complicates the ingress edge router due to sequencing, and requires excessive amount of memory. We propose a single-anchored soft bandwidth allocation system for OBS networks to provide 100 percent reliability without retransmissions. The anchor node, which is the OBS node that blocks the burst, stores the burst, and informs the ingress router by a negative acknowledgment message. The anchor also provides a bandwidth reservation between itself and the egress OBS node using probe-probe-ACK handshake. This handshake provides soft allocation state along the path. The LSPs may be in "anchor mode" or "non-anchor mode". Soft allocation does not allow other "anchor mode" LSPs to use the allocated bandwidth, but "non-anchor mode" LSPs may use with low priority. The high priority is reserved for the "anchor mode" LSP. A soft bandwidth allocation system provides reliability. Another advantage is that no retransmission is needed in this mechanism.
{"title":"Single-anchored soft bandwidth allocation system with deflection routing for optical burst switching","authors":"Timucin Ozugur, Farid Farahmand, Dominique Verchere","doi":"10.1109/HPSR.2002.1024246","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024246","url":null,"abstract":"In optical burst-switched (OBS) networks, the most important problem is the collision problem, which occurs when burst packets contend for the same outgoing interface. TCP-based retransmission is a poor solution since it complicates the ingress edge router due to sequencing, and requires excessive amount of memory. We propose a single-anchored soft bandwidth allocation system for OBS networks to provide 100 percent reliability without retransmissions. The anchor node, which is the OBS node that blocks the burst, stores the burst, and informs the ingress router by a negative acknowledgment message. The anchor also provides a bandwidth reservation between itself and the egress OBS node using probe-probe-ACK handshake. This handshake provides soft allocation state along the path. The LSPs may be in \"anchor mode\" or \"non-anchor mode\". Soft allocation does not allow other \"anchor mode\" LSPs to use the allocated bandwidth, but \"non-anchor mode\" LSPs may use with low priority. The high priority is reserved for the \"anchor mode\" LSP. A soft bandwidth allocation system provides reliability. Another advantage is that no retransmission is needed in this mechanism.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122577791","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 : 2002-11-07DOI: 10.1109/HPSR.2002.1024242
M. Herbert, P. Primet, B. Tourancheau, L. Lefèvre
We propose a new scalable et flexible architecture for Ethernet switching. First we identify the legacy bottlenecks in classical Ethernet backbones: mainly the tree topology and the redundant packet forwarding decisions. Then we describe how a fully distributed communication architecture, based on proven, high-performance computing concepts (source routing, multistage interconnection, wormhole switching) can provide a scalable backbone inside, with Ethernet compatibility outside. Thanks to carefully placed intelligence, materialized as "network processors", we show that not only scalability is enhanced but also flexibility and global transistor count. We present a prototype of this architecture implemented with Myrinet technology.
{"title":"A scalable and fully distributed architecture for Ethernet switching","authors":"M. Herbert, P. Primet, B. Tourancheau, L. Lefèvre","doi":"10.1109/HPSR.2002.1024242","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024242","url":null,"abstract":"We propose a new scalable et flexible architecture for Ethernet switching. First we identify the legacy bottlenecks in classical Ethernet backbones: mainly the tree topology and the redundant packet forwarding decisions. Then we describe how a fully distributed communication architecture, based on proven, high-performance computing concepts (source routing, multistage interconnection, wormhole switching) can provide a scalable backbone inside, with Ethernet compatibility outside. Thanks to carefully placed intelligence, materialized as \"network processors\", we show that not only scalability is enhanced but also flexibility and global transistor count. We present a prototype of this architecture implemented with Myrinet technology.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123612249","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 : 2002-11-07DOI: 10.1109/HPSR.2002.1024207
Po-Chou Lin, Chung-Ju Chang
Within an ATM network domain, the guaranteed frame rate (GFR) service has been designed to accommodate non-real-time application, such as TCP/IP eccentric traffic. The GFR service not only guarantees minimum cell rate (MCR) throughput at the frame level but also supports a fair share of available bandwidth. We propose a high-speed novel fair buffer allocation (HS-NFBA) scheme through a FIFO queueing discipline, which further achieves low average cell queueing delay, and efficient hardware architecture supporting up to 10 Gbps synchronous transfer mode (STM)-64 line rate.
{"title":"A high-speed novel fair buffer allocation for ATM-GFR services in TCP/IP internetworks","authors":"Po-Chou Lin, Chung-Ju Chang","doi":"10.1109/HPSR.2002.1024207","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024207","url":null,"abstract":"Within an ATM network domain, the guaranteed frame rate (GFR) service has been designed to accommodate non-real-time application, such as TCP/IP eccentric traffic. The GFR service not only guarantees minimum cell rate (MCR) throughput at the frame level but also supports a fair share of available bandwidth. We propose a high-speed novel fair buffer allocation (HS-NFBA) scheme through a FIFO queueing discipline, which further achieves low average cell queueing delay, and efficient hardware architecture supporting up to 10 Gbps synchronous transfer mode (STM)-64 line rate.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131674053","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 : 2002-11-07DOI: 10.1109/HPSR.2002.1024213
Hong Huang, J. Copeland
Designing simple and efficient network protection mechanism is an important requirement of optical networks, which is a challenging task for large mesh networks. We extend our previous work on Hamiltonian cycle protection to large-scale multiple domain inhomogeneous optical networks. We proposed two protection schemes with different trade-off between spare resource usage and failure isolation. Our proposed schemes are simple and competitive in network resource usage.
{"title":"Multi-domain mesh optical network protection using Hamiltonian cycles","authors":"Hong Huang, J. Copeland","doi":"10.1109/HPSR.2002.1024213","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024213","url":null,"abstract":"Designing simple and efficient network protection mechanism is an important requirement of optical networks, which is a challenging task for large mesh networks. We extend our previous work on Hamiltonian cycle protection to large-scale multiple domain inhomogeneous optical networks. We proposed two protection schemes with different trade-off between spare resource usage and failure isolation. Our proposed schemes are simple and competitive in network resource usage.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128888664","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 : 2002-11-07DOI: 10.1109/HPSR.2002.1024634
T. Kurimoto
Next generation Internet backbone router, Type-X, will be introduced. Type-X can offer sophisticated QoS support including real time voice, video streaming, and VPN. The Type-X has very-large-capacity 320 Gbps throughput and IPV4N6 dual stack forwarding capability. Note that, employing the-state-of-the-art VLSI technology, both IPV4 and V6 OC-192C full wire-rate forwarding performance is achieved. In addition, Type-X has ATM cell tunneling over MPLS technique and high-reliability technique such as newly structure non-stop software firmware up grade technique and hot swappable technique. This new backbone router offers next generation, real time, high-QoS capability, and VPN backbone IP network to be realized. We will give three demonstrations as follows. (1) Demonstration of the OC192 POS line performance. OC-192C full wire-rate forwarding performance will be
{"title":"High-speed core router, type-X","authors":"T. Kurimoto","doi":"10.1109/HPSR.2002.1024634","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024634","url":null,"abstract":"Next generation Internet backbone router, Type-X, will be introduced. Type-X can offer sophisticated QoS support including real time voice, video streaming, and VPN. The Type-X has very-large-capacity 320 Gbps throughput and IPV4N6 dual stack forwarding capability. Note that, employing the-state-of-the-art VLSI technology, both IPV4 and V6 OC-192C full wire-rate forwarding performance is achieved. In addition, Type-X has ATM cell tunneling over MPLS technique and high-reliability technique such as newly structure non-stop software firmware up grade technique and hot swappable technique. This new backbone router offers next generation, real time, high-QoS capability, and VPN backbone IP network to be realized. We will give three demonstrations as follows. (1) Demonstration of the OC192 POS line performance. OC-192C full wire-rate forwarding performance will be","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130181267","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 : 2002-11-07DOI: 10.1109/HPSR.2002.1024630
K. Mori
{"title":"Small 4 channel 2x2 MEMS optical switch","authors":"K. Mori","doi":"10.1109/HPSR.2002.1024630","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024630","url":null,"abstract":"","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128543036","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 : 2002-11-07DOI: 10.1109/HPSR.2002.1024241
J. Choi, Hyeong-Ho Lee
In this paper, we investigate the performance of a nonblocking input queueing asynchronous transfer mode (ATM) switch with windowing scheme for IP switching system. In particular, this paper analyzes the effects of nonuniform and correlated bursty traffic on the performance of the input queueing switch with windowing scheme. We consider four specific traffic patterns: uncorrelated non-bursty, correlated bursty, uniform and nonuniform traffic. Under the uncorrelated non-bursty traffic, we obtain the maximum throughput as a closed-form solution, and the analytic results are verified by simulation. On the other hand, under correlated bursty traffic, we obtain the maximum throughput through simulation. From the results, we know that the windowing scheme can reduce the head-of-line (HOL) blocking at the input ports and increase the maximum throughput up to 1 under uncorrelated and uniform traffic. However, the windowing scheme can not provide the performance improvement of the input queueing switch because of the congested output blocking under correlated bursty and/or nonuniform traffic. So, the designer should take into account the windowing scheme combined with a separate logical output queue for each output port and traffic load balancing or randomization techniques to build IP switching systems.
{"title":"Performance study of an input queueing ATM switch with windowing scheme for IP switching system","authors":"J. Choi, Hyeong-Ho Lee","doi":"10.1109/HPSR.2002.1024241","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024241","url":null,"abstract":"In this paper, we investigate the performance of a nonblocking input queueing asynchronous transfer mode (ATM) switch with windowing scheme for IP switching system. In particular, this paper analyzes the effects of nonuniform and correlated bursty traffic on the performance of the input queueing switch with windowing scheme. We consider four specific traffic patterns: uncorrelated non-bursty, correlated bursty, uniform and nonuniform traffic. Under the uncorrelated non-bursty traffic, we obtain the maximum throughput as a closed-form solution, and the analytic results are verified by simulation. On the other hand, under correlated bursty traffic, we obtain the maximum throughput through simulation. From the results, we know that the windowing scheme can reduce the head-of-line (HOL) blocking at the input ports and increase the maximum throughput up to 1 under uncorrelated and uniform traffic. However, the windowing scheme can not provide the performance improvement of the input queueing switch because of the congested output blocking under correlated bursty and/or nonuniform traffic. So, the designer should take into account the windowing scheme combined with a separate logical output queue for each output port and traffic load balancing or randomization techniques to build IP switching systems.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125420752","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 : 2002-11-07DOI: 10.1109/HPSR.2002.1024237
J. M. Ng, R. N. Susilo
For some real-time applications, it is necessary to provide a fast switch over to the backup connection whenever the primary connection fails. With the increase in reliability of network links, it is more cost effective if the backup resources are shared among a number of users. However, the sharing of backup resources must be controlled to ensure that users are provided with the required level of availability on the backup connections. In this paper, we proposed an approach for establishing backup connections based on the availability requirements of the applications and provides an evaluation on the effectiveness of such an approach.
{"title":"Evaluation of a controlled shared backup mechanism","authors":"J. M. Ng, R. N. Susilo","doi":"10.1109/HPSR.2002.1024237","DOIUrl":"https://doi.org/10.1109/HPSR.2002.1024237","url":null,"abstract":"For some real-time applications, it is necessary to provide a fast switch over to the backup connection whenever the primary connection fails. With the increase in reliability of network links, it is more cost effective if the backup resources are shared among a number of users. However, the sharing of backup resources must be controlled to ensure that users are provided with the required level of availability on the backup connections. In this paper, we proposed an approach for establishing backup connections based on the availability requirements of the applications and provides an evaluation on the effectiveness of such an approach.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126268483","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}