Stability analysis of a parallel packet switch with bufferless input demultiplexers

نویسندگان

  • Denis A. Khotimsky
  • Santosh Krishnan
چکیده

A class of packet switching systems with inverse multiplexing of switched connections and/or flows (the SCIMUX architecture) has been recently proposed. Such systems open the possibility to use the readily available switch fabrics and port cards in order to build fullyfunctional packet switches, operating at rates much higher than those of their own components taken in isolation. While a comprehensive analytical study of such systems is in high demand, it remains a largely unexplored research area. In this paper we consider the Parallel Packet Switch (PPS) architecture, which is a variation of the SCIMUX architecture with the switching core composed of multiple parallel independent switching planes, restricting our attention to the case with bufferless input demultiplexors. We formalize the notion of comparing “degrees of congestion” in the switches operating at the different port rates by introducing the concept of relative stability. We describe a family of traffic dispatch algorithms and prove that they are necessary and sufficient to guarantee relative stability of a switching plane in the PPS architecture. Finally, we obtain a tight lower bound on the number of required switching planes K, show that the equivalent core speedup bound is given by S K=dK=2e, and prove that meeting these bounds is necessary and sufficient to achieve stability of the entire Parallel Packet Switch. Our results require work-conserving behavior of each switching plane without imposing any restrictions on the queuing structure and service discipline. They naturally generalize to the case of multiple traffic flows between the same pair of external ports and may be applied to QoSaware switches. The described algorithms are fully distributed and do not require state information of any kind to be exchanged between the architecture components. Keywords— Inverse Multiplexing, Multipath Switching Systems, Parallel Switch architectures.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Design and Analysis of a Fully-Distributed Parallel Packet Switch with Buffered Demultiplexers

A Parallel Packet Switch (PPS) is a multistage switch aimed at building a very high-speed switch using much slower devices. A PPS in general has three stages. Several packet switches are placed in the central stage, which operate slower than the external line’s rate. Incoming packets are spread over the center-stage switches by demultiplexers at the input stage. Packets destined to each output ...

متن کامل

WDM Optical Packet Switches with Bufferless and Partially Shared Buffering Wavelength Converters

Abstract-In this paper we introduce two architectures for optical packet switches. One is a bufferless switch employing shared Tunable Wavelength Converters (TWCs), and the other is a packet switch with Partially Shared Buffering (PSB). They were proposed for solving problems of contention resolution. The bufferless optical packet switch employing shared TWCs allows a saving in terms of employe...

متن کامل

A 1-Cycle 1.25 GHz Bufferless Router for 3D Network-on-Chip

In this paper, we propose a 1-cycle high-performance 3D bufferless router with a 3-stage permutation network. The proposed router utilizes the 3-stage permutation network instead of the serialized switch allocator and 7× 7 crossbar to achieve the frequency of 1.25GHz in TSMC 65nm technology. Compared with the other two 3D bufferless routers, the proposed router occupies less area and consumes l...

متن کامل

Parallel Packet Switching Using Multiplexors with Virtual Input Queues

Parallel Packet Switches (PPS) use internal, parallel switch planes that operate at less than line speed. A PPS can scale-up to faster line speeds than a single-plane switch can. Load balancing between planes and providing QoS to flows are open problems. We investigate the performance of a new “VIQ PPS” that distributes cells or variable-length packets to planes and uses multiplexors with virtu...

متن کامل

Efficient Queuing Architecture for a Buffered Crossbar Switch

This paper presents a modified architecture for a buffered crossbar switch that overcomes the memory bottleneck with only a minor impact on performance. The proposed architecture uses two levels of backpressure with different constraints on round trip time. Buffered crossbars are considered an alternative to bufferless crossbars mainly because the latter requires a complex scheduling algorithm ...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2001