Resource Allocation and Queuing Theory

Transcription

1 and Modeling Modeling Networks

2 Outline 1 Introduction Why are we waiting?... 2 Packet-Switched Network Connectionless Flows Service Model Router-Centric versus Host-Centric Reservation Based versus Feedback-Based Window-Based versus Rate-Based Window-Based versus Rate-Based

3 Why are we waiting?... Outline 1 Introduction Why are we waiting?... 2 Packet-Switched Network Connectionless Flows Service Model Router-Centric versus Host-Centric Reservation Based versus Feedback-Based Window-Based versus Rate-Based Window-Based versus Rate-Based

4 Why are we waiting?... Why are we waiting? Fundamentals There are lots of examples for queuing situations: Passport control Supermarkets Planes waiting to take off or land Programs being processed by uniprocessor/multiprocessor Packets at sending side of communication channel Packets at receiving side of communication channel Each of the above involves different states: waiting, being processed, processed, together with transitions between each of the states.

5 Why are we waiting?... Why are we waiting? Fundamentals Waiting is a consequence of the randomness of the processing procedure arrival times - not known in advance processing times - not known in advance number of times we have to wait in a queue for either the same or a different service - not known in advance If these were known then we could schedule in advance and eliminate waiting altogether. In order to model queuing scenarios we employ stochastic processes reflecting the random indeterminate nature of the particular situation being studied.

6 Why are we waiting?... Why are we waiting? How many more times...? We explore operations of interest operating under random processes in order to gain a better understanding of the operations characteristics. We might ask how long do we expect to wait prior to processing? - (customer question) how long is a service likly to be idle? - (service question) how can a reasonable balance be obtained between the two questions above? We seek to identify those characteristics of a system that will allow us to measure the performance of a system.

7 Why are we waiting?... Why are we waiting? Fundamentals In working with queues we need to acquaint ourselves with: Assumptions Actors Arrival and Service Times - (Modeled by Stochastic Processes) Service Discipline - (FIFO, FQ, LIFO, SIRO, Priorities) Design and Execution - (Single, Parallel, Series, Tandem, Network) Queue size - (finite, infinite) nature of Calling Source - (generates finite/infinite number of customers) Customer and service behaviour - (processing speed a function of length of queue; customer: jockeying, balking, reneging)

8 Outline Introduction 1 Introduction Why are we waiting?... 2 Packet-Switched Network Connectionless Flows Service Model Router-Centric versus Host-Centric Reservation Based versus Feedback-Based Window-Based versus Rate-Based Window-Based versus Rate-Based

9 Introduction and Congestion Control are longstanding active areas of research Resource allocation partially implemented in routers / switches in a network the transport protocol running on the end hosts End systems use signalling protocols to indicate resource requirements to network nodes Network nodes respond with information on resource availability

10 Definition is defined to be the process by which network elements try to meet the competing demands that applications have for network resources, by which we mean Link bandwidth (BW) Buffer space in routers / switches Note: it is not always possible to meet all of the demands placed on a resource some users / applications may receive fewer resources than they require

11 Introduction Throughout the following discussion flow control will refer to practices that ensure a fast sender doesn t overpower a slow receiver congestion control will refer to practices that ensure a set of senders do not send too much data into the network due to a lack of resources at some point

12 Outline Introduction 1 Introduction Why are we waiting?... 2 Packet-Switched Network Connectionless Flows Service Model Router-Centric versus Host-Centric Reservation Based versus Feedback-Based Window-Based versus Rate-Based Window-Based versus Rate-Based

13 Introduction Packet Switched Network Multiple links and switches packets may encounter a bottleneck at some point in the network in which, for example high speed links feed into a low speed link Figure: A Potential Bottleneck Router

14 Introduction Figure: A Potential Bottleneck Router

15 - Connectionless Flows Assume network essentially connectionless any connection oriented service implemented in transport protocol running on end hosts Internet model Sequence of packets (datagram delivery service) TCP implements end to end connection abstraction connection set up message traverses network reserves a set of buffers for connection at routers (under utilisation of available resources since not available for general use) we focus on connectionless networks

16 - Connectionless Flows What do we mean by connectionless? All datagrams are completely independent - Modeled by Poisson Distribution - car model Although datagrams are switched independently usually the case that stream of datagrams flow between a particular pair of hosts through a particular pair of hosts - train model Definition Model assumes a flow abstraction as a sequence of packets between source and destination following the same route through the network flows can be defined at different granularities Host to host (same source / destination addresses) process to process (same source / destination host / port pairs

17 - Connectionless Flows Figure: Multiple flows passing through a set of routers

18 - Connectionless Flows State information for each flow kept at routers (soft state) used to make resource allocation decisions regarding packets that belong to the flow Soft state not always explicitly created and removed by signalling a middle ground between purely connectionless network maintaining no state at routers purely connection oriented network maintaining hard state at routers

19 - Connectionless Flows Soft state exists only for purpose of resource allocation not essential for correct operation of network however router better able to handle packets belonging to flow if soft state maintained for flow does not imply relible and ordered delivery of a virtual circuit

20 - Service Model Introduction In the Best Effort Service Model each packet is treated in the same way flows are given no guarantees Defining a service that gives some form of guarantees for example the bandwidth required for a video stream is said to provide multiple qualities of service (QoS). Clearly a range of possibilities exist, regarding the options available here.

21 Outline Introduction 1 Introduction Why are we waiting?... 2 Packet-Switched Network Connectionless Flows Service Model Router-Centric versus Host-Centric Reservation Based versus Feedback-Based Window-Based versus Rate-Based Window-Based versus Rate-Based

22 - Router-Centric versus Host-Centric Two broad groups Those that address the problem of Resource allocation from within the network (routers / switches) Those that address the problem from the edges (end hosts) Real issue is considered to be where does the main burden resides

23 - Router-Centric versus Host-Centric Router-centric each router takes responsibility for deciding when packets are forwarded which packets are dropped informing hosts generating traffic how many packets they are allowed to send Host-centric end hosts observe network conditions (how many packets are getting through) and adjust accordingly We note that these groups are not mutually exclusive, both play their part

24 - Reservation Based versus Feedback-Based Should we make Reservations or wait for Feedback? Reservation Based System End hosts ask for network for a certain amount of capacity at time when flow established Each router allocates resources (buffers and/or % of link bandwidth) If router cannot commit to requirements then request denied

25 - Reservation Based versus Feedback-Based Feedback Based System End hosts send data without recourse to network End Hosts adjust sending rate subject to feedback EXPLICIT feedback - Router sends message slow down IMPLICIT feedback - end host adjusts sending rate subject to network conditions e.g. packet loss Reservation Based implies router centric resource allocation since each router responsible for allocation and policing of its capacity Explicit Feedback-Based implies router centric based mechanism Implicit Feedback-Based implies Host centric mechanism, routers silently drop packets when congested

26 - Window-Based versus Rate-Based Two general ways to express to the sender the amount of data that may be sent Windows based transport protocol indicates the size of window (amount of buffer space) available at the receivers end This limits amount of data that sender can send - so a flow control mechanism a similar mechanism window advertisement can be used to reserve buffer space - so a reource allocation mechanism (e.g. X.25)

27 - Window-Based versus Rate-Based Rate based A Rate based transport protocol indicates the size of indicates the amount of data that the receiver or network is able to absorb A rate based characterisation of flow is felt to be a logical choice in a reservation based system supporting different qualities of service Sender makes reservation for so many bits per second, each router on path determines if it can commit to reservation subject to current committments See summary on page 464