Lecture 23 Reminders: Programming Project 5 (reassigned Homework - - PowerPoint PPT Presentation

Tuesday, November 15 CS 475 Networks - Lecture 23 1

Lecture 23

 Reminders: Programming Project 5 (reassigned

Homework Problem 5.48) due on Thursday. Homework 7 due next Tuesday.

 Questions?

Tuesday, November 15 CS 475 Networks - Lecture 23 2

Outline

Chapter 7 - End-to-End Data

7.1 Presentation Formatting 7.2 Multimedia Data 7.3 Summary

Tuesday, November 15 CS 475 Networks - Lecture 23 3

Introduction

A sender and receiver must agree to a message

• r presentation format. How do we send

traditional (integers, floating-point numbers, characters, arrays, structures) and multimedia (audio, image, video) data? We may want to use data compression to encode the data as efficiently as possible for network

• transmission. However, compression and

decompression processing places increased demands on the end hosts.

Tuesday, November 15 CS 475 Networks - Lecture 23 4

Presentation Formatting

Data must be encoded into a form suitable for transmission over the network. It is decoded at the receiver. (Encoding/decoding are also known as marshalling/unmarshalling.)

Tuesday, November 15 CS 475 Networks - Lecture 23 5

Presentation Formatting

Encoding is necessary because computers represent data in different ways. For example, the native size of an integer may differ between computers and languages. Computers also have different endianess.

Tuesday, November 15 CS 475 Networks - Lecture 23 6

Taxonomy – Data Types

Marshalling systems differ in the data types they

• support. At the lowest level is support for the

base types (integers, floats, characters). At the next level is support for flat types – structures and arrays. At the highest level is support for complex types that contain pointers. The pointer can not be simply copied across. The data pointed to must also be copied. This is known as flattening or serialization.

Tuesday, November 15 CS 475 Networks - Lecture 23 7

Taxonomy – Data Types

Argument marshalling: converting, packing and linearizing

Tuesday, November 15 CS 475 Networks - Lecture 23 8

Taxonomy – Conversion Strategy

Some systems convert all data to canonical intermediate form. For example a system may require that all integers be sent in big-endian regardless of the endianess of either the source

• r receiver.

The alternative approach is receiver-makes-right in which the sender transmits data in its native

• form. The receiver is responsible for translating

the data to its own local form.

Tuesday, November 15 CS 475 Networks - Lecture 23 9

Taxonomy - Tags

Tags can be sent with the data to indicate the type and length of a data item. A tag can also be used to indicate the endianess of the data in a receiver-makes-right system. If tags are not used then the receiver must be programmed to expect data of a particular type in a particular order.

A 32-bit integer encoded in a tagged message

Tuesday, November 15 CS 475 Networks - Lecture 23 10

Taxonomy - Stubs

Stubs are typically used to implement argument marshalling in support of RPC. They may be either compiled or interpreted. Compiled stubs are more efficient while interpreted stubs are more flexible. Interpreted stubs translate data according to a data description file. Changing the data description file does not require recompilation.

Tuesday, November 15 CS 475 Networks - Lecture 23 11

Taxonomy - Stubs

A stub compiler produces stubs from an interface description.

Tuesday, November 15 CS 475 Networks - Lecture 23 12

Examples - XDR

External Data Representation (XDR) is used with SunRPC for argument marshalling.

 It supports the entire C type system except for function

pointers.

 It uses a canonical intermediate form.  It does not use tags (except to indicate array lengths)  It uses compiled stubs.

Tuesday, November 15 CS 475 Networks - Lecture 23 13

Examples - XDR

Example XDR encoding of a structure containing a 32-bit integer, a 7 character string, and a 3 element 32-bit integer array.

Tuesday, November 15 CS 475 Networks - Lecture 23 14

Examples - ASN.1

Abstract Syntax Notation One (ASN.1) is an ISO standard that defines a representation (the Basic Encoding Rules or BER) for data sent over a network. It supports all C types except function pointers, uses a canonical intermediate form, and it uses

• tags. The stubs can be either compiled or

interpreted.

Tuesday, November 15 CS 475 Networks - Lecture 23 15

Examples - ASN.1

Compound types (structures) can be created by means of nesting in ASN.1/BER. ASN.1/BER encoding of a 32-bit integer. The value is preceded by TYPE and LENGTH tags.

Tuesday, November 15 CS 475 Networks - Lecture 23 16

Examples - ASN.1

ASN.1/BER LENGTH tags can be of variable size. If the data item is less than 128 bytes in length, the LENGTH tag is a single byte. If the data item is longer than 127 bytes, that LENGTH tag is multiple bytes long. The MSB of the leading byte is one. The remaining 7 bits indicate the length of the LENGTH tag.

Tuesday, November 15 CS 475 Networks - Lecture 23 17

Examples - NDR

Network Data Representation (NDR) uses receiver makes right. It supports the C type

• system. Data items are not tagged. Compiled

stubs are generated from the Interface Description Language (IDL). NDR inserts an architecture tag (shown below) at the front of each message so that the receiver knows how to translate the data from source format to a local format.

Tuesday, November 15 CS 475 Networks - Lecture 23 18

Markup Languages (XML)

In XML all data are tagged and represented using text.

<?xml version=”1.0”?> <employee> <name>John Doe</name> <title>Head Bottle Washer</title> <id>123456789</id> <hiredate> <day>5</day> <month>June</month> <year>1986</year> </hiredate> </employee>

Tuesday, November 15 CS 475 Networks - Lecture 23 19

Markup Languages (XML)

The XML data types and compound data type formats are defined in an XML Schema Definition (XSD) file. XSDs are also written in XML. XML messages are easy to read and edit. They are naturally portable between architectures. They are inefficient (compared to binary representations of data) and slow to parse.

Tuesday, November 15 CS 475 Networks - Lecture 23 20

In-Class Exercise

 Write a C++ program that writes your name (first and

last names), your age (as an integer) and the first 5 digits of π (3.1415) to memory using XDR (refer to the xdr man page). Normally this block of memory would then be sent across the network to another computer. Simulate this by copying the memory block to a new location (use memcpy) and then have your program read the values from the new location and display them to standard output.

 Email your program to the instructor. It is due by the

beginning of the Thursday's class.