External Data Representation XDR

1
Lecture 11

• External Data Representation (XDR)

2
Overview

• Data Representation
• N-Squared Conversion Problem
• Network Standard Byte Order
• De Facto XDR Standard
• XDR Data Types
• XDR Library Routines
• Building XDR Messages
• Summary

3
Data Representation

• Each computer architecture provides its own
  representation of data
• LSB at the lowest memory address
• MSB at the lowest memory address
• Client/Server applications must contend with data
  representation

decreasing memory address
increasing memory address
0
0
1
4
4
1
0
0
4
N-Squared Conversion Problem

• The underlying issue of data representation is
  software portability
• Asymmetric Data Conversion
• Converts directly from the clients
  representation into the servers implementation
• Must write a client-server pair for each pair of
  architectures
• The programming effort is proportional to the
  square of the number of different data
  representations (N2 N) / 2

5
N-Squared Conversion Problem (Cont)
If client-server software is designed to convert
from the clients native data representation
directly to the servers native data
representation asymmetrically, the number of
versions of the software grows as the square of
the number of architectures
6
N-Squared Conversion Problem (Cont)

• Solving the N-squared conversion problem
• Develop client/server software so that it can be
  compiled and executed on a variety of machines
• Results in a highly portable program
• Makes accessing services easier

7
Network Standard Byte Order

• Solves the data representation problem by
  converting the native byte order to a network
  standard byte order
• TCP/IP protocol software uses symmetric data
  conversion
• Both ends perform the required conversion
• Only one version of the protocol is needed
• Both client and server perform a data conversion
• Data is represented in a standard
  machine-independent format
• The standard format is known as external data
  representation (XDR)
• Advantage of symmetric conversion
• Added flexibility neither the client nor the
  server needs to understand the architecture of
  the other
• Disadvantage of symmetric conversion
• Computational overhead using an intermediate
  format introduces additional computation

8
Network Standard Byte Order (Cont)

• Converting between heterogeneous representations
  requires extra CPU cycles
• Conversion my result in a larger stream of bytes
• Symmetric conversion is still advantageous
• Simplifies programmer
• Reduces errors
• Increases interoperability
• Network management and debugging is easier

9
De Facto XDR Standard

• Sun Microsystems devised an external data format
  called eXternal Data Representation (XDR)
• XDR is a standard for the description and
  encoding of data
• XDR specifies data formats for most data types
  that clients/servers exchange
• XDR specifies that 32-bit binary numbers should
  be represented in big endian order
• Client/Servers using XDR must agree on the exact
  format of messages exchanged
• XDR uses a language to describe data formats. The
  language can only be used only to describe data
  it is not a programming language.
• This language allows one to describe intricate
  data formats in a concise manner.

10
XDR Data Types

• The XDR standard defines various basic data types
  and allows for the definition of compound data
  types (structs), fixed-size and variable-size
  arrays as well as unions.
• XDR provides representations for most C
  programming structures

11
Implicit Types

• Client/Servers using XDR must agree on the exact
  format
• Encoding does not identify its type or length
• XDR message cannot be decoded unless it knows the
  exact format and types of all data fields

12
Software Support for XDR

• Data must be converted from native format to the
  XDR representation before sending across the
  network
• Data must be converted from XDR representation to
  native format when receiving data from the
  network
• XDR conversion libraries are used to eliminate
  potential conversion errors
• Converting floating point representation to the
  XDR standard

13
XDR Library Routines

• XDR routines can convert data items from the
  machines native format to the XDR format and
  vice-versa
• Most XDR implementations use a buffer paradigm
• Uses a program to allocate a buffer large enough
  to hold the external format of a message
• The application calls XDR conversion routines for
  each data item in the message
• After encoding each data item and placing it in
  the XDR stream the application can send the
  message

14
Building an XDR Message

• The XDR buffer paradigm uses a buffer to hold the
  external representation of the message
• Need an XDR Pointer
• XDR xdrs
• Buffer to hold the data
• Using functions you can create and add to the
  buffer, and get data from the buffer
• xdrmem_create(xdrs, buf, BUFSIZE, XDR_ENCODE)
• xdr_int(xdrs, i)

15
XDR Stream Buffer
stream header
0
1
2
3
4
5
6
An XDR stream that has been initialized for
ENCODE and already contains 7 bytes of data
stream header
0
1
2
3
4
5
6
0
0
1
4
XDR stream after a call to xdr_int appends a
32-bit integer with value 260
16
XDR Streams, I/O, TCP UDP

• Method 1 Build entire message, then write
• Method 2 Open Standard I/O stream (FILE object)
  and create using xdrstdio_create()
• now each read/write will convert on the fly
• Method 3 Open UDP datagram interface
• use xdrrec_create() and send data in datagrams

17
Summary

• Client/Server programs must contend with data
  representation problems
• Asymmetric conversion
• The client or server converts from its native
  format to the remote machines native format
• N-Squared Conversion Problem
• Symmetric Conversion
• Uses a standard network representation
• Requires both the client/server programs convert
  between the standard format and native format
• Increased flexibility but added computational
  overhead