Index Journeys
RFC 6340 PROPOSED STANDARD SNMP · MIB · DATA TYPES

Textual Conventions for the Representation of Floating-Point Numbers

Three ready-made SNMP data types for carrying 32-, 64- and 128-bit IEEE floating-point numbers.

On rfc-editor.org ↗
PUBLISHED
August 2011
AUTHORS
R. Presuhn
STREAM
IETF
FLOAT32TC SIZE
4 octets
FLOAT64TC SIZE
8 octets
FLOAT128TC SIZE
16 octets
UNDERLYING TYPE
OCTET STRING
MODULE OID
{ mib-2 201 }
DOI
10.17487/RFC6340

RFC text

Contents
Original text · RFC Editor View on rfc-editor.org ↗ 
Internet Engineering Task Force (IETF)                        R. Presuhn
Request for Comments: 6340                                   Independent
Category: Standards Track                                    August 2011
ISSN: 2070-1721


  Textual Conventions for the Representation of Floating-Point Numbers

Abstract

   This memo defines a Management Information Base (MIB) module
   containing textual conventions (TCs) to represent floating-point
   numbers.

Status of This Memo

   This is an Internet Standards Track document.

   This document is a product of the Internet Engineering Task Force
   (IETF).  It represents the consensus of the IETF community.  It has
   received public review and has been approved for publication by the
   Internet Engineering Steering Group (IESG).  Further information on
   Internet Standards is available in Section 2 of RFC 5741.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   http://www.rfc-editor.org/info/rfc6340.

Copyright Notice

   Copyright (c) 2011 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (http://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.  Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the Simplified BSD License.

Table of Contents

   1. Introduction ....................................................2
   2. The Internet-Standard Management Framework ......................3
   3. Applicability ...................................................3
   4. Structure of the MIB Module .....................................4
      4.1. MIB Modules Required for IMPORTS ...........................4
      4.2. Documents Required for REFERENCE Clauses ...................4
   5. Definitions .....................................................4
   6. Security Considerations .........................................6
   7. IANA Considerations .............................................6
   8. Contributors ....................................................6
   9. References ......................................................7
      9.1. Normative References .......................................7
      9.2. Informative References .....................................7

1.  Introduction

   This memo defines textual conventions for the representation of
   floating-point numbers.  All of these definitions are in terms of the
   IEEE "Standard for Floating-Point Arithmetic", IEEE 754-2008
   [IEEE.754.2008].

   The IEEE "Standard for Floating-Point Arithmetic", IEEE 754-2008
   [IEEE.754.2008], provides for a variety of interchange formats for
   floating-point numbers.  The need for three of these, namely

   o  32-bit,

   o  64-bit,

   o  128-bit,

   has been recognized in network management.  For example, Section
   4.2.3 of the SMIng Objectives [RFC3216] elaborates the need for these
   three floating-point data types in network management protocols.

   The selection of a floating-point format involves many considerations
   and trade-offs.  For an introduction to the fundamentals of floating-
   point representations see Chapter 4 of [KNUTH]; for a discussion of
   these issues specifically with respect to the IEEE formats, see
   [GOLDBERG].

   All of these textual conventions employ the binary interchange format
   defined in [IEEE.754.2008].  Specifically, this means that for all of
   them, the highest-order bit of the first byte is the sign bit, with
   the remaining bits of the octet string corresponding to the exponent
   and fraction parts, in network byte order.

2.  The Internet-Standard Management Framework

   For a detailed overview of the documents that describe the current
   Internet-Standard Management Framework, please refer to section 7 of
   RFC 3410 [RFC3410].

   Managed objects are accessed via a virtual information store, termed
   the Management Information Base or MIB.  MIB objects are generally
   accessed through the Simple Network Management Protocol (SNMP).
   Objects in the MIB are defined using the mechanisms defined in the
   Structure of Management Information (SMI).  This memo specifies a MIB
   module that is compliant to the SMIv2, which is described in STD 58,
   RFC 2578 [RFC2578], STD 58, RFC 2579 [RFC2579] and STD 58, RFC 2580
   [RFC2580].

3.  Applicability

   The following list highlights some of the issues MIB designers need
   to consider when deciding whether to employ these textual
   conventions:

   o  Floating-point numbers are useful if the number space needs to
      cover a large dynamic range.  For number spaces with a limited
      range, fixed-point numbers can be more efficient and more precise.

   o  Floating-point numbers are typically the wrong answer for data
      that is truly decimal or can be handled adequately by re-thinking
      the units and representing the scaled numbers as integers.

   o  The SNMP "lexicographical" ordering for INDEX objects using these
      floating-point textual conventions will simply be that of the
      octet strings corresponding to the floating-point representations,
      which will not always reflect the numerical ordering of the
      corresponding floating-point values.  Even if MIB designers take
      this into account, users might still find the results of a MIB
      "walk" surprising.  Consequently, it is suggested that whenever
      one of these textual conventions is used for an INDEX object, that
      the DESCRIPTION clause should provide some warning.

   o  Embedded systems sometimes lack floating-point support, which can
      complicate the implementation of MIB objects using floating-point
      numbers.

   o  In choosing from among the types defined in this memo, MIB
      designers need to consider both the range and the precision
      needed, as well as recognize that it could be inefficient to use,
      for example, Float128TC when Float64TC would do.

   o  Since these textual conventions are defined in terms of the OCTET
      STRING type, the SMI's mechanisms for formally setting range
      constraints are not available.  MIB designers using these textual
      conventions will need to use DESCRIPTION clauses to spell out any
      applicable range constraints beyond those implied by the
      underlying IEEE types.

   o  Whenever these textual conventions are used in a MIB module, the
      associated DESCRIPTION clause will need to clearly specify whether
      denormalized numbers, NaNs ("not a number") or infinities are
      permitted, along with any special semantics associated with these
      cases.  This is especially important for writeable objects.

4.  Structure of the MIB Module

   This MIB module defines three textual conventions.  It defines no MIB
   objects.

4.1.  MIB Modules Required for IMPORTS

   This MIB module employs definitions from [RFC2578] and [RFC2579].

4.2.  Documents Required for REFERENCE Clauses

   This MIB module contains REFERENCE clauses making reference to IEEE
   754-2008 [IEEE.754.2008].

5.  Definitions

  FLOAT-TC-MIB DEFINITIONS ::= BEGIN

  IMPORTS
      MODULE-IDENTITY,
      mib-2                                 FROM SNMPv2-SMI  -- RFC 2578
      TEXTUAL-CONVENTION                    FROM SNMPv2-TC;  -- RFC 2579

  floatTcMIB    MODULE-IDENTITY
      LAST-UPDATED "201107270000Z"          -- July 27, 2011
      ORGANIZATION "IETF OPSAWG Working Group"
      CONTACT-INFO "WG Email: opsawg@ietf.org

                    Editor: Randy Presuhn
                    randy_presuhn@mindspring.com"

      DESCRIPTION  "Textual conventions for the representation
                    of floating-point numbers.

                    Copyright (c) 2011 IETF Trust and the persons
                    identified as authors of the code.  All rights
                    reserved.

                    Redistribution and use in source and binary forms,
                    with or without modification, is permitted pursuant
                    to, and subject to the license terms contained in,
                    the Simplified BSD License set forth in Section
                    4.c of the IETF Trust's Legal Provisions Relating
                    to IETF Documents
                    (http://trustee.ietf.org/license-info).

                    This version of this MIB module is part of RFC 6340;
                    see the RFC itself for full legal notices."

      REVISION     "201107270000Z"          -- July 27, 2011
      DESCRIPTION  "Initial version, published as RFC 6340."
       ::= { mib-2 201 }

  Float32TC ::= TEXTUAL-CONVENTION
      STATUS       current
      DESCRIPTION  "This type represents a 32-bit (4-octet) IEEE
                    floating-point number in binary interchange format."
      REFERENCE    "IEEE Standard for Floating-Point Arithmetic,
                    Standard 754-2008"
      SYNTAX       OCTET STRING (SIZE(4))


  Float64TC ::= TEXTUAL-CONVENTION
      STATUS       current
      DESCRIPTION  "This type represents a 64-bit (8-octet) IEEE
                    floating-point number in binary interchange format."
      REFERENCE    "IEEE Standard for Floating-Point Arithmetic,
                    Standard 754-2008"
      SYNTAX       OCTET STRING (SIZE(8))


  Float128TC ::= TEXTUAL-CONVENTION
      STATUS       current
      DESCRIPTION  "This type represents a 128-bit (16-octet) IEEE
                    floating-point number in binary interchange format."
      REFERENCE    "IEEE Standard for Floating-Point Arithmetic,
                    Standard 754-2008"
      SYNTAX       OCTET STRING (SIZE(16))

  END

6.  Security Considerations

   This module does not define any management objects.  Instead, it
   defines a set of textual conventions that can be used by other MIB
   modules to define management objects.

   Meaningful security considerations can only be written in the MIB
   modules that define management objects.  Therefore, this memo has no
   impact on the security of the Internet.

7.  IANA Considerations

   The MIB module in this document uses the following IANA-assigned
   OBJECT IDENTIFIER value recorded in the SMI Numbers registry:

      Descriptor        OBJECT IDENTIFIER value
      ----------        -----------------------
      floatTcMIB        { mib-2 201 }

8.  Contributors

   The following people provided helpful comments during the development
   of this document:

   o  Andy Bierman

   o  Martin Duerst

   o  Alfred Hoenes

   o  Juergen Quittek

   o  Juergen Schoenwaeder

   o  Dave Shield

   o  Robert Story

9.  References

9.1.  Normative References

   [IEEE.754.2008]  Institute of Electrical and Electronics Engineers,
                    "Standard for Floating-Point Arithmetic",
                    IEEE Standard 754, August 2008.

   [RFC2578]        McCloghrie, K., Ed., Perkins, D., Ed., and J.
                    Schoenwaelder, Ed., "Structure of Management
                    Information Version 2 (SMIv2)", STD 58, RFC 2578,
                    April 1999.

   [RFC2579]        McCloghrie, K., Ed., Perkins, D., Ed., and J.
                    Schoenwaelder, Ed., "Textual Conventions for SMIv2",
                    STD 58, RFC 2579, April 1999.

   [RFC2580]        McCloghrie, K., Perkins, D., and J. Schoenwaelder,
                    "Conformance Statements for SMIv2", STD 58,
                    RFC 2580, April 1999.

9.2.  Informative References

   [GOLDBERG]       Goldberg, D., "What Every Computer Scientist Should
                    Know About Floating-Point Arithmetic", ACM Computing
                    Surveys Volume 23, Issue 1, March 1991.

   [KNUTH]          Knuth, D., "Seminumerical Algorithms", The Art of
                    Computer Programming (Second Edition) Vol. 2, 1981.

   [RFC3216]        Elliott, C., Harrington, D., Jason, J.,
                    Schoenwaelder, J., Strauss, F., and W. Weiss, "SMIng
                    Objectives", RFC 3216, December 2001.

   [RFC3410]        Case, J., Mundy, R., Partain, D., and B. Stewart,
                    "Introduction and Applicability Statements for
                    Internet-Standard Management Framework", RFC 3410,
                    December 2002.

Author's Address

   Randy Presuhn
   San Jose, CA  95120
   USA

   EMail: randy_presuhn@mindspring.com

AI ELI5

On this page
PROBLEMNetwork management data had no standard way to represent floating-point numbers, even though 32-bit, 64-bit and 128-bit IEEE formats were recognized as needed.
MECHANISMThe memo defines a MIB module with three textual conventions, Float32TC, Float64TC and Float128TC, each an OCTET STRING of 4, 8 or 16 octets holding an IEEE 754-2008 binary interchange value.
WHY IT MATTERSMIB designers can reuse these types instead of inventing their own, though the text warns that floating point is often the wrong choice and that INDEX ordering and range limits need care.

§1Why Floating-Point Types Were Needed

RFC 6340 defines textual conventions (TCs) for representing floating-point numbers in a Management Information Base (MIB) module. All of them are defined in terms of the IEEE Standard for Floating-Point Arithmetic, IEEE 754-2008.

That standard provides a variety of interchange formats. The memo says three of them have been recognized as needed in network management: 32-bit, 64-bit and 128-bit. It points to Section 4.2.3 of the SMIng Objectives (RFC 3216) as an example of that need.

The text also notes that choosing a floating-point format involves many considerations and trade-offs, and refers readers to Knuth and to Goldberg for background on the fundamentals and on the IEEE formats.

People who watch networks sometimes need numbers with decimals. Three sizes of number box were found useful, so this memo makes one standard box of each size.

§2The Management Framework Context

Managed objects live in a virtual information store called the Management Information Base (MIB). MIB objects are generally accessed through the Simple Network Management Protocol (SNMP), and they are defined with the Structure of Management Information (SMI).

This memo specifies a MIB module compliant with SMIv2, described in STD 58 (RFC 2578, RFC 2579 and RFC 2580). For an overview of the whole framework it points to RFC 3410.

Network devices keep a big list of facts about themselves, and a standard language describes what each fact looks like. This memo adds a few new kinds of fact to that language.

§3The Three Textual Conventions

The module, FLOAT-TC-MIB, defines three textual conventions and no MIB objects. Each one is an OCTET STRING of a fixed size:

  • Float32TC: a 32-bit (4-octet) IEEE floating-point number.
  • Float64TC: a 64-bit (8-octet) IEEE floating-point number.
  • Float128TC: a 128-bit (16-octet) IEEE floating-point number.

All three use the binary interchange format of IEEE 754-2008. The highest-order bit of the first byte is the sign bit, and the remaining bits of the octet string hold the exponent and fraction parts, in network byte order. The module imports definitions from RFC 2578 and RFC 2579, and its REFERENCE clauses point to IEEE 754-2008.

There are three boxes: a 4-byte box, an 8-byte box and a 16-byte box. The first bit in each says plus or minus, and the rest holds the number's digits in a fixed order.

THE THREE TEXTUAL CONVENTIONS (RFC 6340)
NameWidthOctets
Float32TC32-bit4
Float64TC64-bit8
Float128TC128-bit16

§4Applicability and Cautions

Section 3 gives MIB designers a checklist of issues to weigh before using these types:

  • Floating point suits a large dynamic range. For a limited range, fixed-point numbers can be more efficient and more precise, and truly decimal data is often better handled by changing units and using scaled integers.
  • Used as an INDEX, the SNMP lexicographic ordering follows the octet strings, which will not always match numeric order. A walk of the table may surprise users, so the DESCRIPTION clause should carry a warning.
  • Embedded systems sometimes lack floating-point support, which complicates implementation.
  • Pick the type for the range and precision needed. Using Float128TC where Float64TC would do could be inefficient.
  • Because the types are OCTET STRING, the SMI cannot set formal range constraints. Designers must state any extra range limits in DESCRIPTION clauses.
  • The DESCRIPTION must say whether denormalized numbers, NaNs ("not a number") or infinities are permitted and what they mean. This matters most for writeable objects.

The memo says: do not use a decimal box if a whole number will do, and do not pick the biggest box if a smaller one is enough. Also, sorting these boxes by their bytes may not put the numbers in the right order.

§5Security and IANA Notes

The module defines no management objects, only conventions that other MIB modules can use. The memo therefore says meaningful security considerations can only be written in the MIB modules that define objects, and that it has no impact on the security of the Internet.

For IANA, the module's identifier floatTcMIB is recorded in the SMI Numbers registry as { mib-2 201 }.

These are just labels for boxes, so they cannot be dangerous on their own. Whoever uses a box in a real list has to think about safety there.

§6References

[IEEE.754.2008]Institute of Electrical and Electronics Engineers, "Standard for Floating-Point Arithmetic", IEEE Standard 754, August 2008.
[RFC2578]RFC 2578McCloghrie, K., Ed., Perkins, D., Ed., and J. Schoenwaelder, Ed., "Structure of Management Information Version 2 (SMIv2)", STD 58, RFC 2578, April 1999.April 1999[RFC2579]RFC 2579McCloghrie, K., Ed., Perkins, D., Ed., and J. Schoenwaelder, Ed., "Textual Conventions for SMIv2", STD 58, RFC 2579, April 1999.April 1999[RFC2580]RFC 2580McCloghrie, K., Perkins, D., and J. Schoenwaelder, "Conformance Statements for SMIv2", STD 58, RFC 2580, April 1999.April 1999[RFC3216]RFC 3216Elliott, C., et al., "SMIng Objectives", RFC 3216, December 2001.December 2001[RFC3410]RFC 3410Case, J., Mundy, R., Partain, D., and B. Stewart, "Introduction and Applicability Statements for Internet-Standard Management Framework", RFC 3410, December 2002.December 2002