Network Working Group K. TsUChiya
Requests for Comments: 2767 H. Higuchi
Category: Informational Y. Atarashi
Hitachi
February 2000

Dual Stack Hosts using the "Bump-In-the-Stack" Technique (BIS)

Status of this Memo

This memo provides information for the Internet community. It does
not specify an Internet standard of any kind. Distribution of this
memo is unlimited.

Copyright Notice

Copyright (C) The Internet Society (2000). All Rights Reserved.

Abstract

In the especially initial stage of the transition from IPv4 to IPv6,
it is hard to provide a complete set of IPv6 applications. This memo
proposes a mechanism of dual stack hosts using the technique called
"Bump-in-the-Stack" in the IP security area. The mechanism allows the
hosts to communicate with other IPv6 hosts using existing IPv4
applications.

1. Introduction

RFC1933 [TRANS-MECH] specifies transition mechanisms, including dual
stack and tunneling, for the initial stage. Hosts and routers with
the transition mechanisms are also developed. But there are few
applications for IPv6 [IPV6] as compared with IPv4 [IPV4] in which a
great number of applications are available. In order to advance the
transition smoothly, it is highly desirable to make the availability
of IPv6 applications increase to the same level as IPv4.
Unfortunately, however, this is eXPected to take a very long time.

This memo proposes a mechanism of dual stack hosts using the
technique called "Bump-in-the-Stack" [BUMP] in the IP security area.
The technique inserts modules, which snoop data flowing between a
TCP/IPv4 module and network card driver modules and translate IPv4
into IPv6 and vice versa, into the hosts, and makes them self-
translators. When they communicate with the other IPv6 hosts, pooled
IPv4 addresses are assigned to the IPv6 hosts internally, but the
IPv4 addresses never flow out from them. Moreover, since the
assignment is automatically carried out using DNS protocol, users do

not need to know whether target hosts are IPv6 ones. That is, this
allows them to communicate with other IPv6 hosts using existing IPv4
applications; thus it seems as if they were dual stack hosts with
applications for both IPv4 and IPv6. So they can expand the territory
of dual stack hosts. Furthermore they can co-exist with other
translators because their roles are different.

This memo uses the Words defined in [IPV4], [IPV6], and [TRANS-MECH].

2. Components

Dual stack hosts defined in RFC1933 [TRANS-MECH] need applications,
TCP/IP modules and addresses for both IPv4 and IPv6. The proposed
hosts in this memo have 3 modules instead of IPv6 applications, and
communicate with other IPv6 hosts using IPv4 applications. They are a
translator, an extension name resolver and an address mapper.

Figure 1 illustrates the structure of the host in which they are
installed.

+----------------------------------------------------------+
+----------------------------------------------------+
IPv4 applications
+----------------------------------------------------+
+----------------------------------------------------+
TCP/IPv4
+-------------------------------------------+
+-----------+ +---------+ +------------+
extension address translator
name mapper +------------+
resolver +------------+
IPv6
+--------+ +-----------+ +---------+ +------------+
+----------------------------------------------------+
Network card drivers
+----------------------------------------------------+
+----------------------------------------------------------+
+----------------------------------------------------------+
Network cards
+----------------------------------------------------------+

Figure. 1 Structure of the proposed dual stack host

2.1 Translator

It translates IPv4 into IPv6 and vice versa using the IP conversion
mechanism defined in [SIIT].

When receiving IPv4 packets from IPv4 applications, it converts IPv4
packet headers into IPv6 packet headers, then fragments the IPv6
packets (because header length of IPv6 is typically 20 bytes larger
than that of IPv4), and sends them to IPv6 networks. When receiving
IPv6 packets from the IPv6 networks, it works symmetrically to the
previous case, except that there is no need to fragment the packets.

2.2 Extension Name Resolver

It returns a "proper" answer in response to the IPv4 application's
request.

The application typically sends a query to a name server to resolve
'A' records for the target host name. It snoops the query, then
creates another query to resolve both 'A' and 'AAAA' records for the
host name, and sends the query to the server. If the 'A' record is
resolved, it returns the 'A' record to the application as is. In the
case, there is no need for the IP conversion by the translator. If
only the 'AAAA' record is available, it requests the mapper to assign
an IPv4 address corresponding to the IPv6 address, then creates the
'A' record for the assigned IPv4 address, and returns the 'A' record
to the application.

NOTE: This action is similar to that of the DNS ALG (Application
Layer Gateway) used in [NAT-PT]. See also [NAT-PT].

2.3 Address mapper

It maintains an IPv4 address spool. The spool, for example, consists
of private addresses [PRIVATE]. Also, it maintains a table which
consists of pairs of an IPv4 address and an IPv6 address.

When the resolver or the translator requests it to assign an IPv4
address corresponding to an IPv6 address, it selects and returns an
IPv4 address out of the spool, and registers a new entry into the
table dynamically. The registration occurs in the following 2 cases:

(1) When the resolver gets only an 'AAAA' record for the target host
name and there is not a mapping entry for the IPv6 address.
(2) When the translator receives an IPv6 packet and there is not a
mapping entry for the IPv6 source address.

NOTE: There is only one exception. When initializing the table, it
registers a pair of its own IPv4 address and IPv6 address into the
table statically.

3. Action Examples

This section describes action of the proposed dual stack host called
"dual stack," which communicates with an IPv6 host called "host6"
using an IPv4 application.

3.1 Originator behavior

This subsection describes the originator behavior of "dual stack."
The communication is triggered by "dual stack."

The application sends a query to its name server to resolve 'A'
records for "host6."

The resolver snoops the query, then creates another query to resolve
both 'A' and 'AAAA' records for the host name, and sends it to the
server. In this case, only the 'AAAA' record is resolved, so the
resolver requests the mapper to assign an IPv4 address corresponding
to the IPv6 address.

NOTE: In the case of communication with an IPv4 host, the 'A' record
is resolved and then the resolver returns it to the application as
is. There is no need for the IP conversion as shown later.

The mapper selects an IPv4 address out of the spool and returns it to
the resolver.

The resolver creates the 'A' record for the assigned IPv4 address and
returns it to the application.

NOTE: See subsection 4.3 about the influence on other hosts caused by
an IPv4 address assigned here.

The application sends an IPv4 packet to "host6."

The IPv4 packet reaches the translator. The translator tries to
translate the IPv4 packet into an IPv6 packet but does not know how
to translate the IPv4 destination address and the IPv4 source
address. So the translator requests the mapper to provide mapping
entries for them.

The mapper checks its mapping table and finds entries for each of
them, and then returns the IPv6 destination address and the IPv6
source address to the translator.

NOTE: The mapper will register its own IPv4 address and IPv6 address
into the table beforehand. See subsection 2.3.

The translator translates the IPv4 packet into an IPv6 packet then
fragments the IPv6 packet if necessary and sends it to an IPv6
network.

The IPv6 packet reaches "host6." Then "host6" sends a new IPv6 packet
to "dual stack."

The IPv6 packet reaches the translator in "dual stack."

The translator gets mapping entries for the IPv6 destination address
and the IPv6 source address from the mapper in the same way as
before.

Then the translator translates the IPv6 packet into an IPv4 packet
and tosses it up to the application.

The following diagram illustrates the action described above:

"dual stack" "host6"
IPv4 TCP/ extension address translator IPv6
appli- IPv4 name mapper
cation resolver

<<Resolve an IPv4 address for "host6".>>

------------> Query of 'A' records for "host6". Name
Server
--------------------------------------->
Query of 'A' records and 'AAAA' for "host6"

<---------------------------------------
Reply only with 'AAAA' record.

<<Only 'AAAA' record is resolved.>>

--------> Request one IPv4 address
corresponding to the IPv6 address.

<<Assign one IPv4 address.>>

<-------- Reply with the IPv4 address.

<<Create 'A' record for the IPv4 address.>>

<------------ Reply with the 'A' record.


Figure 2 Action of the originator (1/2)

"dual stack" "host6"
IPv4 TCP/ extension address translator IPv6
appli- IPv4 name mapper
cation resolver

<<Send an IPv4 packet to "host6".>>

============================> An IPv4 packet.

<------ Request IPv6 addresses
corresponding to the IPv4
addresses.

------> Reply with the IPv6
addresses.

<<Translate IPv4 into IPv6.>>

An IPv6 packet. ===================>

<<Reply an IPv6 packet to
"dual stack".>>

An IPv6 packet. <===================

<<Translate IPv6 into IPv4.>>

<============================ An IPv4 packet.


Figure 2 Action of the originator (2/2)

3.2 Recipient behavior

This subsection describes the recipient behavior of "dual stack."
The communication is triggered by "host6."

"host6" resolves the 'AAAA' record for "dual stack" through its name
server, and then sends an IPv6 packet to the IPv6 address.

The IPv6 packet reaches the translator in "dual stack."

The translator tries to translate the IPv6 packet into an IPv4 packet
but does not know how to translate the IPv6 destination address and
the IPv6 source address. So the translator requests the mapper to
provide mapping entries for them.

The mapper checks its mapping table with each of them and finds a
mapping entry for the IPv6 destination address.

NOTE: The mapper will register its own IPv4 address and IPv6 address
into the table beforehand. See subsection 2.3.

But there is not a mapping entry for the IPv6 source address, so the
mapper selects an IPv4 address out of the spool for it, and then
returns the IPv4 destination address and the IPv4 source address to
the translator.

NOTE: See subsection 4.3 about the influence on other hosts caused by
an IPv4 address assigned here.

The translator translates the IPv6 packet into an IPv4 packet and
tosses it up to the application.

The application sends a new IPv4 packet to "host6."

The following behavior is the same as that described in subsection
3.1.

The following diagram illustrates the action described above:

"dual stack" "host6"
IPv4 TCP/ extension address translator IPv6
appli- IPv4 name mapper
cation resolver

<<Receive data from "host6".>>

An IPv6 packet. <===================

<------ Request IPv4 addresses
corresponding to the IPv6
addresses.

------> Reply with the IPv4
addresses.

<<Translate IPv6 into IPv4.>>

<============================ An IPv4 packet.

<<Reply an IPv4 packet to "host6".>>

============================> An IPv4 packet.

<<Translate IPv4 into IPv6.>>

An IPv6 packet. ===================>


Figure 3 Action of the recipient

4. Considerations

This section considers some issues of the proposed dual stack hosts.

4.1 IP conversion

In common with NAT [NAT], IP conversion needs to translate IP
addresses embedded in application layer protocols, which are
typically found in FTP [FTP]. So it is hard to translate all such
applications completely.

4.2 IPv4 address spool and mapping table

The spool, for example, consists of private addresses [PRIVATE]. So a
large address space can be used for the spool. Nonetheless, IPv4

addresses in the spool will be exhausted and cannot be assigned to
IPv6 target hosts, if the host communicates with a great number of
other IPv6 hosts and the mapper never frees entries registered into
the mapping table once. To solve the problem, for example, it is
desirable for the mapper to free the oldest entry in the mapping
table and re-use the IPv4 address for creating a new entry.

4.3 Internally assigned IPv4 addresses

IPv4 addresses, which are internally assigned to IPv6 target hosts
out of the spool, never flow out from the host, and so do not
negatively affect other hosts.

5. Applicability and Limitations

This section considers applicability and limitations of the proposed
dual stack hosts.

5.1 Applicability

The mechanism can be useful for users in the especially initial stage
where some applications not modified into IPv6 remain. And it can
also help users who cannot upgrade their certain applications for
some reason after all applications have been modified. The reason is
that it allows hosts to communicate with IPv6 hosts using existing
IPv4 applications, and that they can get connectivity for both IPv4
and IPv6 even if they do not have IPv6 applications as a result.

Note that it can also work in conjunction with a complete IPv6 stack.
They can communicate with both IPv4 hosts and IPv6 hosts using IPv4
applications via the mechanism, and can also communicate with IPv6
hosts using IPv6 applications via the complete IPv6 stack.

5.2 Limitations

The mechanism is valid only for unicast communication, but invalid
for multicast communication. Multicast communication needs another
mechanism.

It allows hosts to communicate with IPv6 hosts using existing IPv4
applications, but this can not be applied to IPv4 applications which
use any IPv4 option since it is impossible to translate IPv4 options
into IPv6. Similarly it is impossible to translate any IPv6 option
headers into IPv4, except for fragment headers and routing headers.
So IPv6 inbound communication having the option headers may be
rejected.

In common with NAT [NAT], IP conversion needs to translate IP
addresses embedded in application layer protocols, which are
typically found in FTP [FTP]. So it is hard to translate all such
applications completely.

It may be impossible that the hosts using the mechanism utilize the
security above network layer since the data may carry IP addresses.

Finally it can not combine with secure DNS since the extension name
resolver can not handle the protocol.

6. Security Considerations

This section considers security of the proposed dual stack hosts.

The hosts can utilize the security of all layers like ordinary IPv4
communication when they communicate with IPv4 hosts using IPv4
applications via the mechanism. Likewise they can utilize the
security of all layers like ordinary IPv6 communication when they
communicate with IPv6 hosts using IPv6 applications via the complete
IPv6 stack. However, unfortunately, they can not utilize the security
above network layer when they communicate with IPv6 hosts using IPv4
applications via the mechanism. The reason is that when the protocol
data with which IP addresses are embedded is encrypted, or when the
protocol data is encrypted using IP addresses as keys, it is
impossible for the mechanism to translate the IPv4 data into IPv6 and
vice versa. Therefore it is highly desirable to upgrade to the
applications modified into IPv6 for utilizing the security at
communication with IPv6 hosts.

7. References

[SIIT] Nordmark, E., "Stateless IP/ICMP Translator (SIIT)", RFC
2765, February 2000.

[IPV4] Postel, J., "Internet Protocol", STD 5, RFC791,
September 1981.

[FTP] Postel, J. and J. Reynolds, "File Transfer Protocol",
STD 9, RFC959, October 1985.

[NAT] Kjeld B. and P. Francis, "The IP Network Address
Translator (NAT)", RFC1631, May 1994.

[IPV6] Deering, S. and R. Hinden, "Internet Protocol, Version 6
(IPv6) Specification", RFC2460, December 1998.

[PRIVATE] Rekhter, Y., Moskowitz, B., Karrenberg, D., de Groot, G.
J. and E. Lear, "Address Allocation for Private
Internets", BCP 5, RFC1918, February 1996.

[TRANS-MECH] Gilligan, R. and E. Nordmark, "Transition Mechanisms for
IPv6 Hosts and Routers", RFC1933, April 1996.

[BUMP] D.A. Wagner and S.M. Bellovin, "A Bump in the Stack
Encryptor for MS-DOS Systems", The 1996 Symposium on
Network and Distributed Systems Security (SNDSS'96)
Proceedings.

[NAT-PT] Tsirtsis, G. and P. Srisuresh, "Network Address
Translation - Protocol Translation (NAT-PT)", RFC2766,
February 2000.

8. Acknowledgements

The authors gratefully acknowledge the many helpful suggestions of
the members of the WIDE Project, Kazuhiko YAMAMOTO, Jun MURAI,
Munechika SUMIKAWA, Ken WATANABE, and Takahisa MIYAMOTO, at large.

9. Authors' Addresses

Kazuaki TSUCHIYA
Enterprise Server Division, Hitachi, Ltd.
810 Shimoimaizumi, Ebina-shi, Kanagawa-ken, 243-0435 JAPAN

Phone: +81-462-32-2121
Fax: +81-462-35-8324
EMail: tsuchi@ebina.hitachi.co.jp

Hidemitsu HIGUCHI
Enterprise Server Division, Hitachi, Ltd.
810 Shimoimaizumi, Ebina-shi, Kanagawa-ken, 243-0435 JAPAN

Phone: +81-462-32-2121
Fax: +81-462-35-8324
EMail: h-higuti@ebina.hitachi.co.jp

Yoshifumi ATARASHI
Enterprise Server Division, Hitachi, Ltd.
810 Shimoimaizumi, Ebina-shi, Kanagawa-ken, 243-0435 JAPAN

Phone: +81-462-32-2121
Fax: +81-462-35-8324
EMail: atarashi@ebina.hitachi.co.jp

10. Full Copyright Statement

Copyright (C) The Internet Society (2000). All Rights Reserved.

This document and translations of it may be copied and furnished to
others, and derivative works that comment on or otherwise explain it
or assist in its implementation may be prepared, copied, published
and distributed, in whole or in part, without restriction of any
kind, provided that the above copyright notice and this paragraph are
included on all such copies and derivative works. However, this
document itself may not be modified in any way, such as by removing
the copyright notice or references to the Internet Society or other
Internet organizations, except as needed for the purpose of
developing Internet standards in which case the procedures for
copyrights defined in the Internet Standards process must be
followed, or as required to translate it into languages other than
English.

The limited permissions granted above are perpetual and will not be
revoked by the Internet Society or its successors or assigns.

This document and the information contained herein is provided on an
"AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING
BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION
HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Acknowledgement

Funding for the RFCEditor function is currently provided by the
Internet Society.