Network Working Group M. Gaynor
Request for Comments: 3093 S. Bradner
Category: Informational Harvard University
1 April 2001
Firewall Enhancement Protocol (FEP)
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 (2001). All Rights Reserved.
Abstract
Internet Transparency via the end-to-end architecture of the Internet
has allowed vast innovation of new technologies and services [1].
However, recent developments in Firewall technology have altered this
model and have been shown to inhibit innovation. We propose the
Firewall Enhancement Protocol (FEP) to allow innovation, without
violating the security model of a Firewall. With no cooperation from
a firewall operator, the FEP allows ANY application to traverse a
Firewall. Our methodology is to layer any application layer
Transmission Control Protocol/User Datagram Protocol (TCP/UDP)
packets over the HyperText Transfer Protocol (HTTP) protocol, since
HTTP packets are typically able to transit Firewalls. This scheme
does not violate the actual security usefulness of a Firewall, since
Firewalls are designed to thwart attacks from the outside and to
ignore threats from within. The use of FEP is compatible with the
current Firewall security model because it requires cooperation from
a host inside the Firewall. FEP allows the best of both worlds: the
security of a firewall, and transparent tunneling thought the
firewall.
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in RFC 2119.
Gaynor & Bradner Informational [Page 1]
RFC 3093 Firewall Enhancement Protocol 1 April 2001
The Internet has done well, considering that less than 10 years ago
the telco's were claiming it could not ever work for the corporate
environment. There are many reasons for this; a particularly strong
one is the end-to-end argument discussed by Reed, Seltzer, and Clark
[2]. Innovation at the ends has proven to be a very powerful
methodology creating more value than ever conceived of. But, the
world is changing as Clark notes in [6]. With the connection of the
corporate world to the Internet, security concerns have become
paramount, even at the expense of breaking the end-to-end paradigm.
One example of this is the Firewall - a device to prevent outsiders
from unauthorized access into a corporation. Our new protocol, the
Firewall Enhancement Protocol (FEP), is designed to restore the end-
to-end model while maintaining the level of security created by
Firewalls.
To see how powerful the end-to-end model is consider the following
example. If Scott and Mark have a good idea and some implementation
talent, they can create an artifact, use it, and send it to their
friends. If it turns out to be a good idea these friends can adopt
it and maybe make it better. Now enter the Firewall: if Mark happens
to work at a company that installs a Firewall, he can't experiment
with his friend Scott. Innovation is more difficult, maybe
impossible. What business is it of an IT manager if Scott and Mark
want to do some experiments to enable them to better serve their
users? This is how the web was created: one guy with talent, a few
good ideas, and the ability to innovate.
Firewalls are important, and we do respect the right of anybody to
protecting themselves any way they want (as long as others are not
inconvenienced). Firewalls work, and have a place in the Internet.
However, Firewalls are built to protect from external threats, not
internal ones. Our proposed protocol does not break the security
model of the Firewall; it still protects against all external risks
that a particular Firewall can protect against. For our protocol to
work someone inside the Firewall must run an application level
protocol that can access TCP port 80. Our concept allows a
consistent level of security while bypassing the IT manager in charge
of the Firewall. We offer freedom to innovate without additionally
compromising external security, and the best part, no need to waste
time involving any managers for approval.
We got this idea from the increasing number of applications that use
HTTP specifically because it can bypass Firewall barriers. This
piecemeal deployment of specific applications is not an efficient way
to meet the challenge to innovation created by Firewalls. We decided
to develop a process by which TCP/IP itself is carried over HTTP.
Gaynor & Bradner Informational [Page 2]
RFC 3093 Firewall Enhancement Protocol 1 April 2001
With this innovation anyone can use any new TCP/IP application
immediately without having to go through the laborious process of
dealing with Firewall access for the particular application. An
unintended byproduct of this proposal is that existing TCP/IP
applications can also be supported to better serve the users. With
FEP, the users can decide what applications they can run.
Our protocol is simple and is partly based on the Eastlake [3]
proposal for MIME encoding of IP packets. We use the ubiquitous HTTP
protocol format. The IP datagram is carried in the message body of
the HTTP message and the TCP packet header information is encoded
into HTTP headers of the message. This ASCII encoding of the header
fields has many advantages, including human readability, increasing
the debuggability of new applications, and easy logging of packet
information. If this becomes widely adopted, tools like tcpdump will
become obsolete.
Figure 1 shows a high level view of our protocol. The application
(1) in host A (outside the Firewall) sends a TCP/IP datagram to host
B (within the firewall). Using a tunnel interface the TCP/IP
datagram is routed to our FEP software (2), which encodes the
datagram within a HTTP message. Then this message is sent via a
HTTP/TCP/IP tunnel (3) to host B on the normal HTTP port (4). When
it arrives at host B, this packet is routed via the tunnel to the FEP
software (5), which decodes the packet and creates a TCP/IP datagram
to insert into host's B protocol stack (6). This packet is routed to
the application on host B (7), as if the Firewall (8) never existed.
Gaynor & Bradner Informational [Page 3]
RFC 3093 Firewall Enhancement Protocol 1 April 2001
host A host B
---------- ----------
| App | (1) | App | (7)
|----------| |----------|
| TCP | | TCP |
|----------| |----------|
| IP | | IP | (6)
|----------| |----------|
| FEP dvr | (2) | FEP dvr | (5)
|----------| |----------|
| TCP | | TCP |
|----------| |----------|
| IP | Firewall (8) | IP |
---------- --- -----------
| (3) | | ^ (4)
+---------------->| |-----------------------+
| |
| |
---
Figure 1
FEP allows either side to look like a client or server. Each TCP/IP
packet is sent as either a HTTP GET request or a response to a GET
request. This flexibility work well with firewalls that try to
verify valid HTTP commands crossing the Firewall stopping the
unwanted intercepting of FEP packets.
The TCP/IP packet is encoded into the HTTP command in two (or
optionally three) steps. First, the IP packet is encoded as the
message body in MIME format, as specified in [3]. Next, the TCP [4]
packet header is parsed and encoded into new HTTP headers. Finally,
as an option, the IP header can also be encoded into new optional
HTTP headers. Encoding the TCP and optionally the IP header is
strictly for human readability, since the entire IP datagram is
encoded in the body part of the HTTP command.
This proposal defines the following new HTTP headers for representing
TCP header information.
TCP_value_opt - This ASCII string represents the encoding type for
the TCP fields where a mandatory encoding type is not specified.
The legitimate values are:
Gaynor & Bradner Informational [Page 4]
RFC 3093 Firewall Enhancement Protocol 1 April 2001
TCP_binary - ASCII representation of the binary representation of the
value of the field.
TCP_hexed - ASCII representation of the hex representation of the
value of the field.
TCP_Sport - The 16-bit TCP Source Port number, encoded as an ASCII
string representing the value of port number.
TCP_Dport - The 16-bit TCP Destination Port number, encoded as an
ASCII string representing the value of the port number.
TCP_SeqNum - The 32-bit Sequence Number, encoded as an ASCII string
representing the hex value of the Sequence number. This field
MUST be sent as lower case because it is not urgent.
TCP_Ackl - The 32-bit Acknowledgement Number, encoded as ASCII string
representing the value of the Acknowledgement number.
TCP_DODO - The 4-bit Data Offset value, encoded as an ASCII string
representing the base 32 value of the actual length of TCP header
in bits. (Normally this is the Data value times 32.)
TCP_6Os - The 6 reserved bits, encoded as a string of 6 ASCII
characters. A "O" ("Oh") represents an "Off" bit and "O" ("Oh")
represents an "On" bit. (Note these characters MUST all be sent
as "off" and MUST be ignored on receipt.)
TCP_FlgBts - The TCP Flags, encoded as the set of 5 comma-separated
ASCII strings: [{URG|urg}, {ACK|ack}, {PSH|psh}, {RST|rst},
{SYN|syn}, {FIN|fin}]. Capital letters imply the flag is set,
lowercase means the flag is not set.
TCP_Windex - The 16-bit TCP Window Size, encoded as an ASCII string
representing the value of the number of bytes in the window.
TCP_Checkit - The 16-bit TCP Checksum field, encoded as an ASCII
string representing the decimal value of the ones-complement of
the checksum field.
TCP_UP - The 16-bit TCP Urgent Pointer, encoded as the hex
representation of the value of the field. The hex string MUST be
capitalized since it is urgent.
Gaynor & Bradner Informational [Page 5]
RFC 3093 Firewall Enhancement Protocol 1 April 2001
TCP_Opp_Lst - A comma-separated list of any TCP options that may be
present. Each option is encoded as an ASCII string representing
the name of the option followed by option-specific information
enclosed in square brackets. Representative options and their
encoding follow, other IP options follow the same form:
End of Options option: ["End of Options"]
Window scale option: ["Window scale", shift_count], where
shift_count is the window scaling factor represented as the
ASCII string in decimal.
This proposal defines the following new HTTP headers for representing
IPv4 header information:
These optional headers are used to encode the IPv4 [5] header for
better readability. These fields are encoded in a manner similar to
the above TCP header fields.
Since the base IP packet is already present in an HTTP header, the
following headers are optional. None, some or all of them may be
used depending on the whim of the programmer.
IP_value_opt - This ASCII string represents the encoding type for the
following fields where a mandatory encoding type is not
specified. The legitimate values are the same as for
TCP_value_opt.
IP_Ver - The IP Version number, encoded as an UTF-8 string. The
legitimate values for the string are "four", "five", and "six."
The encapsulation of the fields in the IP header are defined in
this section if the value is "four", and in section 3.3 if the
value is "six". Encapsulations for headers with IP_Ver value of
"five" will be developed if the right orders are received.
Encapsulations for headers with the IP_Ver value of "eight" are
empty. Implementations MUST be able to support arbitrary native
languages for these strings.
IP4_Hlen - The IP Internet Header Length field, it is encoded in the
same way as TCP_DODO.
IP4_Type_of_Service (this name is case sensitive) - This is an
obsolete name for a field in the IPv4 header, which has been
replaced with IP_$$ and IP_CU.
Gaynor & Bradner Informational [Page 6]
RFC 3093 Firewall Enhancement Protocol 1 April 2001
IP_$$ - The 6-bit Differentiated Services field, encapsulated as an
UTF-8 string representing the name of the DS codepoint in the
field.
IP_CU - The 2-bit field that was the two low-order bits of the TOS
field. Since this field is currently being used for experiments
it has to be coded in the most general way possible, thus it is
encoded as two ASCII strings of the form "bit0=X" and "bit1=X,"
where "X" is "on" or "off." Note that bit 0 is the MSB.
IP4_Total - The 16-bit Total Length field, encoded as an ASCII string
representing the value of the field.
IP4_SSN - The IP Identification field, encoded as an ASCII string
representing the value of the field.
IP4_Flags - The IP Flags, encoded as the set of 3 comma separated
ASCII strings: [{"Must Be Zero"}, {"May Fragment"|"Don't
Fragment"}, {"Last Fragment"|"More Fragments"}]
IP4_Frager - The 13-bit Fragment Offset field, encoded as an ASCII
string representing the value of the field.
IP4_TTL - The 8-bit Time-to-Live field, encoded as an UTF-8 string of
the form "X hops to destruction." Where "X" is the decimal value
-1 of the field. Implementations MUST be able to support
arbitrary languages for this string.
IP4_Proto - The 8-bit Protocol field, encoded as an UTF-8 string
representing the common name for the protocol whose header
follows the IP header.
IP4_Checkit - The 16-bit Checksum field, encoded in the same way as
TCP_Checkit.
IP4_Apparent_Source - The 32-bit Source Address field. For user
friendliness this is encoded as an UTF-8 string representing the
domain name of the apparent sender of the packet. An alternate
form, to be used when the domain name itself might be blocked by a
firewall programmed to protect the innocence of the corporate
users, is an ASCII string representing the dotted quad form of the
IPv4 address.
IP4_Dest_Addr - The 32-bit Destination Address field, encoded in the
same way as is IP4_Apparent_Source.
Gaynor & Bradner Informational [Page 7]
RFC 3093 Firewall Enhancement Protocol 1 April 2001
IP4_Opp_Lst - A comma-separated list of all IPv4 options that are
present. Each option is encoded as an ASCII string representing
the name of the option followed by option-specific information
enclosed in square brackets. Representative options and their
encoding follow, other IP options follow the same form:
End of Options option: ["End of Options"]
Loose Source Routing option: ["Loose Source Routing", length,
pointer, IP4_addr1, IP4_addr2, ...], where length and pointer
are ASCII strings representing the value of those fields.
This proposal defines the following new HTTP headers for representing
IPv6 header information:
These optional headers encode the IPv6 [5] header for better
readability. These fields are encoded in a manner similar to the
above TCP header fields.
Since the base IP packet is already present in an HTTP header the
following headers are optional. None, some or all of them may be
used depending on the whim of the programmer. At this time only the
base IPv6 header is supported. If there is sufficient interest,
support will be developed for IPv6 extension headers.
IP_$$ - the 6-bit Differentiated Services field - see above
IP_CU - the 2-bit unused field - see above
IP6_Go_with_the_Flow - The 20-bit Flow Label field. Since this field
is not currently in use it should be encoded as the UTF-8 string
"do not care".
IP6_PayLd - The 16-bit Payload Length field, encoded as an ASCII
string representing the value of the field. The use of FEP with
IPv6 jumbograms is not recommended.
IP6_NxtHdr - The 8-bit Next Header field, encoded in the same way as
IP4_Proto.
IP6_Hopping - The 8-bit Hop Limit field, encoded in the same way as
IP4_TTL.
Gaynor & Bradner Informational [Page 8]
RFC 3093 Firewall Enhancement Protocol 1 April 2001
IP6_Apparent_Source - The 128-bit Source Address field. For user
friendliness, this is encoded as an UTF-8 string representing the
domain name of the apparent sender of the packet. An alternate
form, to be used when the domain name itself might be blocked by a
Firewall programmed to protect the innocence of the corporate
users, is an ASCII string representing any one of the legitimate
forms of representing an IPv6 address.
IP6_Dest_Addr - The 128-bit Destination Address field, encoded the
same way as IP6_Apparent_Source.
We wish to thank the many Firewall vendors who have supported our
work to re-enable the innovation that made the Internet great,
without giving up the cellophane fig leaf of security that a Firewall
provides.
Mark Gaynor
Harvard University
Cambridge MA 02138
EMail gaynor@eecs.harvard.edu
Scott Bradner
Harvard University
Cambridge MA 02138
Phone +1 617 495 3864
EMail sob@harvard.edu
Gaynor & Bradner Informational [Page 9]
RFC 3093 Firewall Enhancement Protocol 1 April 2001
References
[1] Carpenter, B., "Internet Transparency", RFC 2775, February 2000.
[2] Saltzer, J., Reed, D., and D. Clark, "End-to-End Arguments in
System Design". 2nd International Conference on Distributed
Systems, Paris, France, April 1981.
[3] Eastlake, D., "IP over MIME", Work in Progress.
[4] Postel, J., "Transmission Control Protocol", STD 7, RFC 793,
September 1981.
[5] Postel, J., "Internet Protocol", STD 5, RFC 791, September 1981.
[6] Clark, D. and M. Blumenthal, "Rethinking the Design of the
Internet: The end-to-end argument vs. the brave new world". 2000.
Gaynor & Bradner Informational [Page 10]
RFC 3093 Firewall Enhancement Protocol 1 April 2001
Full Copyright Statement
Copyright (C) The Internet Society (2001). 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 RFC Editor function is currently provided by the
Internet Society.
Gaynor & Bradner Informational [Page 11]