Do you have a question? Post it now! No Registration Necessary. Now with pictures!
- Richard E. Silverman
December 3, 2005, 7:38 am
rate this thread
SSH Communications Security (SCS, at ssh.com) recently made the following
It contains this paragraph:
"Both SSH Tectia and OpenSSH are based on the Secure Shell version 2 (SSH2)
protocol specifications, originally developed by SSH Communications
Security and standardized by the IETF. However, OpenSSH deviates from the
standards in its SCP (Secure Copy Protocol) implementation. SSH Tectia
Client and Server now incorporate a compatibility mode for OpenSSH SCP,
which still uses the old Secure Shell version 1 (SSH1). In addition, the
new SSH Tectia product versions will support the OpenSSH public-key file
format, eliminating the need for manual key conversions."
This is FUD, pure and simple, adulterated only by some outright errors.
Let's deconstruct the text to see just how wrong it is.
"Both SSH Tectia and OpenSSH are based on the Secure Shell version 2
(SSH2) protocol specifications, originally developed by SSH
Communications Security and standardized by the IETF. "
True as far as it goes, but it would be more accurate to note that the
IETF SSH working group has made substantial modifications, additions, and
improvements to the protocol specs. The IETF did not just accept and
"standardize" SCS's work as given.
"However, OpenSSH deviates from the standards in its SCP (Secure Copy
This is a disturbing sentence. First let's dispose of the outright error:
"scp" does not stand for "secure copy protocol." "scp" is not even the
name of a protocol; it's the name of a program. "scp" is a play on "rcp,"
meaning a secure version of the venerable Unix remote file-copying program
-- just as "ssh" is a secure counterpart to "rsh." For the company
started by the man who wrote the original ssh programs and coined these
words, publishing this is an embarrassment. After all, the writer could
just have gone upstairs and asked Tatu about it.
Worse, though, is the claim that "OpenSSH deviates from the standards in
its SCP." Unmentioned, of course, are the standards from which OpenSSH
supposedly "deviates." It can't be the core protocols (transport,
userauth, and connection): since scp just runs the ssh program in a
subprocess for communication, it complies with these protocols exactly as
much as ssh does. All that's left is the rcp protocol scp uses to
actually transfer files. rcp can't "deviate" from SSH, because rcp is a
totally different protocol about which SSH says nothing -- if it does,
then running rsync or cvs, or forwarding an IMAP connection over an SSH
connection, are all deviant as well. SCS appears to imply here that
OpenSSH is somehow non-compliant because its scp does not use SFTP, the
file-transfer protocol developed under the SSH umbrella. And that is
absurd; there is no standard anywhere stating that "any program
transferring files over an SSH connection must use SFTP."
"SSH Tectia Client and Server now incorporate a compatibility mode for
OpenSSH SCP, which still uses the old Secure Shell version 1 (SSH1)."
I'm not absolutely sure what the first part means, since ssh2 and sshd2
have always had an SSH-1 compatibility mode, by which each would just exec
"ssh" or "sshd" instead. However, last year Tectia Server got an SSH-1
"internal emulation" mode, meaning it can now handle protocol 1
connections by itself. So I assume SCS is announcing here a similar
change to scp2. That's certainly good. However, the release then claims
that OpenSSH scp "still uses the old Secure Shell version 1 (SSH1)." This
is just plain false. Firstly, scp doesn't directly use *any* version of
the SSH protocol: it runs a program in a subprocess to connect to the
remote host, and that program is "ssh" by default -- hence, it uses
whatever protocol version ssh is configured to use. It might use either
(or neither, if the user selects a different program with scp -S). And of
course, the OpenSSH default is SSH-2.
"In addition, the new SSH Tectia product versions will support the
OpenSSH public-key file format, eliminating the need for manual key
That, of course, is just peachy.
SCS has a good product in Tectia, which has many feature advantages over
its competitors, both commercial and free. The truth will sell itself;
it's a shame SCS feels the need to use misleading and false statements in
its advertising. These are not qualities a customer looks for in any
vendor, and a security vendor least of all.
Re: ssh.com FUD
Actually, even as I write this, the IETF hasn't actually standardized
the SecSH protocols yet. It's in the "any day now" phase, but it's not
a standard yet.
The current status can be found at
Disclosure: for those that don't know, I'm one of the OpenSSH developers.
Darren Tucker (dtucker at zip.com.au)
GPG key 8FF4FA69 / D9A3 86E9 7EEE AF4B B2D4 37C9 C982 80C7 8FF4 FA69
Good judgement comes with experience. Unfortunately, the experience
usually comes from bad judgement.
- » ssh on command line: force using a group size (prime size) of 1024 (and no...
- — Newest thread in » Secure Shell Forum