Hello Albert and All,

At 11:35 AM 3/1/2002 +0100, albert.mietus wrote:
>Hi all,
>
>We have discussed a lot about SHA/DSA/RSA/MAC for securing syslog on small
>systems.
>My point was it should be possible to make a "more secure then nothing"
>syslog for small systems. I think anybody agrees on that.
>
>But, ... my suggestions was to do it with syslog-sign, as it is the most
>simple security extension to syslog we have discussed until now.
>As Jon pointed out, by doing this we have a huge risk of making -sign a lot
>more complex and a lot less "beautiful" at it is.
>We should do that! I fully agree with Jon.
>
>So I have a new proposal.
>
>First, I would like to suggest to minimise the change to -sign. Basically,
>leave it the way it is. But make the key length variable. I don't see any
>reason why the key length should be fixed (as it is now). The keylength can
>be specified in the certificate blocks. Or even in the signature blocks.
>The current (fixed) size is:
>        a) to long for small(er) systems
>        b) insecure, in some future time.
>By making the key length a "variable", we allow "small" (but less secure)
>keys, for small systems.
>And, maybe even more important, allow an easy upgrade for "very secure"
>extreme long keys. The will be needed somehow, someday. The RFC should
>recommend an (minimum) key length; it should be the current one.
>
>
>Second, I would like to propose to discuss an alternative "secure syslog",
>especially for "small" systems. This could be a "syslog-mac", it could be a
>kerborized one(if that is "small"), it could be something else.  I have to
>think about it more. But I'm willing to do so.
>
>But I don't know iff this can be done within our charter (Chris, can you
>answer this question?)

I feel sure that we can proceed if others are interested
in doing so.  I would ask everyone to think about this
before we agree, however.  If the proposed alternative
"syslog-mac" were to be much simpler than syslog-sign,
would people prefer to implement that rather than doing
the harder work to implement syslog-sign?  

I'm a bit undecided on this as well.  On one hand, I'd
like to say that syslog-sign has the attributes that we
want.  If we start doing other things, then we may
reduce the importance of that work.  If the proposed 
alternative "syslog-mac" were to be much simpler than 
syslog-sign, would people prefer to implement that rather 
than doing the harder work to implement syslog-sign?  On 
the other hand, I'd like solutions that work and that will 
be implemented.  It does sound like "syslog-mac" may have 
qualities that would appeal to people working on low-end 
devices.  

Let's hear from other people about this and then decide.

Thanks,
Chris

Reply via email to