Re: Secure logging (was: King COunty)

From: Charlie Strauss <cems_at_earthlink_dot_net>
Date: Tue Nov 09 2004 - 22:59:49 CST

Thanks for the tip on how to search. I checked and did not find a
discussion on this point.

The point I was inquiring about was in response to Jim Marchs
observation that the audit logs in King County are missing entries that
should be there. The supposition is the logs were edited.

So the question I was asking was if there is a simple means to assure a
write-once log. What I was wondering is if this is technically
feasible on a CD-R.

It appears that OVC is writing cd's in iso standard data format.

is it technically possible to write short records on demand to a CD-R
in this format?

I've heard of multi-session CR-r (where you dont 'finalize' the disk
after each write). But is this suited to something with the number of
independent short writes a log-file will likely have? is there perhaps
a either a limit on the minimum size of a write or the maximum number
of writes one can do on a CD-r?

On common OS's one actually buffers the writes to disk then dumps the
buffer to the CD. But buffered writes are not what you want for a
real-time logfile.

On Nov 9, 2004, at 9:36 PM, Ed Kennedy wrote:

> Hello Fred and Charlie:
>
> Just in case you weren't aware of this, there is a Searchable
> correspondance archive
> http://gnosis.python-hosting.com/voting-project/ . If OVC has
> previously discussed a subject like this, that is where you will find
> it. BTW, I don't understand the issue. Could you please clarity it
> for this simple minded civil engineer?
>
> Thanks, Ed Kennedy
>
>
>
> Fred McLain wrote:
>> A safe, inexpensive and well tested mechanism exists for logging, cash
>> register printers. It's not as techie, but they work! Integration of
>> one of these with the tally software we have shouldn't be overly
>> difficult. What might be more interesting is making them physically
>> secure as they are in cash registers.
>>
>> -Fred-
>>
>> On Tue, 2004-11-09 at 16:56, charlie strauss wrote:
>>> In response to Jim March's pointing out that the paper log
>>> generation events dont show up in the computer logs. I am wondering
>>> what technical means can not only make audit log changes detectable
>>> but also make them very difficult.
>>>
>>> If the systems were on line one could use a remote, crytographic
>>> time-stamped-hash server, but I dont think we want them on-line.
>>>
>>> Is it technically feasible for a conventional CD-R in normal CD
>>> iso-data format to write short records without buffering them in
>>> large batches (that would delay their commitment to the CD)? I've
>>> used multi-session CD-Rs but the sessions tend to be large so
>>> perhaps the iso format limits the number of these events??
>>>
>>> If so one could use serial number stamped CDs for the loggers and
>>> not drive up the price with special logger hardware.
>>>
>>> Or has OVC already dealt with this another way.
>
==================================================================
= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
==================================================================
Received on Tue Nov 30 23:17:23 2004

This archive was generated by hypermail 2.1.8 : Tue Nov 30 2004 - 23:17:44 CST