Re: The Vernacular of Security Narrative
From: Walter Roberson (roberson_at_ibd.nrc-cnrc.gc.ca)
Date: 01/03/05
- Next message: jupiter: "Re: how to encript"
- Previous message: Boudewijn van Breukelen: "Two ADSL routers in one network"
- In reply to: krasicki: "Re: The Vernacular of Security Narrative"
- Next in thread: krasicki: "Re: The Vernacular of Security Narrative"
- Reply: krasicki: "Re: The Vernacular of Security Narrative"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Date: 3 Jan 2005 16:35:22 GMT
In article <1104766160.929162.273240@c13g2000cwb.googlegroups.com>,
krasicki <krasicki@consultant.com> wrote:
:Let's talk about your assumptions. First the corporation needs to
:process millions of transactions per time frame. How many are
:encrypting everything sensitive?
For remote access? If they have that many transactions per second
then they're likely encrypting -everything-. That's what VPN
Concentrators and devices such as Cisco Secure Sockets Module
for the 650x router series are for [$US 50000 per module!]
Financial institutions already encrypt all remote transactions.
There have been a number of descriptions over the years about how
Automatic Teller Machines (ATM) work; there are both session keys
and per-device keys (which get changed periodically.)
:But access is not about location, it is about trusted individuals. And
:my concern is not that trusted individuals go rogue but that some
:combination of trust is compromised by a fellow employee who socially
:engineers their way into unauthorized access and chooses to exploit it.
:What I am suggesting is that in such a case, what additional measures
:can be applied to sensitive applications to reduce the risk of
:compromise?
DCE (Distributed Computing Environment.) It's a pain to set up the
first time, and your application has to be DCE aware, but with it
you can protect by any of a number of criteria, including location
[right down to the device], username, time, and 'view'. Information
on the LAN is encrypted (kerebos).
:The task is to increase the temporal element so that the window of
:opportunity to successfully exploit an insider's glimpse of security is
:eliminated. Let's say a password rotation happens every three months.
If we are making the assumption that this is a big business with
a lot of money at stake and a lot of transactions going on, then
unless the business is very stupid, the business isn't going to rely
on just usernames and passwords. Businesses as big as the ones
you are hinting at usually aren't -allowed- to use simple
username/password protections.
In my opinion, the greater risk is in businesses which are
large enough to handle a fair bit of money, and yet small enough
or distributed enough that the cost of implimenting real security
would put a noticable dent in their profitability. If a company
has a *net* profit of [say] $10 million per year [say on $100 million
in sales], and implimenting a real security architecture is going to
cost [say] $3 million, then they are probably not going to impliment
serious security unless the costs can be distributed over a number of
quarters. Cheaper for them to absorb the occasional half-million
dollar theft than to put in $3 million in real security... especially
if they can get insurance against theft. But for a company like a bank
or Microsoft, $3 million in security is peanuts. Similarily,
$3 million is not much for a hospital, which could be fined
a million or more dollars for not putting in sufficient security
to meet patient privacy laws.
--
WW{Backus,Church,Dijkstra,Knuth,Hollerith,Turing,vonNeumann}D ?
- Next message: jupiter: "Re: how to encript"
- Previous message: Boudewijn van Breukelen: "Two ADSL routers in one network"
- In reply to: krasicki: "Re: The Vernacular of Security Narrative"
- Next in thread: krasicki: "Re: The Vernacular of Security Narrative"
- Reply: krasicki: "Re: The Vernacular of Security Narrative"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
|
|