|
As a place to start, how about in the current STIX model anything that extends from Related___Type is a relationship object, anything with a normal @idref is not?
Sighting wasn’t top-level before so we get to make it up.
We need to make sure it's very clear where the relationship object starts and ends. If I am not forced to use the relationship object for a sighting relationship, then I go back to supporting an atomic sighting object.
Aharon
Sent using OWA for iPhone
Are you actually forced to use the relationship object to have sighting work there? I see that as a separate use case from relationships, and in cases like this we can still have the target_id field directly there without the need for a new object.
Same thing for an indicator pattern, for example.
While those are references between top-level constructs they seem different than the normal type of relationships we’ve been discussing.
John
BTW: Any thoughts on setting up a slack channel for STIX dev discussions? Sometimes I think that would be more conducive to these questions than e-mail.
Bret, I almost always prefer atomic objects.
If we do both a relationship object and a Sightings atomic object together, it just seems... well weird.... (not very scientific I know)
Example Sightings Object -
ID: Sighting GUID
Marking: Sighting TLP
Producer: Who made the sighting
Timestamp:
Target_ID: Replaced by Relationship Object
Now I am going to be forced to use the relationship object to make the Sighting work. I am also going to be forced to make a potentially large number of new Sighting Objects (since there is a timestamp). Also, a sighting by itself, without the looking
into the Relationship object is kind of useless.
We can eliminate this extra complexity by eliminating the atomic Sightings object and replacing it with a relationship type.
Just debating
<OutlookEmoji-😊.png>
Aharon Chernin
CTO
SOLTRA
| An FS-ISAC & DTCC Company
18301 Bermuda green Dr
Tampa, fl 33647
One thing to keep in mind is that we want the objects as small and simple as possible. Some times to make them more broad you have to add a lot of extra fields. This should be avoided. We want them to be as atomic as possible. Also, if they
are separate then they can grow and evolve independently.
This is one of the many things I do not like about how STIX and CybOX is done today. The excessive use of object oriented reuse makes it nearly impossible to fix or change certain things as that would have adverse effects on other areas that
can not take those changes.
Object reuse is not always a good thing.
Thanks,
Bret
Bret Jordan CISSP
Director of Security Architecture and Standards | Office of the CTO
Blue Coat Systems
PGP Fingerprint: 63B4 FC53 680A 6B7D 1447 F2C0 74F8 ACAE 7415 0050
"Without cryptography vihv vivc ce xhrnrw, however, the only thing that can not be unscrambled is an egg."
Great discussion topic!
The conversation seemed (to me) to settle on the idea that there were three concepts that are related in some way:
1. Relationships
– A link between objects (e.g., this TTP is related to that Indicator)
2. Assertions
– The +1/-1 concept
3. Sightings
– “I saw that, too!”
It seems that the structures are similar across the three concepts (e.g., id, from, to, assertion, source/confidence/rationale) and that the larger open question is
whether humans are benefitted by these things being variations of the same concept or three different concepts (or something else).
I personally think there is a single set of common properties that can do Relationships, Assertions, and Sightings, and that it looks roughly like what Aharon posted.
However, there was a counter-point that this combining of concepts makes it more difficult to understand.
I’ll leave the group with these questions:
1. Is
there a single set of properties that makes sense for Relationships, Assertions, and Sightings?
2. If
there is a single set of properties, does it make sense to combine them, as Aharon has mentioned?
3. What
clarifying questions, if any, do you have that will help you answer #1 or #2?
a. Note
that this might be the most important of the three questions!
Thank you.
- Mark
This should be "sightings object rethought". While coming up with a proposal, I spotted a different way of thinking about Sightings. In my opinion, the most important thing is determining which STIX object
is being sighted. However, there is some other bits of information that is useful: sightings producer and date/time of sighting.
Now take a look at the recent relationship object discussions:
Relationship Object Discussion:
ID [1]: The ID of the relationship,
a simple random GUID
Marking[1]: The ID of the marking object that
you should reference
Version [1]: The version of the relationship; a simple number to be used with the ID for version control
Type [1]: The “type” of relationship being expressed. (Not sure of how this works yet)
Description [1]: A single simple and short description
Source [1] : The ID of one or more source entities in the relationship as a URI (not QName)
Targets [1..N]: The ID of one or more targets in the relationship as a URI (not QName)
Start [1]: A timestamp in UTC stating when the relationship between the objects started, or the text 'unknown'.
End [1]: A timestamp in UTC stating when the relationship between the objects ended, or the text 'ongoing', or the text 'unknown'.
Reliability/Confidence [1]: A measure of confidence in the relationship using the Information Reliability scale.
Producer [1]: A simple producer object like what John calls out
Timestamp [1]: A timestamp in UTC stating
when the relationship object was created.
Could a sighting be a type of Relationship?
Relationship Object Discussion:
Marking[1]: TLP Green
Version [1]: 1
Type [1]: Sighting
Description [1]: Soltra Edge reported Sighting
Source [1] : Soltra
Targets [1..N]: soltra:indicator-<GUID>
Start [1]:
End [1]:
Reliability/Confidence [1]:
Producer [1]: Soltra
Timestamp [1]: <timestamp>
Or is there more meta data we need to collect regarding sightings that a sighting deserves it's own object?
SOLTRA |
An FS-ISAC & DTCC Company
|