Sorting out generic website fw rules

[DragomirGaliaIT]

Hey everyone,

We are trying to sort out generic firewall alerts that we get as the incidents.

Currently, when there's site blocked that someone browsed through, we get the incident to check for it.

I would like to implement some correlation rule that will only trigger alert for suspicious ad/website when there are more than 1 connection, especially established one, and not to give alert if there's one blocked connection.
So if someone for example has one blocked connection to the site that is suspicious, but other one is established, that would also be alert and incident to check.

Any ideas on how to complete that?

BR,

Dragomir.

[anlynch]

Hi @DragomirGaliaIT,

Thanks for reaching out through LIVEcommunity!

I'm going to do some research on this and get back to you.

[DragomirGaliaIT]

Any update @anlynch . I've did something similar in this rule, don't know if this can help you out:

/*
Query finds the last connection of a suspicious domain and then displays all connections 5 minutes prior and 10 seconds after the connection to a suspicious domain.
Fields AGENT and DOMAIN are mandatory.
You may change variables minutes_before_connection and seconds_after_connection to include more or less results around the connection.

Last connection time used to filter for results is based both on established connections and dns queries.
*/

dataset = xdr_data
| filter agent_hostname = "AGENT" // <-- Agent that you are investigating
| alter suspicious_domain = "DOMAIN" // <-- Domain that caused the alert
| alter minutes_before_connection = 5
| alter seconds_after_connection = 10
| filter action_external_hostname != null or dns_query_name != null
| join type = inner (
dataset = xdr_data
| alter domain = if(action_external_hostname != null, action_external_hostname, dns_query_name != null, dns_query_name, null)
| filter domain != null and agent_hostname != null
| comp max(_time) as last_visit by domain, agent_hostname
) as X _time <= X.last_visit and X.domain = suspicious_domain and X.agent_hostname = agent_hostname
| filter
timestamp_diff(last_visit, _time, "MINUTE") <= minutes_before_connection and
timestamp_diff(_time, last_visit, "SECOND") <= seconds_after_connection
| fields
_time,
actor_process_image_name,
actor_process_image_command_line,
action_external_hostname as established_connection,
dns_query_name as dns_query,
_product
| sort desc _time

As description shows, it is used to tell us number of connection before and after generic alert and incident. If this can help you, it's great.