Kubernetes Anomalous Outbound Network Activity from Process
THIS IS A EXPERIMENTAL DETECTION
This detection has been marked experimental by the Splunk Threat Research team. This means we have not been able to test, simulate, or build datasets for this detection. Use at your own risk. This analytic is NOT supported.
Description
This detection detects outbound network traffic volume anomalies from processes running within containerised workloads. Anomalies are provided with context identifying the Kubernetes cluster, the workload name, and the type of anomaly. This detection leverages Network performance Monitoring metrics harvested using an OTEL collector, and is pulled from Splunk Observability cloud using the Splunk Infrastructure Monitoring Add-on. (https://splunkbase.splunk.com/app/5247). This detection compares the tcp.bytes, tcp.new_sockets, tcp.packets, udp.bytes, udp.packets metrics for source (transmitting) workload process pairs over the last 1 hout, with the average of those metrics for those pairs over the last 30 days in order to detect any anonymously high outbound network activity. Anonymously high outbound network traffic from a process running in a container is a potential indication of data exfiltration, or an indication that the process has been modified. Anomalously high outbound network activity from a process running within a container suggests the potential compromise, which may lead to unauthorized data exfiltration, communication with malicious entities, or the propagation of malware to external systems. The compromised container could also serve as a pivot point for further attacks within the containerized environment.
- Type: Anomaly
-
Product: Splunk Enterprise, Splunk Enterprise Security, Splunk Cloud
- Last Updated: 2024-01-10
- Author: Matthew Moore, Splunk
- ID: dd6afee6-e0a3-4028-a089-f47dd2842c22
Annotations
Kill Chain Phase
- Installation
NIST
- DE.AE
CIS20
- CIS 13
CVE
Search
1
2
3
4
5
6
7
8
9
10
11
12
13
14
| mstats avg(tcp.*) as tcp.* avg(udp.*) as udp.* where `kubernetes_metrics` AND earliest=-1h by k8s.cluster.name source.workload.name source.process.name span=10s
| eval key='source.workload.name' + ":" + 'source.process.name'
| join type=left key [ mstats avg(tcp.*) as avg_tcp.* avg(udp.*) as avg_udp.* stdev(tcp.*) as stdev_tcp.* avg(udp.*) as stdev_udp.* where `kubernetes_metrics` AND earliest=-30d latest=-1h by source.workload.name source.process.name
| eval key='source.workload.name' + ":" + 'source.process.name' ]
| eval anomalies = ""
| foreach stdev_* [ eval anomalies =if( '<<MATCHSTR>>' > ('avg_<<MATCHSTR>>' + 3 * 'stdev_<<MATCHSTR>>'), anomalies + "<<MATCHSTR>> higher than average by " + tostring(round(('<<MATCHSTR>>' - 'avg_<<MATCHSTR>>')/'stdev_<<MATCHSTR>>' ,2)) + " Standard Deviations. <<MATCHSTR>>=" + tostring('<<MATCHSTR>>') + " avg_<<MATCHSTR>>=" + tostring('avg_<<MATCHSTR>>') + " 'stdev_<<MATCHSTR>>'=" + tostring('stdev_<<MATCHSTR>>') + ", " , anomalies) ]
| fillnull
| eval anomalies = split(replace(anomalies, ",\s$$$$", "") ,", ")
| where anomalies!=""
| stats count(anomalies) as count values(anomalies) as anomalies by k8s.cluster.name source.workload.name source.process.name
| where count > 5
| rename k8s.cluster.name as host
| `kubernetes_anomalous_outbound_network_activity_from_process_filter`
Macros
The SPL above uses the following Macros:
kubernetes_anomalous_outbound_network_activity_from_process_filter is a empty macro by default. It allows the user to filter out any results (false positives) without editing the SPL.
Required fields
List of fields required to use this analytic.
- tcp.*
- udp.*
- k8s.cluster.name
- source.workload.name
- dest.workload.name
- udp.packets
How To Implement
To gather NPM metrics the Open Telemetry to the Kubernetes Cluster and enable Network Performance Monitoring according to instructions found in Splunk Docs https://docs.splunk.com/observability/en/infrastructure/network-explorer/network-explorer-setup.html#network-explorer-setup In order to access those metrics from within Splunk Enterprise and ES, the Splunk Infrastructure Monitoring add-on must be installed and configured on a Splunk Search Head. Once installed, first configure the add-on with your O11y Cloud Org ID and Access Token. Lastly set up the add-on to ingest metrics from O11y cloud using the following settings, and any other settings left at default:
- Name sim_npm_metrics_to_metrics_index
- Org ID <Your O11y Cloud Org Id>
- Signal Flow Program data('tcp.packets').publish(label='A'); data('tcp.bytes').publish(label='B'); data('tcp.new_sockets').publish(label='C'); data('udp.packets').publish(label='D'); data('udp.bytes').publish(label='E')
- Metric Resolution 10000
Known False Positives
unknown
Associated Analytic Story
RBA
Risk Score | Impact | Confidence | Message |
---|---|---|---|
25.0 | 50 | 50 | Kubernetes Anomalous Outbound Network Activity from Process in kubernetes cluster $host$ |
The Risk Score is calculated by the following formula: Risk Score = (Impact * Confidence/100). Initial Confidence and Impact is set by the analytic author.
Reference
Test Dataset
Replay any dataset to Splunk Enterprise by using our replay.py
tool or the UI.
Alternatively you can replay a dataset into a Splunk Attack Range
source | version: 1