Chaos Engineering for
Cloud Native Security
Kennedy TorkuraBSides Berlin 2020
1
Cloud Native
Security
the 4Cs of cloud native security
Cloud native infrastructure is infrastructure
that is hidden behind useful abstractions,
controlled by APIs, managed by software,
and has the purpose of running applications
(cloud native applications).
- Kris Nova et. al
Cloud native security is about securing
cloud native infrastructure.
1
Most cloud security incidents are
caused by users
■ User focused security (cloud customer)
□ Shared security responsibility model
□ Users are responsible for security e.g.
configuration of cloud assets
■ CSA Top Threats to Cloud Computing:
□ #1- Data breaches
– Cloud storage
– Cloud IAM
□ #2 Misconfiguration and inadequate
change control
■ OWASP Top 10 - Security
misconfiguration
Cloud Security
Cloud Mis-configurations
users bucket
groups
policy
policy
virtual
machine
Cloud Mis-configurations
* Deploy time/orchestration
■ Terraform
■ CloudFormation
■ Chef ….
⇒ Visibility into the cloud state
⇒ Validation of the efficiency of security
tools
Chaos Engineering
Chaos Engineering is the discipline of experimenting on a
system in order to build confidence in
the system’s capability to withstand turbulent conditions in
production
- principlesofchaos.org
Immunity by Infection -> Vaccination
■ Protect our bodies by introducing disease-like substances
■ Our bodies are tricked to believe the substances are real
■ Our immune systems are triggered to develop appropriate defences
Chaos Engineering
■ Chaos engineering is a fancy name for fault injection
■ Fault injection is a concept that emanates from
scientific research
■ Chaos engineering is not about recklessness but
entails careful planning, experimentation and
observation
Security Chaos Engineering
■ The benefits of chaos engineering go
beyond performance and availability
aspects of resiliencel
■ Resilience is defined as the “ability to
persist dependability”
■ Security is a core attribute of
dependability
■ Chaos engineering concepts are also
suitable for gaining confidence is security
attributes
Risk Driven Fault Injection
Risk-driven Fault Injection aims at injecting
security faults (perturbations) into cloud systems in
order to detect security weaknesses and enable
remediations e.g. security hardening
Risk Driven Fault Injection
Chaos Actions
■ Create
■ Delete
■ Modify
users bucket
groups
policy
policy
virtual
machine
Note
Do not delete
databases,
storage etc
Risk Driven Fault Injection
High level architecture of CloudStrike Results of an experiment
Risk Driven Fault Injection
Security Fault Engine
Risk Driven Fault Injection
start
create user
Bob
get cloud
buckets
select random
bucket
create
malicious
policy
assign policy
to Bob &
bucket
end
An example of an experiment
hypothesis: cloud buckets are secure
DEMO
State Recovery After Experiment
Risk Driven Fault Injection
■ Modes of operation:
□ Low- 30%
□ Medium - 60%
□ High - 90%
■ Attack scenario: chaining
of multiple attack points
Conclusion
● Cloud native security entails securing cloud native infrastructures
● Traditional security tooling is not sufficient for the contemporary
security challenges
● Security chaos engineering is a viable option for improving cloud
native security
Thank you for listening !
Chaos Engineering
for Cloud Native
Security
Email: run2obtain@gmail.com
Twitter: @run2obtain
Website: http://ktorkura.me/portfolio/
Meduim:https://medium.com/@run2obtain

Chaos engineering for cloud native security

  • 1.
    Chaos Engineering for CloudNative Security Kennedy TorkuraBSides Berlin 2020
  • 2.
    1 Cloud Native Security the 4Csof cloud native security Cloud native infrastructure is infrastructure that is hidden behind useful abstractions, controlled by APIs, managed by software, and has the purpose of running applications (cloud native applications). - Kris Nova et. al Cloud native security is about securing cloud native infrastructure.
  • 3.
    1 Most cloud securityincidents are caused by users ■ User focused security (cloud customer) □ Shared security responsibility model □ Users are responsible for security e.g. configuration of cloud assets ■ CSA Top Threats to Cloud Computing: □ #1- Data breaches – Cloud storage – Cloud IAM □ #2 Misconfiguration and inadequate change control ■ OWASP Top 10 - Security misconfiguration Cloud Security
  • 4.
  • 5.
    Cloud Mis-configurations * Deploytime/orchestration ■ Terraform ■ CloudFormation ■ Chef …. ⇒ Visibility into the cloud state ⇒ Validation of the efficiency of security tools
  • 6.
    Chaos Engineering Chaos Engineeringis the discipline of experimenting on a system in order to build confidence in the system’s capability to withstand turbulent conditions in production - principlesofchaos.org Immunity by Infection -> Vaccination ■ Protect our bodies by introducing disease-like substances ■ Our bodies are tricked to believe the substances are real ■ Our immune systems are triggered to develop appropriate defences
  • 7.
    Chaos Engineering ■ Chaosengineering is a fancy name for fault injection ■ Fault injection is a concept that emanates from scientific research ■ Chaos engineering is not about recklessness but entails careful planning, experimentation and observation
  • 8.
    Security Chaos Engineering ■The benefits of chaos engineering go beyond performance and availability aspects of resiliencel ■ Resilience is defined as the “ability to persist dependability” ■ Security is a core attribute of dependability ■ Chaos engineering concepts are also suitable for gaining confidence is security attributes
  • 9.
    Risk Driven FaultInjection Risk-driven Fault Injection aims at injecting security faults (perturbations) into cloud systems in order to detect security weaknesses and enable remediations e.g. security hardening
  • 10.
    Risk Driven FaultInjection Chaos Actions ■ Create ■ Delete ■ Modify users bucket groups policy policy virtual machine Note Do not delete databases, storage etc
  • 11.
    Risk Driven FaultInjection High level architecture of CloudStrike Results of an experiment
  • 12.
    Risk Driven FaultInjection Security Fault Engine
  • 13.
    Risk Driven FaultInjection start create user Bob get cloud buckets select random bucket create malicious policy assign policy to Bob & bucket end An example of an experiment hypothesis: cloud buckets are secure
  • 14.
  • 15.
  • 16.
    Risk Driven FaultInjection ■ Modes of operation: □ Low- 30% □ Medium - 60% □ High - 90% ■ Attack scenario: chaining of multiple attack points
  • 17.
    Conclusion ● Cloud nativesecurity entails securing cloud native infrastructures ● Traditional security tooling is not sufficient for the contemporary security challenges ● Security chaos engineering is a viable option for improving cloud native security
  • 18.
    Thank you forlistening ! Chaos Engineering for Cloud Native Security Email: [email protected] Twitter: @run2obtain Website: http://ktorkura.me/portfolio/ Meduim:https://medium.com/@run2obtain