NCTV | Cyber Security Assessment Netherlands 2021
In the days that follow, more information about the incident slowly
comes to light. It seems that attackers were able to generate a huge
amount of traffic from the inside, via a botnet of virtual machines.
This internal DDoS attack overwhelmed the virtual machine
manager (VMM) and caused it to fail. The VMM is software that
controls the virtualisation of the hardware (servers in a data centre)
and distributes the available resources such as memory and CPU
among the connected users (customers’ virtual machines). Because
the VMM crashed, all virtual machines connected to the VMM and
in use at the time have been lost.
The VMM has been reset to restore service. Cirrocumulus Networks
will discuss with all affected customers whether their virtual
machines can also be reset or whether further analysis is required
to determine whether data that was being processed at the time of
the crash should and can be restored. This depends on the
configuration of a user's cloud environment and the type of work
the customer performs on the affected virtual machines. For
customers for whom (part of ) the virtual machines are reset, the
availability of their cloud environment is restored a few minutes or
at most a few hours after the VMM is reset. For users where further
investigation is required, this may take days or even several weeks.
In reporting on the incident, much attention is also paid to how
this attack could have taken place. To make the virtual machines
function as a botnet, the attackers placed malware on the virtual
machines. This means that they must have had access to these
virtual machines. This leads to speculation about a connection
with a recent Cirrocumulus Networks user incident, where
attackers were able to exploit a vulnerability during a recovery
operation after an extreme-weather system failure. These attackers
then gained access to the cloud environment of several users,
presumably to exfiltrate data. It now seems that the same
perpetrators started preparing this internal DDoS attack at the
same time. According to experts, now that their activities have
been discovered, the attackers may well have launched this DDoS
attack to hamper the investigation and cause as much damage and
disruption as possible.
Interpretation
For cloud services, DDoS attacks are seen as a concrete risk. One
example is an attack on Amazon's cloud services in 2019. Since
then, much attention has also been devoted to measures to
counteract DDoS attacks, aimed at identifying and repelling
improper traffic. However, when the attack is carried out with
legitimate traffic (e.g. from the service's customers), identifying
and stopping the flow of traffic is much more difficult. DDoS
attacks on cloud environments can either come from outside (for
example, an external botnet attacking a set of virtual machines in a
cloud environment) or from within (an internal botnet of virtual
machines attacking a target within the same cloud environment).
Internal attacks in particular are seen as a serious risk, as they can
disrupt the entire virtual infrastructure.
52
The consequences of virtual machines going down - due to the
VMM crash - are similar to a computer crashing. The data that is
currently being processed and has not yet been saved is lost. How
much data loss there is depends on the settings of the virtual
machine. For complex calculations or data processing that may
take hours or days, a crash is much more drastic than losing the last
few sentences in a word processing document. The way in which
data is stored also influences the impact of such an incident. For
example, it is possible to replicate data in different locations. These
are matters that are not automatically taken care of by a cloud
service provider and that a user should therefore think about when
setting up the (cloud) network infrastructure.
Key terms
Virtualisation: one of the core technologies of cloud services. With
virtualisation, a virtual (simulated) computer environment is
created, whereby one physical computer environment is divided
into several virtual computers, also called virtual machines. Cloud
Service Providers have physical servers in a data centre and the
cloud environment of the customers is built via virtualisation. As a
result, these parties do not need to have their own physical server.
Virtual Machine Manager (VMM) or Hypervisor: a software
program (comprising several modules) that sits between the CSP's
physical server (physical hardware and host operating system) and
the customers' virtual machines (guest operating system). It
enables virtualisation and regulates performance by distributing
memory, CPU and other resources to virtual machines.
Key questions for the reader
1. When designing your cloud environment, did you take the
failure of this infrastructure into account (design for failure)?
2. What activities does your organisation perform in the cloud
environment and how sensitive are these processes to
interruption?
3. How is the data processed in the cloud environment stored?
For complex or sensitive data processing, has replication at
multiple data centre locations or 'availability zones’ been
considered? Please note: Replication can ensure that important
data is not lost in the event of disruption at one location but
remains available at another location.
4. Do you know the basis upon which your organisation chose a
public, private or hybrid cloud environment? Does this include
the complex data processing and sensitive or unique data that
plays a role in your organisational processes?
Back to
Table of Contents
Select target paragraph3
Connect to a paragraph
Connect to an entity
Disable highlights
Add to table of contents