Jay Coley, UK technical director, Trend Micro explains why cybersecurity for cloud installations must be proactive, agile, and built-in from the start.
In the third quarter of 2019, the cloud market reached an important tipping point. With public cloud spending soaring by 24 per cent from the previous quarter, over half of all IT vendor revenues from infrastructure came from cloud rather than on-premises, according to IDC. The message was clear: cloud computing has become the new normal in IT.
But as the market matures, so the use of cloud technologies becomes more complex: multi- and hybrid cloud set-ups are now increasingly common. IT security leaders are struggling to manage this complexity and uphold their part of the shared responsibility model with limited in-house skills. They are also faced by an increasingly cloud-savvy cybercrime underground, ready, and able to exploit any gaps in protection and user error.
The bottom line is that organisations cannot afford to do security in the cloud as they treated security in legacy environments – as an afterthought. It must be proactive, agile, and built-in from the start.
A shared responsibility
Organisations migrate data and services to the cloud for a few well understood reasons: to drive efficiency, reduce costs, improve business agility, and enhance the customer experience. Yet there are misconceptions: cloud infrastructure might accelerate time-to-market eventually, but it can take years to fully migrate apps and servers to the cloud. Similarly, Internet as a Service (IaaS) providers might have better security capabilities and resources than you can afford in-house, but they will not do everything. The likes of AWS, Azure and Google Cloud Platform (GCP) might secure the platform infrastructure, but customers are responsible for protecting the data residing there.
This is a common misconception, and one that is driving one of the greatest threats to corporate cloud security today: misconfiguration. Every single day, Trend Micro identifies 230 million configuration errors that can leave organisations unwittingly exposed to cryptojacking, digital skimming, data exfiltration, ransomware and more. In AWS at least, default settings are secure. But they are often changed by under pressure IT teams looking to streamline internal workflows and processes.
A people problem
Even when organisations are not hosting websites in their S3 buckets, there are still risks. If they are publicly writable, the AWS repositories could be abused by cybercriminals to host command-and-control (C&C) servers; store child exploitation material; or be used as an entry point to steal sensitive corporate information.
Containers under fire
Containers represent another front in the war against cybercrime. These lightweight, isolated environments are increasingly favoured by DevOps teams as an easier way to rapidly test and deploy new applications and integrate hybrid clouds. Docker is one of the most popular tools out there to create and deploy containers. Yet here, too, user error is exposing organisations to greater cyber risks.
Trend Micro has discovered numerous Docker servers exposed to the public internet without any protections in place. Often the Docker images were found to already be running cryptocurrency miners. There are several ways a hacker can achieve this: perhaps by installing it from an image that contains the code, hosted on a repository like docker.io, or by using a popular base image like Alpine Linux or Ubuntu and installing the mining software at boot time. Exposed container infrastructure could also invite hackers to install other malware to steal data or hold your IT infrastructure hostage. Remember: any container on an exposed Docker system can be introspected and its data exfiltrated, while they can also allow attackers to move laterally to different parts of the network.
Exposed by error
We also found common security weaknesses with AWS Lambdas, lightweight processes that run for only a limited time, as part of serverless architectures. Some developers think that because the Lambda function names are not directly known to the attacker, they are already more secure and therefore require little in the way of authentication. In reality, they can be relatively easily found by inspecting proxy server logs, examining the source code of web pages that use Lambdas with an API gateway on the back end, or by listening to network traffic with a sniffer. They could, in so doing, expose sensitive internal data or provide attackers with access to other cloud resources.
Another common sight in cloud infrastructure is container orchestration platform Kubernetes. Unfortunately, in just one month in 2019 we found over 32,000 Kubernetes servers that were left wide open. This is not recommended as it allows hackers to probe the infrastructure for remotely exploitable vulnerabilities which could allow them access.
From the start
Most of the mistakes we have highlighted above are preventable. They do not even cover the challenges associated with cloud credentials, which are often leaked via Pastebin and similar sites, and are bought and sold on a thriving black market. The bad news is that the trend for human error in cloud security will only increase as organisations embrace DevOps, containers, and IaaS to drive digital transformation-fuelled growth.
To avoid unintentionally exposing corporate data and systems to the internet, and therefore also to the black hats, IT leaders must first understand their cloud infrastructure. Consider cloud deployment scripts like AWS CloudFormation to see how it all fits together, what controls are in place by default and where there may be gaps. Then it is a case of following best practices such as least privilege access policies and applying a Cloud Security Posture Management (CSPM) tool to monitor closely and continuously for misconfigurations.
Finally, integrate security into your DevOps culture. This is easier said than done, especially if the function has a reputation for being a block on innovation. But with newer, API-led security services now available, developer teams do not have to sacrifice speed and agility for better protection. The benefits of the cloud are undeniable, but they can swiftly be undermined if multi-layered security is not put in place from the start.
Read more – A model for the cloud