Cloud Security Pitfalls

What Is Cloud

The cloud is a paradigm in which computing resources are hosted and managed by third-party providers and delivered on demand via the Internet. Users can access, configure, and pay for these services as needed, without owning or maintaining the underlying hardware. AWS, Google Drive, and even TryHackMe are all "clouds" of some sort. There are three main cloud service models you should know about: IaaS, PaaS, and SaaS. Let's explore them one by one!

IaaS

Managing a server room or data center can be challenging: You are fully responsible for its physical security, hardware stability, software updates, network routing, and many other tasks. Every IT company needs some computing power to run its business, but not every company has the expertise or resources to maintain and properly secure its own servers.

That's the problem Infrastructure as a Service (IaaS) tries to solve. IaaS is a cloud service model where computing infrastructure is provided online on demand. For example, with an IaaS provider like Amazon AWS, Google Cloud, or Microsoft Azure, you can launch any virtual machine in the cloud just by clicking a button in a web GUI, without worrying about power outages and hardware failures.

PaaS

For some companies, IaaS is not enough. For example, software developers don't want to bother with launching virtual machines - all they want is to write the source code, click a button, and see their application up and running, without caring much how and where it actually runs. Such requests are covered with Platform as a Service (PaaS) - a cloud service model for simple development and hosting web applications.

Interestingly, TryHackMe also offers PaaS features, as you can create your own private or public rooms and then host them in the cloud, without knowing how it works internally. In turn, TryHackMe uses IaaS (Amazon AWS) to host its infrastructure, including in-room virtual machines. Other PaaS offerings include Vercel, Heroku, and Google App Engine.

SaaS

Software as a Service (SaaS) allows users and companies to launch complex applications in the cloud without installing any software on their computers. Slack, Zoom, Gmail, Dropbox, and Google Docs are among the thousands of SaaS offerings. Unlike the previous models, SaaS is always a final product that can be used by a non-technical audience. Let's see a comparison table below for a simpler understanding:

It's fun to think about how you can combine the models. For example, if you develop a cat image generator, make it a public web service, and host it in AWS, you'll get SaaS on IaaS (or CataaS). What you should remember is that the "as a service" approach always implies some form of abstraction, where a maintenance burden is taken from you and delegated to a cloud service provider.

Security of the Cloud

Cloud computing is a complex topic, but it is built on top of the same core technologies as traditional on-premises. For instance, AWS often shares insights into its internal architecture (video 1, video 2), where familiar concepts like TCP/IP and Linux play a key role. There is even a saying that "the cloud is just someone else's computer". This is why you must know that clouds are not invulnerable and, just like your computer, are at risk of attacks. Security of the cloud provider's internal infrastructure is often called Security of the cloud:

Risk of Cloud Vulnerabilities

Biggest clouds are built with security in mind, and it's extremely rare to see them getting breached. However, it's still possible, and when they occur, attackers often target the provider's largest customers. Treat this as a form of supply chain risk - don't blindly trust the cloud provider, but apply the same defensive principles: segment the network, analyze login activity, and monitor endpoint behavior.

For less popular cloud services, the chance of compromise is much higher. There were many cases where a breach of a local IaaS provider resulted in the deployment of malware on all hosted VMs, or where a breach of SaaS led to the exposure of sensitive data. This is why you should carefully choose your cloud vendor and decide what to entrust them with.

Risk of Poor Cloud Visibility

The risks associated with cloud breaches are amplified by the fact that, as an end user, you have no visibility into the cloud provider's internal environment. For example, between August 8 and August 18, 2025, adversaries used stolen Salesloft (SaaS) OAuth tokens to exfiltrate data from certain SaaS customers' tenants. As an end user, you can't detect it, as the malicious activity occurred entirely within SaaS infrastructure - essentially, "someone else's computer".

No matter how mature your SOC is, using cloud services means entrusting your data to third-party vendors, who will never provide their internal security logs upon request. This is especially critical to maintain control over SaaS usage. Other departments may unknowingly upload sensitive data to untrusted SaaS applications, creating shadow IT risks that can later lead to unexpected breaches.

Incidents of the Cloud

Security in the Cloud

The previous task was about the security of the cloud - something that the cloud provider takes care of. However, it's still your responsibility to protect resources in the cloud: VMs you host in IaaS, applications you build with PaaS, or credentials to your SaaS accounts. Everything you have in the cloud requires the same level of monitoring and hardening as on-premises systems. Check out the Shared Responsibility Model for more details.

Cloud Migration Pitfalls

There is a misconception that migrating an old, unpatched virtual machine to the cloud somehow makes it new and secure, or that moving files to Google Drive guarantees protection from ransomware. Neither is true. While cloud environments can reduce exposure to certain traditional attack vectors, they also introduce new, cloud-specific threats that are often overlooked by IT teams.

Another issue is that people tend to apply their on-premises security practices to cloud environments. For instance, it is acceptable to use 12-character passwords without MFA in an isolated Active Directory network, but it is critically dangerous in public clouds accessible from anywhere. Cloud security requires a unique approach, which you will learn about in the following rooms.

Logging in Clouds

Since you can't install the SIEM agent in the cloud and collect logs from there the same way as from on-premises, you have to rely on the cloud provider's solution. Some vendors provide comprehensive logging services, such as AWS CloudTrail, which you will learn in the next room. However, in the majority of cases, cloud logging is limited, especially in SaaS, where:

• Paid Logs: Logging to SIEM may require an additional payment or license
• Poor Format: Log fields may be incomplete, unstructured, or not documented
• Lack of Integration: In some cases, solutions don't support logging to SIEM at all

Incidents in the Cloud

What to Protect and Monitor

The effort required to build monitoring in the cloud differs depending on the cloud service model. For example, SaaS is the simplest, as all you have to do is ingest the logs in SIEM via the provider's API and monitor for risky actions, such as when a confidential Google Drive document is made public, an internal GitHub repository is downloaded, or when someone logs in to Notion from a suspicious IP and exports all corporate pages. On the other hand, covering IaaS is a more challenging task, as you need to cover both:

• Workloads: Monitor virtual machines or containers, same as in on-premises
• Cloud Services: Monitor database queries, storage access, and many more
• Control Plane: Monitor logins and actions within the cloud admin console

Data Sources for Monitoring

On-premises endpoint monitoring typically relies on EDR, SIEM, and forensic tools. But in the cloud, the tools change drastically. EDRs are often unsupported due to containerized workloads and auto-scaling; SIEM integration with the provider's logging APIs may be difficult; and forensic investigation is constrained by a lack of accessible memory or disk artifacts. The image below illustrates the differences in cloud and on-premises logging:

Common Cloud Security Tools

Many specialized solutions begin to pop up to cover what traditional tools can't cover in the clouds. Cloud Access Security Brokers (CASB) enforce security policies, Cloud Workload Protection Platforms (CWPP) protect against malware, Cloud Security Posture Management (CSPM) alert on misconfigurations, and so on. While we encourage you to read about them more, you can still set up monitoring just with SIEM and have a decent SOC coverage:

Cloud Security Challenge

In this task, you will explore the differences between cloud service models and repeat the Shared Responsibility Model - what you manage versus what the vendor manages. Visit the static site below, complete the exercises, and get the flags!

Access	Granted
URL	Static Site

In this room, you explored the differences between IaaS, PaaS, and SaaS cloud service models, uncovered often-overlooked risks, and examined common pitfalls organizations face when migrating to the cloud. You also learned that protecting clouds is not easy, and that there are many challenges in log collection and SOC coverage to ensure proper security monitoring. In the upcoming rooms, you will dive deeper into the technical details using AWS (IaaS) and Entra ID (SaaS) as examples. Hope you enjoyed the room!