The Dutch Tax Administration (Belastingdienst) sent me a trophy and formal letter of appreciation on behalf of the Dutch government. The front engraving on the trophy playfully reads, “I hacked the Dutch Tax Administration and never got a refund”. Together with the trophy was a formal letter of appreciation.On behalf of the Dutch Tax and Customs Administration, we would like to thank Jacob for participating in our Coordinated Vulnerability Disclosure program. For that, we present this letter of appreciation to Jacob.At the Dutch Tax and Customs Administration, we consider the security of our systems a top priority. But no matter how much effort we put into system security, there can still be vulnerabilities present. For that, we warmly welcome people like Jacob as a particpiant in the Coordinated Vulnerability Disclosure program.We appreciate not only that you have reported a security issue to us, but also that you have professionally done this. I would like to extend my thanks to the Belastingdienst Security Operations Center for sending me such a kind gift.
The UK Ministry of Defence (MoD) sent me a VDP challenge coin for my finding and responsible disclosure of a critical (9.6 CVSS) severity vulnerability. Together with the coin was a small thank you note.Thank you! The Ministry of Defence takes the security of our systems seriously. To show our appreciation for all your time and effort, we would like to reward you with a Vulnerability Disclosure hacker coin. This echos a similar response to when I hacked the Dutch government and they sent me a t-shirt.
The NCSC-NL (National Cyber Security Centre – Netherlands) sent me a ‘lousy’ t-shirt on behalf of the Dutch government. Together with the t-shirt was a thank you letter.Thank you for bringing a vulnerability to our attention. Together with vulnerability reporters like you we can increase the resilience of Dutch society in the digital domain and better protect our systems and systems of our partners. This was a pleasant response to receive and illustrates a far better approach to engaging with ethical hackers than traditional threats of prosecution. However, there’s been some past controversy within the security community on whether this type of reward disincentivises ethical hacker participation by undermining the value inherent in VDP and bug bounty programs. Some argue that the effort researchers need to invest in helping to find and responsibly report vulnerabilities to government organisations far outweighs the level of compensation value these novelty rewards are worth. My view is that expectations should be managed realistically, and maybe the focus should shift away from pursuing personal gain to instead encouraging wider public sector adoption of better security practices. I’m happy with my t-shirt and appreciate the efforts the NCSC-NL went to. It’s certainly a better response than some of the other governments I’ve reported vulnerabilities to.
Believe it or not, sometimes admins will design systems to attract attackers. Working in information security, one of my favourite defensive strategies to deploy in operational practice involves the use of honeypots and honeynets. These tightly controlled decoy mechanisms are designed to entice attackers, rob them of their time, and help with profiling attack intent, objectives, and origin. Honeypots Honeypots are a useful way for admins to learn more about an adversary's objectives by intentionally exposing a machine that appears to be a highly valuable and sometimes unprotected target. Although a honeypot may seem legitimate to an attacker, honeypots are typically isolated from normal internal networks and configured in such a way that all interactive activity can be monitored and logged. This has several benefits from a defensive point of view. By convincing an attacker to focus their efforts on a designated honeypot endpoint, an administrator can gain insight into an attacker's tactics, techniques, and procesures (TTPs). This can be used to predict attack execution behaviour, or even aid in attribution and identifying where the attack may have originated from. Furthermore, honeypots may delay an attacker, buying administrators crucial time to respond, or force attackers to exhaust their own resources pursuing fruitless tasks. Honeypots have been in use for several decades, but they have often been costly to deploy because this typically meant dedicating actual hardware to face attackers, thus reducing what infrastructure could be reserved for production purposes. Furthermore, in order to engage an attacker for any significant amount of time, a honeypot needs to look like a real (and ideally valuable) network node, which means sitting in the attacker's seat and putting some thought into what software and data should be deployed on it. This all takes lots of time and traditionally was not practical for very large deployments. However, virtualisation addresses many of the challenges associated with administering honeypot machines because virtualised infrastructure is designed to scale easily. Honeynets A honeynet is an entire network designed to attract attackers. The benefits of its use are the same as that of honeypots, but honeynets are designed to look like real network environments, complete with real operating systems, applications, services, and associated network traffic. Honeypots can be thought of as a highly interactive set of honeypots, providing realistic feedback just as a real network would. For both honeypots and honeynets, deployed services are not actually used in production, so there shouldn't be any reason for any legitimate interaction with the servers. This makes it easy to recognise that any prolonged interaction with deployed services usually implies malicious intent. It follows that traffic from external hosts in a honeynet is usually indicative of attack behaviour and not as likely to be a false positive generated by expected network traffic.As with individual honeypots, all activity is monitored, recorded, and security controls optimised to balance the liklihood of any attack occurence with the ease of attack execution. As with honeypots, virtualisation has also improved the performance of honeynets, allowing for varied and easily scalable network configurations on existing hardware infrastructure.
Mobile computing devices have revolutionised the way people interact with each other and the Internet, though just like any other computing device, mobile devices are subject to vulnerabilities. Some of these vulnerabilities are based on the design of the architecture and how data is processed, stored, and transmitted to and from the device. Mobile device architecture Smartphone and tablet devices are composed of various hardware and software components (e.g, an operating system and software applications). A battery provides the external power source, and a keypad or touchscreen allows the user to interact with the device. Most mobile devices are built with a system on a chip (SoC). The SoC is a small, integrated circuit that connects together common components that make up a mobile device. SoC are designed to reduce overall system costs, increase performance, and lower power consumption. Just like in a personal computer (PC), the CPU is used for decision logic and the GPU is responsible for visual processing. RAM provides temporary memory storage for applications, and ROM provides the long-term storage, such as for firmware and operating systems. When the mobile device is configured for a subscriber network like Three, EE, or Vodafone etc, the modem allows mobile devices to communicate over cellular networks, using basic phone services to make phone calls and send text messages. A SIM (subscriber identity module) card is unique, and is required in order to identify and authenticate a user's device on the cellular network. Once authenticated, the user's communications are encrypted. SIM cards have a limited storage capacity (up to 256KB) and contain information regarding the user's identity, location, network authentication data, phone number, stored contact lists, and even stored text messages. Setting a SIM personal identification number (PIN) on the mobile device can help protect your data in the event the device is lost or stolen. Two of the most common mobile operating systems on the market are iOS (iPhone Operating System) and Android. The iOS operating system is proprietary and runs exclusively on Apple mobile devices (i.e, iPhone, iPad, etc). Android, which is developed by Google, is open-source and found on a variety of hardware such as mobile phones, televisions, and other technological items. Android Android is a mobile operating system based on the Linux 2.x and 3.x kernels. Much like iOS, the Android platform is made up of different layers (stacks) that offer distinct services and interface with other components within the stack. On a mobile Android device, users interact within the application layer. This layer is also home for the native system apps that are installed by default such as the calendar app, camera, and email. Android applications are developed in Java. Applications run their own processes within a virtual machine (i.e, an instance of ART, which is short for Android Runtime), as if they were separate user accounts with separate home directories. This provides isolation among all the other applications running on the device. The Java application programming interface (API) framework exposes features of the Android OS to simplify access to application data and other system components. The primary components of an Android application are: Activities - Parts of the application the user can see. Fragments - A behaviour that is placed in an activity. Intents - Used for sending messages between other components. Broadcast receivers - Allow an application to receive notifications from other apps. Content providers - A SQLite database to store data in the form of a flat file. Services - Used to start intents, send notifications, and process data. The hardware abstraction layer (HAL) interfaces with built-in hardware components of the device. The native C and C++ libraries provide support for applications developed in native code, such as HAL and ART. The kernel provides foundational services to other components within the platform, such as drivers, memory management, display functionality, etc. iOS The iOS is based on Darwin, which is an open-source, Unix-based OS that was first released by Apple in 2000. iOS is a layered architecture that is made up of four levels of abstraction. Cocoa Touch - User interface (UI) framework for developing software apps, like games, to run on iOS. Media Services - Provides audio, graphics, video, and over-theair (AirPlay) capabilities. Core Services - Fundamental services like networking, file access, address book, etc. Core OS - Provides OS functionality such as power management, file system, etc. Each layer contains different frameworks, which are groups of libraries and resources (i.e, images, header files, etc) that can be used for developing an application. Smaller applications typically contain all the resources they need to function directly in the application bundle. In relation to iOS development, another word for framework is a bundle. Objective-C and Swift are high-level programming languages specifically for Apple operating systems like iOS, whereas the low-level programming language C is used for operating system and kernel development. The six core features of the iOS security architecture are: Hardware security Secure boot (secure boot chain) Code signing Sandbox Encryption and data protection General exploit mitigations When an iOS device is booted, it goes through a process that Apple calls the secure boot chain. Apple uses an Apple Root CA (Certificate Authority) certificate, which is loaded in read-only memory (boot ROM) for verifying other certificates to establish explicit trust relationships. Each step of the boot process contains components that are cryptographically signed by Apple. This signature represents a chain of trust and is verified every time the device is booted to ensure the device has not been tampered with. This process is similar to the applications that are allowed to run on the device. Apple use code signing to ensure only approved applications are deployed on the device. Users are forced to visit the Apple store to download authorised applications that have been signed by Apple, kind of like being in application prison. Ironically, jailbreaking the device is the only way to bypass the security mechanisms and run third-party applications. Jailbreaking is the process of exploiting a software vulnerability in iOS that enables low-level execution with elevated privileges to bypass the security mechanism in iOS. The hardware security feature provides cryptographic opertions to secure technologies operating on the iDevice. This is probably the most important security feature of the device. There are two Advanced Encryption Standard (AES) 256-bit encryption keys included on every iDevice, called group ID (GID) and unique ID (UID) values. The GID key is used to prevent modification to firmware files, outside of the user's private data. UIDs are created during manufacturing and are unique to every device. They are used in conjunction with passcodes and other data protection mechanisms for file encryption and decryption. If hardware-like memory chips are removed and reused on another iDevice, encrypted files would not be accessible. The keys are fused into the application processor and are not recoverable, not even when using a JTAG or other debugging interface. The AES-256 crypto engine, which works with a SHA-1 cryptographic hash function, is built into every iDevice to encrypt data and optimise overall performance. A JTAG (Joint Test Action Group) is a type of hardware mechanism used for debugging and connecting to embedded devices on a circuit board. The sandbox is a restricted area where applications are executed from. It is a general mitigation technique to prevent escalation attacks. If an application were to be compromised, the damage would be limited to the data managed by the vulnerable application and possibly the data from other applications, like your contacts, depending on the access restrictions enabled by the iOS user. Conclusion Android and iOS application developers perform the majority of the software development higher up the stack, since most of the resources and libraries for working with subcomponents are readily available and easy to work with. Because most of the development activity happens at the application layer, mobile users tend to fall victim to vulnerabilities derived from poor security development practices. This is why it's important for security conscious mobile users to understand not only the underlying architecture, but also how to balance their own security considerations with the growing need for varied mobile device usage in practice.
In this era, cyber-attacks are gaining in popularity with more sophisticated techniques and attack deployment strategies being seen in the wild than ever before. DDoS (Distributed Denial of Service) attacks have gained popularity over the years, and as network systems advanced these types of attacks grew to become more refined. Many types of DDoS attacks now exist including HTTP flood attacks, SYN flood attacks, volumetric attacks, DNS amplification attacks, UDP flooding, ICMP flood attacks, and many others. In this blog post, I will walkthrough the basic and advanced concepts of a DDoS attack, what they are, the different types, and how companies can mitigate DDoS attacks with preventive techniques or measures to mitigate their effects. What are DDoS attacks? DDoS attacks or Denial of Service attacks are a malicious attempt to disrupt services offered by a server. In a typical DDoS attack, the intent of the attacker is to disrupt normal traffic of the target by maliciously overwhelming it with a coordinated flood of traffic, consequently denying any legitimate traffic access to the service. This attack is possible by bridging various types of computers and devices into a centrally controlled botnet and collectively pointing them all towards a target endpoint. As multiple systems are required to carry out a DDoS attack successfully, IoT devices are the low-hanging fruit for attackers aiming to grow their own botnet. This is because the nature of the IoT market means manufacturing processes are often rushed and security controls are considered secondary. As a result, many IoT devices are released with high severity vulnerabilities which are trivial for opportunistic hackers to exploit. Types of DDoS attacks There are many ways to disrupt or suspend a server’s resources or computing operations, and hackers have found various ways to overwhelm servers using various simple techniques. Approaches like using botnets and automation to issue requests demand the targeted servers employ more computational power to handle them. Thus, most of the time, such attacks are successful. DDoS attacks can be broadly divided into three types, which reside within specific layers of the OSI model. The layers of the OSI Model The first type are volumetric attacks. In this type, the attacker intends to saturate bandwidth of the target server and measures the intensity in bits per second. These include spoofed packet floods, ICMP floods, and UDP floods. The second type of attacks are protocol attacks, which are together termed Layer 3 and Layer 4 attacks. These are initiated by targeting vulnerabilities in the network and transport layers of the OSI model. In this type, the main intention is to consume computational resources and can impact the infrastructure used to manage network traffic, such as firewalls and load balancers. The third type are known as application layer Attacks. These include flooding GET / POST requests for images, files, or other large file size assets. These attacks are intended to crash the server and the intensity is measured in requests per second (rp/s). A higher rp/s rate leads to a faster server crash. Layer 3 (L3) attacks This type of attack takes advantage of flaws within the network layer protocols of the OSI model, and commonly include: Ping flood attacks – in this type of attack, the attacker attempts to crash the server by flooding it with ping requests until the server crashes. Smurf DDoS attack – early implementations of ICMP inherently had poor validation measures which made it easy for an attacker to spoof an IP address in an ICMP request. When using this in the context of a DDoS attack, the attacker sends a series of ping requests to thousands of servers. Once the requests are sent, the attacker then spoofs the target IP address and directs the response to the target system rather than their own IP address. However, this is now an uncommon attack vector as modern infrastructure is rarely vulnerable to this type of attack. ICMP ping of death attack – in this type of attack, the attacker deliberately sends ping requests with IP packets that are greater in size than the maximum allowed by the IP protocol. The network infrastructure, in accordance with TCP/IP, then divides the whole request into chunks that are processed and sent in request fragments. However, when all these fragments are combined together at the target server for processing, the server is unable to handle the computational demand and crashes. For this reason, many modern servers block ICMP requests altogether to prevent any variations of this attack getting through. Layer 4 (L4) attacks In this type, the objective is to increase the number of packets sent in under a second to the point that the server crashes. The impact is measured by packet per second (pp/s). In this type of attack, a service running on the server is usually targeted rather than the server itself. SYN DoS/DDoS attack – in this attack, the attackers exploit the ‘three-way hand shake’. This is done by exhausting the target’s resources by waiting through the TCP Timeout period one after another. In this way, the TCP handshake is never completed and the buffer is full of other waiting TCP handshake requests coming from legitimate sources. Eventually, the server crashes due to timeout requests and an overwhelming amount of resource consumption. SYN-ACK flood attack – In this attack, the attacker attempts to occupy all the space in a target server state space by sending SYN requests from spoofed IP addresses. Once the state space of the target server is full, it has insufficient free space to handle legitimate requests. Layer 7 (L7) attacks These attacks are prevalent on the application layer of the OSI model, and commonly include: HTTP Flood attack – in this attack, the upper layer of the of the target application is overwhelmed with HTTP requests. The attacker exhausts the server’s resources by saturating it with GET and POST requests for images, files, or some other asset from a targeted server. DNS amplification attack – in this attack, the goal is to flood a target with fake DNS lookup requests from open DNS servers that consume network bandwidth to the point that the site fails. WordPress DoS/DDoS attack – in this attack, the concept of XML-RPC (Remote Procedure Call) is leveraged by an attacker. The pingback feature is exploited when multiple links can be generated from a compromised host which sends more requests for data from a WordPress server than the target can handle. The impact to companies DDoS attacks create more requests to a server than it can handle, thus refusing the incoming requests of legitimate traffic and customers. As websites are the common gateway to a company’s services, disrupting a web server’s capacity to handle requests can temporarily take a company offline. Where business operations might rely on consistent uptime and availability, service disruptions and outages have the potential to cause significant financial loss and may result in irreparable reputational damage. This not only affects customer trust and satisfaction, but also makes customers move away from online platforms which ultimately brings the customer base down. With a lower customer base, companies are likely to experience reduced revenue and lower conversions. This is why DDoS mitigation measures and techniques are important to consider befoe they occur. DDoS Attack Detection Detection of an active DDoS attack depends on the attack size, the targeted server’s capacity to handle the incoming requests, and whether or not any uptime or performance monitoring systems are already deployed to trigger alerts. The immediate impact of attacks can range from intermittent performance issues such as slow form submissions and page rendering, to issuing 503 error responses and complete server crashes. The following server-side commands can help manually investigate and identify an attack source: Windows: netstat -ano Lists all the listening ports and their connections to remote IPs. netstat -ano | find /i /c “80” Lists all active IP connections being made to port 80. netstat -na 1 | find “{216.58.204.238}” Lists all active IP connections being made to 216.58.204.238 (example IP). Linux: netstat -ntu | awk '{print $5}' | cut -d: -f1 | sort | uniq -c | sort -n Lists the number of connections each IP address makes to the server. netstat -anp | grep 'tcp|udp' | awk '{print $5}' | cut -d: -f1 | sort | uniq -c | sort -n Lists the number of connections the IP's are making to the server using the TCP or UDP protocol. netstat -ntu | grep ESTAB | awk '{print $5}' | cut -d: -f1 | sort | uniq -c | sort -nr Lists only ESTABLISHED connections, and displays the number of connections for each IP address. netstat -plan | grep :80 | awk {'print $5'} | cut -d: -f 1| sort | uniq -c| sort -nk 1 Shows a list of IP addresses and the number of connections that are connecting to port 80 on the server. For coordinated attacks from multiple source IP addresses, it can be useful to determine whether or not the connections originate from common subnets (such as /16 or /24). netstat -ntu | awk '{print $5}' | cut -d: -f1 -s | cut -f1,2 -d'.' | sed 's/$/.0.0/' | sort | uniq -c | sort -nk1 -r Lists any connected IPs from the same /16 subnet which start with the same two octets (e.g. 216.58.xxx.xxx), and their number of connections. netstat -ntu | awk '{print $5}' | cut -d: -f1 -s | cut -f1,2,3 -d'.' | sed 's/$/.0/' | sort | uniq -c | sort -nk1 -r Lists any connected IPs from the same /24 subnet which start with the same three octets (e.g. 216.58.204.xxx), and their number of connections. DDoS Attack Mitigation DDoS attacks require proficient knowledge and understanding of network security controls to properly mitigate. Once an attack has been detected and the abusing IP address identified, manual steps can be taken to block it. route add 216.58.204.238 reject Blocks 216.58.204.238 from reaching the server. route -n | grep 216.58.204.238 A way to validate if the block was successful. For automation, one or more of following solutions are typically deployed in hardened network infrastructure: Intelligent DDoS mitigation solutions – these are full-fledged systems which deflect and absorb large DDoS attack requests. The providers of such solutions provide load balancing with a scalable distributed architecture and allow for assets to be served through a CDN. Blocking bad traffic – protocol attacks can be mitigated by blocking bad traffic and allowing only legitimate traffic from authorised hosts. These solutions rely on a system which determines from request characteristics whether or not specific requests to a server are legitimate (through a human) or an attack (through a bot or automated). Absorbing based mitigation –a global network of scrubbing centers that scale, as needed and on demand, to absorb large-scale DDoS attacks. DDoS protection for IPs – this mitigation technique includes protection for L7 attacks that target particular websites or services hosted in the cloud. This is now mostly adopted by companies who opt to use cloud-based infrastructure. DDoS protection for websites – this approach detects malicious requests to web servers and protects websites against L7 attacks that target web applications. It is achieved by deploying a cloud-based web application firewall (WAF) to block malicious bots and requests. DDoS protection for networks – this type of solution protects networks with high-packet processing capabilities which mitigate some of the largest DDoS attacks. The various deployment models used to achieve this mitigation technique include GRE tunnels, Equinix Cloud Exchange, and Cross Connect. A flow-based monitoring approach is also deployed where needed along with support for automatic switch-over. Top DDoS Mitigation Solution Providers There are a number of third-party providers on the market offering high-capacity on-demand DDoS attack prevention, detection, and mitigation services. Below are some of the top providers I would recommend. Cloudflare Akamai Imperva