Making Sense of the Information Security Landscape

There are hundreds of different information security solutions out there and choosing which one to pick can be hard. Usually decisions are driven by recommendations, vendor familiarity, successful upsells, compliance needs, etc. I’d like to share my understanding of the security landscape by providing one-line descriptions of each of the different categories of products.

Note that these categories are not strictly defined sometimes and they may overlap. They may have evolved over time and a certain category can include several products from legacy categories. The explanations will be simplified. For a generalization and summary, skip the list and go to the next paragraph. This post aims to summarize a lot of Gertner and Forester reports, as well as product data sheets, combined with some real world observations and to bring this to a technical level, rather than broad business-focused capabilities. I’ll split them in several groups, though they may be overlapping.

Monitoring and auditing (more info)

  • SIEM (Security Information and Event Management) – collects logs from all possible sources (applications, OSs, network appliances) and raises alarms if there are anomalies. Also, allows for forensic investigations into security incidents.
  • IDS (Intrusion Detection System) – listening to network packets and finding malicious signatures or statistical anomalies. There are multiple ways to listen to the traffic: proxy, port mirroring, network tap, host-based interface listener. Deep packet inspection is sometimes involved, which requires sniffing TLS at the host or terminating it at a proxy in order to be able to inspect encrypted communication (especially for TLS 1.3), effectively doing an MITM “attack” on the organization users.
  • IPS (Intrusion Prevention System) – basically a marketing upgrade of IDS with the added option to “block” traffic rather than just “report” the intrusion.
  • UEBA (User and Entity Behavior Analytics) – a system that listens to system activity (via logs and/or directly monitoring endpoints for user and system activity, including via screen capture) that tries to identify user behavior patterns (as well as system component behavior patterns) and report on any anomalies and changes in the pattern, also classifying users as less or more “risky”. Recently UEBA has been part of next-gen SIEMs
  • SUBA (Security user Behavior Analytics) – same as UEBA, but named so after the purpose (security) rather than the entities monitored. Used by Forester (whereas UEBA is used by Gartner)
  • DAM (Database Activity Monitoring) – tools that monitor and log database queries and configuration changes, looking for suspicious patterns and potentially blocking them based on policies. Implemented via proxy or agents installed at the host
  • DAP (Database Audit and Protection) – based on DAM, but with added features for content classification (similar to DLPs), vulnerability detection and more clever behavior analysis (e.g. through UEBA)
  • NTA (Network Traffic Analysis) – monitor network traffic to spot anomalies and threats
  • NDR (Network Detection and Response) – uses NTA and adds machine learning, SOAR integrations and other nice-to-haves.
  • FIM (File Integrity Monitoring) – usually a feature of other tools, FIM is constantly monitoring files for potentially suspicious changes
  • SOC (Security Operations Center) – this is more of an organizational unit that employs multiple tools (usually a SIEM, DLP, CASB) to fully handle the security of an organization.

Access proxies

  • CASB (Cloud Access Security Broker) – a proxy (usually) that organizations go through when connecting to cloud services that allow them to enforce security policies and detect anomalies, e.g. regarding authentication and authorization, input and retrieval of sensitive data. CASBs may involve additional encryption options for the data being used.
  • CSG (Cloud Security Gateway) – effectively the same as CASB
  • SWG (Secure Web Gateway) – a proxy for accessing the web, includes filtering malicious websites, filtering potentially malicious downloads, limiting uploads
  • SASE (Secure Access Service Edge) – like CASB/CSG, but also providing additional bundled functionalities like a Firewall, SWG, VPN, DNS management, etc.

Firewalls

  • WAF (Web Application Firewall) – a firewall (working as a reverse proxy) that you put in front of web applications to protect them from typical web vulnerabilities that may not be addressed by the application developer – SQL injections, XSS, CSRF, etc.
  • NF (Network Firewall) – the typical firewall that allows you to allow or block traffic based on protocol, port, source/destination
  • NGFW (Next Generation Firewall) – a firewall that combines both network firewall, (web) application firewall and providing analysis of the traffic thus detecting potential anomalies/intrusions/data exfiltration
  • DBFW (Database Firewall) – a firewall that sits between applications and the database, preventing SQL injections and buffer overflows, also logging all queries. Similar to DAMs and DAPs

Data protection

  • DLP (Data Leak Prevention / Data Loss Prevention) – that’s a broad category of tools that aim at preventing data loss – mostly accidental, but sometimes malicious as well. Sometimes involves installing an agent in each machine, in other case it’s proxy-based. Many other solutions provide DLP functionality, like IPS/IDS, WAFs, CASBs, but DLPs are focused on inspecting user activities (including via UEBA/SUBA), network traffic (including via SWGs), communication (most often email) and publicly facing storage (e.g. FTP, S3), that may lead to leaking data. DLPs include discovering sensitive data in structured (databases) and unstructured (office documents) data. Other DLP features are encryption of data at rest and tokenization of sensitive data.
  • ILDP (Information Leak Detection and Prevention) – same as DLP
  • IPC (Information Protection and Control) – same as DLP
  • EPS (Extrusion Prevention System) – same as DLP, focused on monitoring outbound traffic for exfiltration attempts
  • CMF (Content Monitoring and Filtering) – part of DLP. May overlap with SWG functionalities.
  • CIP (Critical Information Protection) – part of DLP, focused on critical information, e.g. through encryption and tokenization
  • CDP (Continuous Data Protection) – basically incremental/real-time backup management, with retention settings and possibly encryption

Vulnerability testing

  • RASP (Runtime Application Self-protection) – tools (usually in the form of libraries that are included in the application runtime) that monitor in real-time the application usage and can block certain actions (at binary level) or even shut down the application if a cyber attack is detected.
  • IAST (Interactive Application Security Testing) – Similar to RASP, the subtle difference being that IASP is usually used in pre-production environments while RASP is used in production
  • SAST (Static Application Security Testing) – tools that scan application source code for vulnerabilities
  • DAST (Dynamic Application Security Testing) – tools that scans web applications for vulnerabilities through their exposed HTTP endpoints
  • VA (Vulnerability assessment) – a process helped by many tools (including those above, and more) for finding, assessing and eliminating vulnerabilities
  • Vulnerability scanners – tools that scan the organization infrastructure for (potential) vulnerabilities.

Identity and access

  • IAM (Identity and Access Management) – products that allow organizations to centralize authentication and enrollment of their users, providing single-sign-on capabilities, centralized monitoring authentication activity, applying access policies (e.g. working hours), enforcing 2FA, etc.
  • SSO – the ability to use the same credentials for logging into multiple (preferably all) applications in an organization.
  • WAM (Web Access Management) – the “older” version of IAM, lacking flexibility and some features like centralized user enrollment/provisioning
  • PAM (Privileged access management) – managing credentials of privileged users (e.g. system administrators). Instead of having admin credentials stored in local password managers (or worse – sticky notes or files on the desktop), credentials are stored in a centralized, protected vault and “released” for use only after a certain approval process for executing a given admin task, in some cases monitoring and logging the executed activities. The PAM handles regular password changes. It basically acts as a proxy (though not necessarily in the network sense) between a privileged user and a system that requires elevated privileges.

Endpoint protection

  • AV (Anti-Virus) – the good old antivirus software that gets malicious software signatures form a centrally managed blacklist and blocks programs that match those signatures
  • NGAV (Next Generation Anti-Virus) – going beyond signature matching, NGAV looks for suspicious activities (e.g. filesystem, memory, registry access/modification) and uses policies and rules to block such activity even from previously unknown and not yet blacklisted programs. Machine learning is usually said to be employed, but in many cases that’s mostly marketing.
  • EPP (Endpoint Protection Platform) – includes NGAV as well as a management layer that allows centrally provisioning and managing policies, reporting and workflows for remediation
  • EDR (Endpoint Detection and Response) – using an agent to collect endpoint (device) data, centralize it, combine it with network logs and analyze that in order to detect malicious activity. After suspected malicious activity is detected, allow centralized response, including blocking/shutting down/etc. Compared to NGAV, EDR makes use of the data across the organization, while NGAV usually focuses on individual machines, but that’s not universally true
  • ATP (Advanced threat protection) – same as EDR
  • ATD (Advanced threat detection) – same as above, with just monitoring and analytics capabilities

Coordination and automation

  • UTM (Unified Threat Management) – combining multiple monitoring and prevention tools in one suite (antivirus/NGAV/EDR), DLP, Firewalls, VPNs, etc. The benefit being that you purchase one thing rather than finding your way through the jungle described above. At least that’s on paper; in reality you still get different modules, sometimes not even properly integrated with each other.
  • SOAR (Security Orchestration, Automation and Response) – tools for centralizing security alerts and configuring automated actions in response. Alert fatigue is a real thing with many false positives generated by tools like SIEMs/DLPs/EDRs. Reducing those false alarms is often harder than just scripting the way they are handled. SOAR provides that – it ingests alerts and allows you to use pre-built or custom response “cookbooks” that include checking data (e.g. whether an IP is in some blacklist, are there attachments of certain content type in a flagged email, whether an employee is on holiday, etc.), creating tickets and alerting via multiple channels (email/sms/other type of push)
  • TIP (Threat Intelligence Platform) – threat intelligence is often part of other solutions like SIEMs, EDRs and DLPs and involves collecting information (intelligence) about certain resources like IP addresses, domain names, certificates. When these items are discovered in the collected logs, the TIP can enrich the event with what it knows about the given item and even act in order to block a request, if a threat threshold is reached. In short – scanning public and private databases to find information about malicious actors and their assets.

Email

  • SEG (Secure email gateway) – a proxy for all incoming and outgoing email that scans them for malicious attachments, potential phishing and in some cases data exfiltration attempts.
  • MFT (Managed File Transfer) – a tool that allows sharing files securely with someone by replacing attachments. Shared files can be tracked, monitored, audited and scanned for vulnerabilities, and access can be cut once the files was downloaded by the recipient, reducing the risk of data leaks.

DDoS

  • DDoS mitigation/protection – services that hide your actual IP in an attempt to block malicious DDoS traffic before it reaches your network (when it’s too late). They usually rely on large global networks an data centers (called “scrubbing centers”) to send clean traffic to your servers.

Compliance

  • GRC (Governance, Risk and Compliance) – a management tool for handling all the policies, audits, risk assessments, workflows and reports regarding different aspects of compliance, including security compliance
  • IRM – allegedly, philosophically different and more modern and advanced, in reality – the same as GRC with some additional monitoring features

So let’s summarize the ways that all of these solutions work:

  • Monitoring logs and other events
  • Inspecting incoming traffic and finding malicious activities
  • Inspecting outgoing traffic and applying policies
  • Application vulnerability detection
  • Automating certain aspects of the alerting, investigation and response handling

Monitoring (which is central to most tools) is usually done via proxies, port mirroring, network taps or host-based interface listeners, each having its pros and cons. Enforcement is almost always done via proxies. Bypassing these proxies should not be possible, but for cloud services you can’t really block access if the service is accessed outside your corporate environment (unless the SaaS provider has an IP whitelist feature).

In most cases, even though machine learning/AI is advertised as “the new thing”, tools make decisions based on configured policies (rules). Organizations are drowned in complex policies that they should keep up to date and syncrhonize across tools. Policy management, especially given there’s no indsutry standard for how policies should be defined, is a huge burden. In theory, it gives flexibility and should be there, in practice it may lead to a messy and hard to manage environment.

Monitoring is universally seen as the way to receive actionable intelligence from systems. This is much messier in reality than in demos and often leads to systems being left unmonitored and alerts being ignored. Alert fatigue, which follows from the complexity of policy management, is a bug problem in information security. SOAR is a way to remedy that but it sounds like a band-aid on a broken process rather than a true solution – false alarms should be reduced rather than being closed quasi-automatically. If handling an alert is automatable, then tha tool that generates it should be able to know it’s not a real problem.

The complexity of the security landscape is obviously huge – product categories are defined based on multiple criteria – what problem they solve, how they solve it, or to what extent they solve it. Is a SIEM also a DLP if it uses UEBA to block certain traffic (next-gen SIEMs may be able to invoke blocking actions even if requiring another system to carry it out). Is a DLP a CASB if it does encryption of data that’s stored in cloud services? Should you have an EPP and a SIEM, if the EPP gives you good enough overview of the events being logged in your infrastructure? Is a CASB a WAF for SaaS? Is a SIEM a DAM if it supports native database audit logs? You can’t answer these questions at a category level, you have to look at particular products and how well they implement a certain feature.

Can you have a unified proxy (THE proxy) that monitors everything incoming and outgoing and collects that data, acting as WAF, DLP, SIEM, CASB, SEG? Can you have just one agent that is both a EDR, and a DLP? Well, certainly categories like SASE and UTM go in that direction, trying to ease the decision making process.

I think it’s most important to start from the attack targets, rather than from the means to get there or from the means to prevent getting there. Unfortunately, enterprise security is often driven by “I need to have this type of product”. This leads to semi-abandoned and partially configured tools for which organizations pay millions. Because there is never enough people to be able to go into the intricate details of yet another security soluion, and organizations rely on consultants to set things up.

I don’t have solutions to the problems stated above, but I hope I’ve given a good overview of the landscape. And I think we should focus less on “security products” and more on “security techniques” and on people that can implement them. You don’t have a billion dollar corporation to sell you a silver bullet (which you can’t fire). You need traind experts. That’s hard. There aren’t enough of them. And the security team is often undervalued in the enterprise. Yes, cybersecurity is very important, but I’m not sure whether this will ever get enough visibility and be prioritized over purely business goals. And maybe it shouldn’t, if risk is properly calculated.

All the products above are ways to buy some feeling of security. If used properly and in the right combination, it can be more than a feeling. But too often a feeling is just good enough.

1 thought on “Making Sense of the Information Security Landscape”

Leave a Reply

Your email address will not be published. Required fields are marked *