Category Archives: Security Learning

Random Thoughts, Security Learning

How good is our current Security Strategy?

Gallery 4 Comments

Few years ago, none of the “Hacktivist Groups” existed or even if they did, they lurked in the underworld. But today, they have the guts to come out in public and declare war on the Internet. They have also been very successful in bringing big corporations loss in terms of data and money. And how much wager would you like to place that this is just the beginning. This begs us to the very question – How good is our current Security Strategy? 

Traditionally we have been building the Security Regime using one or both of the approaches in tandem: Known Bad Security (Blacklisting) and Known Good Security (Whitelisting). To all those signature and behavior based thinkers, don’t fret, for this approach is a superset of Signature and Behavior based approach.

First let us look at Known Bad Security or Blacklisting:
One of the things we are very good at is, “KNOWN BAD” detection and response. By this I mean, we are good at identifying Vulnerabilities based on Vendor releases, patching them once the vendor releases the patches, updating AV/AS, IDS/IPS, Content Filtering etc to protect against exploitation. This is what “KNOWN BAD” Security is all about. You know it is bad and you defend against it. But a recent survey by Verizon shows that only 1% of the total data breaches are identified by IDS/IPS or AV solutions. This is a clear indicator that Signature based detection or Blacklisting based response is not giving us the results. So even though we are very good at Known Bad Security, we are being compromised day in and day out!!!!

Known Good Security or Whitelisting is just the opposite of Known Bad:
By this I mean, we identify and maintain a list of KNOWN GOOD items in our IT Infra. What connections are good, What users are good, What files are good, What is allowed, What is unauthorized etc as our data points for Known Good Security. Based on this data, we identify Security Abnormalities, anomalous pattern detection etc that don’t conform to the Whitelist and go after them as Rogues/Attackers. We investigate them, if found bad follow remediation process for them or if found good add them to the Whitelist. Once we know What is bad, we automate it by feeding to the Blacklist detection and Response. This while being effective is a slow and tedious process thereby giving gracious amounts of time for an attacker to wreak havoc.

Some Good, Some Bad:
Most of the Enterprises today effectively use a combination of Blacklisting and Whitelisting to achieve their Information Security needs. But based on the threats being propagated today, we can say with enough confidence that this approach is failing. The main reason for the failure is that, “Actual Good and Actual Bad are way more than Known Good and Known Bad”. Since we are unable to quantify these numbers scientifically, we end up doing good of nothing.

What we lack?
Our current strategy towards security has some gaping holes. Some of them are listed below:

  • Over Relying on External Sources: We still rely on Vendor input, community input and other public disclosures to define Blacklisting. One vendor’s threat detection efficiency is different from the other. One vendor might rate a Malicious Code as High Severity, but the other may rate it as Low. This kind of disparity does not help in determining what is “Actually bad”.
  • Poor Knowledge of our environment: How many times have you identified a Security incident and while investigating found out something new about the environment. I can bet that it is literally every time. Without knowing the exact nature of our environment, we would not be able to do any effective Whitelisting. Without effective Whitelisting, effective Blacklisting also is impacted
  • One cure for All diseases: We think that if one organization is compromised by a specific exploit, it is applicable to all. We seldom think or evaluate the Controls we have may differ significantly from the controls other organizations have. Security should be tailored to suit not vice-verse.
  • Once we Whitelist something, we never re-evaluate. We perceive that “it is clean” and pay little attention till hell breaks loose. This is more related to human nature than anything else I guess. Once we “move past” we never look back. This will hurt us because, a whitelist today might turn bad tomorrow due to IT dynamics, thereby leading to an exploit.
  • We live by and die by More tools – more security, Latest signatures – more protection, more resources – more coverage, more training – more knowledge. Most organization just buy Security tools or technologies to fill a check box in their Audit/Compliance needs. If the company execs have caught wind of some Security attack that happened at some other company, they are paranoid that it will happen to them as well. Hence “Gimme more security” approach.
  • More the HooHa More Serious the Threat: The amount of publicity received is directly proportional to the severity of the threat. We would have several other threats in our environment, several gaps we need to fix, but we would still look for the famous Conficker, Flame, Stuxnet, Aurora, Zeus etc. Even though some of them were big in terms of spread, every organization had different infection rates.
  • We still think Security as a “Operations function“. We still go by the number of Alerts worked, number of incident raised, time to solve, time to respond etc. Security is more Analytical, investigative field. Looking beyond the noise, finding the needle in the haystack, attacker attribution and all sound cool on paper, but to bring it to reality the current strategy doesn’t help.
  • Security is not a culture. In everyday life, you lock the front door, keep important things in a safe, put on safety gear, wear a seat belt etc, but we don’t treat our IT systems development, implementation and management with a Security mindset. Bad products are developed, bad implementations happen, bad administration and monitoring happen, and finally mistakes from people too happen, leading to a Security breach, data theft and loss.

I am sure there is more than the above list in terms of flaws in our current strategy. What do you think? Please comment on!!!

 

Random Thoughts, Security Learning, What you need to know?

SIEM Use Cases – What you need to know?

Gallery 11 Comments

My previous post “Adopting SIEM – What you need to know” would give a better starting point if you are new to SIEM and want to implement it in your organization. If you already use/manage/implement a SIEM, then read on.
To start with, SIEM tools take a lot of effort to implement. Once implemented, they need to be taken care like babies. If care is not given, within a few months you would be staring at a million dollar museum artifact. Now there are two parts of care:

  1. Making sure that the systems are updated regularly, not only for patches and configurations but also the content put in them.
  2. Second and the most important part is making the SIEM relevant to the current Threat Landscape.

Anyone who has worked on SIEM for some time would agree with me, that Administration is generally easier compared to making the system relevant to the Threat Landscape. Before people hit me with “Administration is also a pain”, I would like to offer a defense saying that mostly, all SIEM products have documentation attached that give fair amount of information on how to install, update, upgrade and operate these systems. However, Translating Threat Landscapes to nuts and bolts for SIEM purposes is the biggest challenge and there are no guides that can help do that.

In this blog post, my attempt is to make this translation as easy as possible. In SIEM parlance, we call the translation as a Use Case. If there is well-defined Use Case, implementing them, responding to them and managing them would become easier. Such Use Cases would eventually become the cornerstone on which a SOC (Security Operations Center) is built. As usual, I would like to start with defining a Use Case, running through its stages and then finally wrapping it up with an example. So here we go.

Use Case Definition: A Use Case by definition is nothing but a Logical, Actionable and Reportable component of an Event Management system (SIEM). It can be either a Rule, Report, Alert or Dashboard which solves a set of needs or requirements.

A Use Case is actually “developed” and this development is a complete process and not just a simple task. Like a mini project it has several stages. The various Stages involved in Use Case Development are as follows:

  • First stage is the “Requirements”Definition. It can be any of the following high level requirements and is unique to every company:
    1. Business
    2. Compliance
    3. Regulatory
    4. Security
  • Once the requirements are finalized, the next stage would be to “Define the scope” of the requirement. This would typically mean the IT Infrastructure that needs to be protected and is a high priority for the specific requirement.
  • Once the scope is finalized, we can sit down and list the “Event Sources” that would be required to implement the Use Case. These would be Log Data, Configuration Data, Alert Data etc coming out of IT Systems under the above Requirements Scope.
  • The next stage would be to ensure that the Event Sources are going through “Validation Phase” before use. Many times, we would have an Event source but the required data to trigger an Event may not be available. This needs to be fixed before we proceed with the Use Case development.
  • Post validation, we need to “Define the Logic”. This is where we exactly define what and how much data is needed to alert along with the Attack Vector we would like to detect.
  • Use Case “Implementation and Testing” is the next stage. This is where we actually configure the SIEM to do what it does best – Correlation and Alerting. During Implementation the definition of the desired output can also be done. The output can be one of the following:
    1. Report
    2. Real Time Notification
    3. Historical Notification
  • Once implementation is done, we need to “Define Use Case Response” procedures. These procedures help you to make the Use Case Operational.
  • Finally, Use Case “Maintenance” is an ongoing process to keep the Use Case relevant by appropriate tuning.

Now that we have defined in detail the Use Case Development methodology, it is time to take an example and see how this actually looks in Real Life Implementation terms.

The Requirement: Outbound Spam Detection.
The Scope: Mail Infrastructure, End User Machine, Security Detection Infrastructure
The Event Source:
  • IDS/IPS at Network and Host – Signature Based Detection
  • Mail Hygiene or Mail Filtering Tools – Signature Based Detection
  • Events from Network Devices – Traffic Anomaly Based Detection
  • Events from End User Detection tools – Signature and Traffic Anomaly Based Detection

The Event Validation: The devices logging to SIEM should be normalized and parsed properly. Typically, SIEM products would allow Content development based on their native Field Mappings (Through Parsing). If the fields are not mapped, then the SIEM does a poor job of Event Triggering and Alerting. The required fields for the above Use Case would typically be Source IP, Source user ID, Email Addresses, Target IP, Host information of Source and Target, Event Names for SPAM detection, Port and Protocol for SMTP based traffic detection etc.

Use Case Logic Flow: The Logic definition is something unique to the environment and needs to be defined accordingly. The logic can be either Signature based or behavior based. You can have it restricted to certain subset of data (based on the Event Sources above) or expand it to be more generic. Some samples are given below:
  • One machine doing Port 25 Outbound connections at the rate of 10 in a minute
  • SPAM Signatures originating from the same source from IDS/IPS, Mail Filter etc having the same destination Public domain
  • SYN Scans on port 25 constantly from a single source etc

Implementation and Testing: Once the logic is defined, Configuration of SIEM and tuning the implementation to trigger more accurately is the next phase. After Implementation of the Use Case, we would need several iterations of Incident Analysis along with data collection to ensure that the Use Case is doing what it is intended to do. This is done at the SIEM level and may involve aggregation, threshold adjustments, logic tightening etc.

Use Case Response: After implementation, the Use Case need to be made as a valuable resource by Defining a Use Case Response. This is the stage where you would define “What action needs to be taken and how it needs to be taken”. You can look at Episode 4 of my Security Investigation series to get an idea of how to Investigate SPAM cases. Other Security Investigation Series Articles are located here – Security Investigation Series.

SIEM Use Cases are really the starting point for good Incident detection. If you want to run a SOC, having well-defined SIEM Use Cases would ease management and increase efficiency of Operations. This post is my humble attempt to simplify and regularize Use Case development for SIEM implementations.

As always, I would love to hear comments and thoughts on this topic.

Random Thoughts, Security Learning, What you need to know?

Adopting SIEM – What you need to know?

Gallery 13 Comments

“SIEM stands for Security Information and Event Management”

Oh wait, I have heard of SIM, I have heard of SEM, what is this SIEM??.
Originally, the Security Information Management (SIM) and Security Event Management (SEM) systems were two different technologies performing similar but distinct functions. Gartner in 2005, coined the term SIEM to encompass both. As the name suggests, It is nothing but a collection of tools and technologies to manage Incident and Events pertaining to Security alone. Some of the tell-tale capabilities of a typical SIEM platform are:

  1. Collect Logs from various Log Sources/Devices
  2. Store these logs for a decent amount of time
  3. Provide Fast Search/Retrieval capabilities
  4. Provide meaningful interpretation of Log received
  5. Provide capabilities to correlate between logs of different devices
  6. Basic Ticketing/Alerting capabilities.

The first 4 points are typical of a SIM and the remaining 2 are typical of a SEM.
Any tool that does all of these is a SIEM. There are more than 50 different products that cater to the SIEM space. Just like any other product, they cater to various market segments at various price points.
If you Google for SIEM reviews you would get a lot of information on various products. In my experience, I have worked with at least 4 SIEM vendors. Each one of them have their own pros and cons. Comparing a product in a DEMO and comparing it after use are two different things. So, in this blog post, I am going to highlight few things as “What you need to know” when you are planning to adopt SIEM technology

  1. Have a defined Logging process in your environment. This is very crucial because a SIEM is useless without a good Logging Program. This not only helps in making the SIEM implementation easier, but also helps in getting a measure of the volume you are dealing with. In my experience, often times, despite having an Industry leading SIEM, the log Management made it look pedestrian and a waste of money.
  2. Every SIEM vendor has something called as Collector/Connector/Receiver/Agent that collects logs from the devices and converts them to their proprietary format. This conversion or parsing as we call is important for the product developers to store data in a format they can understand and process quickly. Most of the vendors offer something of a Custom Collector/Parser development for their “unsupported” log sources. This costs money, skills in-house and may require regular maintenance. Hence Native Parsing Support for Log Sources is better. Establish this before you move ahead with SIEM implementations. Either source a in-house resource to help build and manage such customizations or spend more money to get the vendor to do it.
  3. Identify primary focus areas from an Organizational perspective. This will help you configure your SIEM solutions appropriately. These focus areas should be broadly classified and then expanded to the ground level. For example, if your requirement is compliance, start with control requirements, see what logs need to be collected to fulfill them, see how integration needs to be done, see what needs to be reported, alerted, retained, etc.
  4. Get a dedicated SIEM administrator or rather train someone in-house to be that person. This is very important because, in my experience I have always felt that SIEM is as good as the administrator is. Without proper maintenance and care, it will decay over time. If you really need to generate value out of it, manage it well. By managing a SIEM I mean not only the system itself but also the ecosystem it resides in.
  5. Understand that SIEM alone cannot solve all your Security Problems. It is NOT A MAGIC WAND. If setup and configured correctly, a SIEM can at best point you in the right direction, a direction where you can identify and fix several security issues in your enterprise thereby strengthening it. So, be prepared to have a Response/Remediation team that will investigate the alerts generated and take appropriate action.
  6. Correlation is a vital part of SIEM offerings. Before Adopting SIEM, make sure you understand and possibly catalog the various Attack Vectors, Threat Scenarios you would want looked at for correlation in your organization. This will give a fair direction for the basic rules you would put in place to start with. Once you are comfortable and start seeing the various alerts generated, you can play around and experiment more. In my experience, start with built-in rules, understand them, investigate them, tune them and then slowly start building your own content. For more details on the various rules available in SIEM Look at Rules Rule in SIEM Kingdom
  7. Architecture wise, make sure your SIEM solutions are in tandem with your Logging solutions. Also, build your SIEM as modular as possible thereby making upgrades, technology refresh etc seamless.
  8. Don’t forget the filtering aspect. Correlation Engines will perform faster and will get you better results if they are attacking a smaller set of “known bad” logs rather than all. This is crucial in large enterprises as the Log Volume can easily overwhelm the SIEM systems. Note: Many SIEM tools have limitations in the number of events they can process. This is denoted in Events Per Second (EPS). Even though the vendors advertise several thousands, an effective correlation system can have only around 2000 – 5000 EPS tops. Anything more will make your system painstakingly slow. So understand and work through this. Look at my posts What and How much to Collect and High Log Volume – What to Filter and What to Keep? to get more information on how to log, what to log and what to filter.
  9. Remember, more processing layers, less EPS. This means that the Log Collection layer will have more EPS processing capability than the Correlation engine and so on. Visualize it as a pyramid with the Log Collection at the Base and the Correlation at the top
  10. Last but not the least, “Stay Alert and Eager. The Logs Don’t Lie”

Hope this post helped you in getting a fair idea of SIEM technologies. I have worked on HP ArcSight, Symantec SSIM, Novell E-Sentinel. If you need details about them in terms of practical setup, configuration, architecture etc, shout out and I will help as much as possible.
 [pdf]Save as PDF[/pdf]