Oracle Security: The Big Picture - Page 3
January 15, 2008
Major Data Theft Incidents
Despite the myriad of regulations governing data security, and a clear increase in focus on security issues in general, there have been a deluge of successful thefts of data on a vast scale. There have been so many incidents made public that the Privacy Rights Clearinghouse was established as a publicly available chronology of data breaches, and a count of the number of personal records that have been disclosed since February 2005. At the time of this writing in May 2007, over 150,000,000 records containing sensitive personal information have been compromised in several hundred incidents. Check out the latest chronology at www.privacyrights.org/ar/ChronDataBreaches.htm.
The problem is not entirely limited to theft. There have been a significant number of cases where information was disclosed inadvertently, often because of an innocent or silly mistake that seemingly lies outside the realm of data security. There have been far too many cases to cover them all here, too many even to cover just the really interesting ones, so we have chosen four incidents to describe in some detail. These examples are intended to paint a picture of the various ways in which sensitive data has become compromised in the recent past.
CardSystems Solutions--June 2005
In the spring of 2005, a hacker was able to exploit vulnerabilities at CardSystems Solutions in order to gain access to their internal network and retrieve detailed data on approximately 40 million credit card transactions that the company had stored in a database. CardSystems was an Atlanta-based credit card processing company, responsible for handling $15 billion dollars in annual transactions on behalf of more then 100,000 small- to medium-sized businesses. The attack was detected first by a MasterCard fraud detection system when several likely fraudulent transactions were detected. MasterCard was able to trace the cardholder information used back to CardSystems, whom they immediately notified of a possible breach. A short investigation by CardSystems confirmed that their systems had been hacked. While the exact details of the attack remain the secret of the credit card companies, several clues have been disclosed that allow us to piece together the most likely scenario for how the attack worked.
During September 2004, a hacker found vulnerabilities in an Internet-facing application that CardSystems customers used to access data. The attacker was able to gain access to the Web application, likely through an easily guessed password, and then begin the process of looking for ways to access the underlying servers and databases directly. The most common method for using a Web application to gain access to internal databases and other systems is via Structured Query Language (SQL) injection. The attacker was able to locate points in the application where weak input validation allowed him to inject SQL into some forms, and interact directly with the database that housed the application data. From there, the attacker gained access to the database servers operating system, likely using database functions to do so, such as xp_cmdshell on MS SQL Server and Sybase. Since databases generally run with full administrative access on their host servers, once the attacker could access the OS, they assumed full control. From there, a script was created on a server in the CardSystems internal network. The script was designed to search the network for a particular type of file that contains Credit Card Track Data (the data on the magnetic strip on the back of your credit cards), and then send that data to the attacker via File Transfer Protocol (FTP).
In clear violation of the credit card companies data security rules, CardSystems had been storing files that contained complete track data for failed transactions. These files were the ones targeted and stolen by the script. The files contained a complete set of information on each transaction, including the cardholders name, card number, expiration date, and CVV code. With this data, the attacker was able to initiate the fraudulent transactions that led to the attack being detected. Within 60 days of discovering the hack, Visa declared that they would revoke CardSystems authority to process their transactions. American Express quickly followed. CardSystems was to pay the ultimate price for their negligence; they were out of business.
The attack against ChoicePoint was more of a brilliantly executed con than what most of us think of as a hack. ChoicePoint is a data aggregator and information broker, collecting, mining, and selling information on the backgrounds and spending patterns of, well, of everyone. This data is then sold to insurance companies, loan officers, media companies, law enforcement, or really any business looking for background checks on potential employees or customers. The attackers took advantage of ChoicePoints business model and poor customer screening processes to essentially convince them to hand over the data.
Attackers set up approximately 50 fake companies, giving them legitimate names and phone numbers, dreaming up business models, and even establishing false Tax IDs. They used these companies to open accounts with ChoicePoint, who were only too happy to sign up a few more customers and start feeding them data. Over a period of months, these companies requested and received data on 145,000 adults from ChoicePoint, acting and operating in much the same way as any of their legit customers. But these folks were not legit and not interested in targeted marketing campaigns. They were interested in identity theft, in destroying the finances of others for their own personal gains. The breach was eventually uncovered, but the exact data collected could not be determined. This is when things got ugly for ChoicePoint.
Making the mistake of allowing these accounts to be set up was shameful, and punitive action was certainly unavoidable, but when the prosecuting attorneys realized that ChoicePoint had no way to determine what records had been accessed, they went out for blood. ChoicePoint executives were subpoenaed to testify before Congress, where many difficult questions were asked about the lack of tracking and user auditing within the sensitive information systems that the company maintained. The incident cost ChoicePoint a fortune, kicked off a flurry of legislation governing data mining companies that collect and sell personal information, and made crystal clear to the business world that if you are going to store personal information, youd better take steps to protect it and ensure that you properly track access to it, and retain the audit logs.
Currently on record as the largest single incident of data theft of all time, information on nearly 50 million credit card transactions was disclosed to what appears to be a group of professional computer criminals. The attack was technically complex, devastatingly effective, and was in place providing data to the attackers for nearly two years before it was noticed. In a 2006 research study by the Ponemon Institute, 31 companies that had experienced a data breach were analyzed, and the cost to the company of each compromised record was estimated at $182.00. Using that number as a guideline, TJX could be forced to pay out nearly $9,000,000,000. This staggering figure does not include the less tangible costs such as loss of business, loss of productivity, brand damage, or loss of market capitalization. With risks on this scale, it is hard to believe that companies dont do more to secure their data. Hopefully, by following the steps in this book, you can help your organization avoid this ultimate nightmare scenario.
Details of the attack remain somewhat sketchy, but we have a general timeline of events and a sound theory of what went on. TJX first found evidence of unauthorized software on their systems on December 18, 2006. They immediately hired two major consulting firms to perform a detailed analysis of the incident and provide guidance on how to respond. Within a few days, it was becoming clear that the software was indeed malicious and that the intruder responsible continued to have access to critical financial systems within TJXs network, access that dated back to July 2005. The attack was multi-faceted, including elements of stealing files, intercepting communications, and breaking encryption on protected data.
Files stolen contained historical information about payment card transactions. TJX has not been able to completely identify which files were stolen, primarily because they periodically delete these files during the normal course of business. Some of the files that were likely accessed dated back to 2003, when security rules about protecting cardholder information were far more lax than they are today. These older files generally contained clear-text (unencrypted) data, giving the attacker easy access to the credit card details.
Taking the attack to the next level, the attacker was able to gain access to the network used during the payment card transaction approval process. During this process, cardholder information, including names, card numbers, expirations dates, and CVV2 numbers are transmitted to the payment card issuer without encryption. The intruder was able to watch this approval process in action, collecting the credit card data as it traversed the network.
The final blow to TJX was the discovery that the attacker had likely gained access to the software and decryption keys needed to decipher the card data that had been stored, properly encrypted, in their databases. Even when the company believed that it had strong security in place to protect their sensitive data, the attacker was able to find holes in the system and gain full access.
The methods used to initially penetrate the TJX network have not been disclosed, however many experts agree that it likely began with the attacker gaining access to an unprotected wireless network at one of TJXs retail stores. Sitting in a parked car in the parking lot, the attacker could have connected to the wireless network and attacked the system used to manage the stores cash registers. Once that system was breached (probably using a freeware hacker tool downloaded from the Internet), it became a simple manner to connect to the central processing systems in TJX headquarters. Corporate security is often compared to a Tootsie-Pop; hard and crunchy on the outside, soft and chewy in the middle. The analogy refers to strong security at the network perimeter, but little to no security inside the network. Once the hacker found his or her way past TJXs network perimeter, it was game over; weak internal security controls allowed them to steal everything but the kitchen sink.
Department of Veterans Affairs--May 2006
The breach at Veterans Affairs (VA) was an interesting one,
as it demonstrates how a seemingly innocent (but poorly thought through) act
can lead to the biggest inadvertent disclosure of sensitive information in the
history of the
That evening, the analysts house was broken into by a petty burglar looking to steal electronics and other small household valuables. Included among the thiefs booty was a laptop, the very same laptop that came home from work that day, along with that very important hard disk. The next morning, the break-in was noticed and the missing laptop reported to the VA. The data on the laptop was never an issue, but the hard disk was a big deal and while it seemed clear that the data was never the target of the theft, there was no way to know what had happened with the data once it was in criminal hands. A mad FBI search for the laptop began, lasting for about eight weeks before the missing machine and the all important external disk was found. But it was too late; the data was out there and there was no way to prove it had not been accessed or transferred to others (although the FBI did perform a forensic analysis and found no evidence that the data had been touched).
Everyone affected needed to be informed, credit monitoring services were contracted to watch for potential identity theft, and the government took action first requiring the tracking of all data extracts from systems containing sensitive data, and soon thereafter mandating hard drive encryption on all mobile PCs. VA also implemented more training and education programs on data handling policies, but apparently not enough. It was not even three months later before the VA experienced a similar breach, this time involving 18,000 records stolen with a laptop that a contractor for Unisys had taken home for the night.
A Step-by-step Approach to Securing Oracle
This book focuses on a practical, step-by-step approach to securing Oracle databases. Well define three levels of Oracle security and provide you guidance on how to determine what level is appropriate for each of your systems, then explain exactly how to get yourself from where you are today to the right level for you. Well take things a step or two beyond just helping you secure your Oracle databases. We will build a mechanism for tracking your progress, establishing a plan and then demonstrating continuous improvement. We will also create a system to measure and prove the security of your systems to both internal and external compliance officers and auditors, giving you the option of using a checklist and a set of scripts, or automating the process with commercially available tools. Your journey to secure Oracle databases begins here.
Tools & Traps...
One Size Fits All Doesnt Fit Most
Weve seen quite a few companies try to start a comprehensive database security program only to end up nowhere but frustrated. The most common reason for their failure: establishing a one size fits all approach to securing their systems. The database environment within a typical enterprise is extremely diverse, and different systems have different needs. This makes it next to impossible to create one overarching set of standards to which every system must comply. The one size fits all approach can fail in several different ways, but the end result is always the same: no real database security program, and an easy target for any hacker who cares to attack.
The first step in the security program is to establish a plan and a set of guidelines, which is a common point of failure. Some companies will establish a working group in order to build a set of security and configuration guidelines. Problems often arise when the members of the working group, who often represent different functional business areas, are unable to come to agreement on common standards. In the most severe cases, this leads to the database security effort being abandoned before it really begins, leaving each administrator to decide on their own if and how they will protect their data. More commonly, working groups establish a watered down set of standards, leaving plenty of room for the exceptions required by each business area, but also leaving a knowledgeable attacker with open doors into the systems.
Once a set of standards have been established, they are often passed to the database administration teams with little or no guidance on how to implement the requirements. The issue is not that the DBAs dont know how to configure the security settings; they know the databases better then anyone else. The problem lies in prioritization and workload. In organizations with dozens, hundreds, or thousands of databases, the task of implementing a security standard enterprise-wide can easily be overwhelming. When open questions are left about which systems to secure first, or how far to go with each system before moving on, efforts often fizzle out after tightly locking down a small handful of randomly chosen databases. Hackers will use automated tools to search for weaknesses across any database they can access; securing only a few is almost like securing none at all.
With a set of security configuration standards that explain how to secure each database based on business value and risk, and a clear plan that sets out a priority order in which to secure the databases, an enterprise can be extremely successful in implementing database security across the organization. This book will help you build that plan for your Oracle databases, but the lessons learned can easily be translated to secure any database platform.
Appropriate Security For Each Class of Database System
To get started, you will need to assign each of the databases you are responsible for a priority rating based primarily on its business value. This activity should not be performed in a vacuum. Rather you should get together with other stakeholders within your organization, folks such as the database administrators, IT security engineers, application owners, business line managers, and internal auditors to review and assign each system a business value rating. It is often easiest to use a set of categories rather than come up with an absolute priority list. Try using three categories, such as:
Once your databases have been assigned to categories, you can make smart decisions about which to secure first and how far to go in securing each one. Your most critical systems will get the highest level of security, while the least critical systems get the lowest level. This way, each database gets only the security features that it requires, eliminating the excess workload caused by forcing every database to be secured to the same standard of protection. Like a system of building blocks, each category takes over where the last leaves off, building a foundation of security and then expanding it to the needs of the system it is protecting. This allows you to quickly establish a baseline level of security on every database, preventing the vast majority of simple attacks and lowering business risk, then moving on to conquer more complex issues with more rigorous security standards on only the most sensitive databases.
It is not enough to secure your databases; you have to prove it. Achieving compliance with anything from government regulations to internal security standards means providing evidence that you have taken the proper steps to secure your system. To enable you to demonstrate and communicate your Oracle security to whomever you must report to, each level includes a systematic approach to maintaining and assessing compliance. This includes references to checklists, scripts, and tools you can use to automate the assessment and monitoring process.
With a solid understanding of the history and evolution of security features in Oracle, we hope that you can understand how the many different security controls came to be and why they were added to the system. Government and Industry regulations have been established to govern the handling of sensitive personal and financial information, forcing companies to comply or face significant punitive actions. Even with increasing pressure to secure systems, major organizations have experienced high profile breaches, with costs as high as going out of business. Hacking is a type of theft that companies must target with real investments in people, process, and technology.
This book gives you a prescription to secure a single database, or to build an enterprise-wide strategy to lock down your critical Oracle database systems and tightly integrate your database security strategy with your corporate security infrastructure. By establishing a strong Oracle security program, you can play a lead role in ensuring your business meets regulatory requirements, properly protects corporate secrets, and acts as a good custodian of the sensitive personal and financial information that you store.
Solutions Fast Track
A Brief History of Security Features in Oracle
The Regulatory Environment Driving Database Security
Major Data Theft Incidents
A Step-by-step Guide to Securing Oracle
Frequently Asked Questions
Q: What is the top Oracle security issue you see out there today?
A: By far, it is default accounts with default passwords enabled in the database. Read more about that in Chapter 4.
Q: Ive been a DBA for 20 years. Why am I only now being asked to secure my databases?
A: The world has changed. The bad guys have pulled off some pretty significant crimes, so the good guys have put controls in place to try and stop the attackers. Today, your business is faced with increasing regulatory pressure to secure your critical data, while at the same time there are more hackers than ever trying to get at it.
Q: My database is behind a firewall in a secure network, am I protected?
A: Not a chance. Today, most of the bad guys operate from inside your network. More then 70 percent of attacks involve an insider. Also, that firewall is full of holes, designed to allow applications to function. Attackers have found ways to exploit vulnerable applications to gain direct access to the databases.
Q: How much work is involved in securing an Oracle database?
A: Every database is different. Depending on the sensitivity of the data, the environment it operates in, and several other factors, the answer can be anything from an hour to a couple of weeks. The important thing is to take the task step-by-step, eliminating the most critical issues first and making real progress from the beginning. Security is never complete and no system is unbreakable, however, we can build significant defenses so that the effort to break in far outweighs any potential payoff to the attacker.