+1.800.357.1019

+1.800.357.1019

Feel free to call us toll free at +1.800.357.1019.

If you are in the area you can reach us at +1.360.359.4400.

Standard support
6:30am - 4:00pm PST, Monday - Friday, Free

Premium support
If you own Townsend Security 24x7 support and
have a production down issue outside normal
business hours, please call +1.800.349.0711
and the on-call person will be notified.

International customers, please dial +1.757.278.1926.

Townsend Security Data Privacy Blog

Patrick Townsend

Recent Posts

Microsoft SQL Server Encryption Key Management

Posted by Patrick Townsend on Mar 20, 2017 2:01:48 PM

The hardest part of an encryption strategy is the proper management of encryption keys. Failing to protect encryption keys puts protected data at risk, and fails to meet security best practices and compliance regulations. For Microsoft SQL Server customers who have already implemented Transparent Data Encryption (TDE) or Cell Level Encryption (CLE) the biggest cause of an audit failure is the lack of good encryption key management.

Encryption-Key-Management-SQL-Server This is the fourth in a series on the topic of Microsoft SQL Server encryption. Let’s look at some of the characteristics of good encryption key management for SQL Server.

Extensible Key Management (EKM) Providers
As we’ve discussed previously it is the responsibility of key management vendors to provide the Extensible Key Management (EKM) Provider software that is installed and registered to the SQL Server database enabling either TDE or CLE encryption. The software from the key management vendor is installed on the SQL Server instance and provides both encryption and key management services. The SQL Server database administrator does not need to be involved in the actual retrieval of an encryption key - that is the job of the EKM Provider software.

EKM Provider software must handle the encryption and decryption of the database key for Transparent Data Encryption, and must handle the retrieval of a symmetric key for Cell Level Encryption. Key retrieval should be performed in a manner that protects the encryption key from loss on the network, protects the key while in memory, and should properly log the key retrieval event in a system log repository. Encryption key retrieval is normally protected through the use of a secure TLS network connection between the EKM Provider software on SQL Server and the key manager hardware or virtual machine. There are many other critical aspects of EKM Provider key management implementations, and these will be discussed in a future series.

Enterprise Key Management Solutions
The proper generation, storage, protection and management of encryption keys is the core purpose of professional encryption key management solutions. As security devices an encryption key manager is responsible for creating strong encryption keys that meet industry standards, and protecting those keys from loss during the lifecycle of the keys. Encryption key managers may be hardware security modules (HSMs), virtual servers (VMware, Hyper-V, etc.), or multi-tenant or dedicated cloud instances. In addition to implementing industry standards for encryption key management, key servers will provide a variety of authentication, systems management, and audit functions to meet security best practices and compliance regulations. Microsoft SQL Server customers who want to achieve compliance with common regulations should look to deploy a professional, certified and validated key management solution.

Key Management Industry Standards
Encryption key management systems are cryptographic modules that perform a variety of functions. As a cryptographic module they fall under the standards of the National Institute of Standards and Technology (NIST) and key managers should provably meet NIST standards. The relevant NIST standard for encryption key management is the Federal Information Processing Standard 140-2 (FIPS 140-2), “Security Requirements for Cryptographic Modules”. Key management solutions which implement FIPS 140-2 standards will insure the generation of strong encryption keys, the protection of those keys from corruption or substitution, and the implementation of encryption that provably meets NIST cryptographic standards.

In addition to provide standards for encryption key management NIST also provides a method for vendors to validate that their solutions meet the standard. Encryption key management solutions are tested by chartered security testing laboratories and solutions are then approved directly by NIST. NIST publishes the solutions that have passed FIPS 140-2 testing and Microsoft SQL Server customers should look for FIPS 140-2 validation of any key management solution used to protect the database.

Migrating Locally Stored Keys to Key Management
Many Microsoft SQL Server users start their encryption projects by using the option to locally store the database encryption key on the local SQL Server instance. While this is not a security best practice, it is a common way to start an encryption project.

Fortunately, it is easy to migrate a locally stored encryption key to a proper key management solution. The migration involves the protection of the SQL Server database key to key management protection and does not require the decryption of the database. The database key which is currently protected by local keys and certificates is placed under the protection of the key manager. The EKM Provider software of your vendor then becomes responsible for unlocking the database key (TDE) or retrieving the symmetric key for Cell Level Encryption (CLE).

OASIS Key Management Interoperability Protocol (KMIP)
Many SQL Server customers ask about the KMIP standard for integrating with key managers. While KMIP is important for many reasons, it does not apply to the Microsoft EKM Provider interface. The EKM Provider interface leaves it to the key management vendor to perform the needed cryptographic functions on the key server. These functions do not map to KMIP operations and attributes. While it is advisable to deploy key management solutions that meet KMIP standards, it is not required for SQL Server encryption.

To this point we have defined the SQL Server encryption architecture, options for implementing SQL Server encryption (TDE and CLE), and basic requirements for encryption key management. In the next part of this series we will look at EKM Provider implementation topics as well as business continuity topics.

Patrick
Encryption and Key Management for Microsoft SQL Server

Topics: SQL Server, SQL Server encryption

Microsoft SQL Server Automatic Encryption - Cell Level Encryption

Posted by Patrick Townsend on Feb 21, 2017 9:11:00 AM

In this third part of the series on Microsoft SQL Server encryption we look at Cell Level Encryption, or CLE, which is Microsoft terminology for Column Level Encryption. With CLE the manner and timing of SQL Server’s call to the EKM Provider software is quite different than for Transparent Data Encryption. It is important to understand these differences in order to know when to use CLE or TDE. Let’s look at some aspects of the CLE implementation:

Encrypted Columns
Encryption-Key-Management-SQL-Server Cell Level Encryption is implemented at the column level in a SQL Server table. Only the column you specify for encryption is protected with strong encryption. You can specify more than one column for CLE in your tables, but care should be taken to avoid performance impacts of multiple column encryption (see below).

With Cell Level Encryption you may be able to minimize some of the encryption performance impacts on your SQL Server database. Because the EKM Provider is only called when the column must be encrypted or decrypted, you can reduce the encryption overhead with careful implementation of your database application code. If a SQL query does not reference an encrypted column, the EKM Provider will not be invoked to perform decryption. As an example, if you place the column Credit_Card under CLE encryption control, this query will not invoke the EKM Provider for decryption because the credit card number is not returned in the query result:

SELECT Customer_Number, Customer_Name, Customer_Address FROM Orders ORDERBY Customer_Name;

You can see that judicious use of SQL queries may reduce the need to encrypt and decrypt column data.

SQL Application Changes
Unlike Transparent Data Encryption you must make a change to the SQL statement in order to implement Cell Level Encryption. The SQL Server functions “encryptbykey” and “decryptbykey” are used on SQL statements. Here is an example of a SQL query that encrypts a CLE-encrypted column:

select encryptbykey(key_guid('my_key'), 'Hello World');

Implementing CLE encryption in your SQL Server database requires modifications to your applications, but may be well worth the additional work.

Encryption and Key Retrieval
The EKM Provider software is called for each column value to perform encryption and decryption. This means a larger number of calls to the EKM Provider compared to Transparent Data Encryption. Because the number of calls to the EKM Provider may be quite large it is important that the encryption and key management functions of the EKM Provider are highly optimized for performance (see the next section).

The EKM Provider software from your key management vendor is responsible for performing encryption of the data. From a compliance point of view it is important to understand the encryption algorithm used to protect data. Be sure that the EKM Provider software uses a standard like the Advanced Encryption Standard (AES) or other industry recognized standard for encryption. It is common to use 128-bit or 256-bit AES for protecting data at rest. Avoid EKM Providers which implement non-standard encryption algorithms.

Encryption Key Caching
When deploying CLE it is important that the EKM Provider software optimize both encryption and key management. The number of calls to the EKM Provider software can be quite high. Good EKM Providers will securely cache the symmetric key in the SQL Server context rather than retrieve a key on each call. The retrieval of an encryption key from a key server takes precious time and multiple calls to retrieve a key can have severe performance impacts. Secure key caching is important for CLE performance. The use of the Microsoft Windows Data Protection Application Program Interface (DPAPI) is commonly used to protect cached keys.

Performance Considerations
When properly implemented Cell Level Encryption can reduce the performance impact of encryption on your SQL Server database. For very large tables with a small number of columns under encryption control, the performance savings can be substantial. This is especially true if the column is used less frequently in your applications.

CLE Vendor Note
Note that each vendor of EKM Provider software implements encryption and key management differently. Some EKM Providers only implement Transparent Data Encryption (TDE). If you suspect you will need Cell Level Encryption be sure that your key management support includes this capability.

In the next part of this series we will look at encryption key management in SQL Server.

Patrick

Encryption and Key Management for Microsoft SQL Server

Topics: SQL Server, Cell Level Encryption, SQL Server encryption

Microsoft SQL Server Automatic Encryption - Transparent Data Encryption

Posted by Patrick Townsend on Feb 14, 2017 8:33:00 AM

In this second part of the series on Microsoft SQL Server encryption I want to focus on Transparent Database Encryption, or TDE. Most Microsoft customers who implement encryption in SQL Server use Transparent Data Encryption as it is the easiest to deploy. No code changes are required and enabling encryption requires just a few commands from the SQL Server console. Let’s look at some of the characteristics of TDE implementation.

Database Encryption
Download the Webinar - Just Click! TDE involves the encryption of the entire database space in SQL Server. There is no need or ability to select which tables or views are encrypted, all tables and views in a database are encrypted at rest (on disk). When data is read from disk (or any non-volatile storage) SQL Server decrypts the entire block making the data visible to the database engine. When data is inserted or updated the SQL Server database encrypts the entire block written to disk.

With TDE all of the data in your database is encrypted. This means that non-sensitive data is encrypted as well as sensitive data. There are advantages and disadvantages to this approach - you expend computing resources to encrypt data that may not be sensitive, but you also avoid mistakes in identifying sensitive data. By encrypting everything at rest you are also protected from expansion of regulatory rules about sensitive data protection.

sql-decrypts-db.png

Protection of the Symmetric Key
When you enable Transparent Data Encryption on your SQL Server database the database generates a symmetric encryption key and protects it using the EKM Provider software from your key management vendor. The EKM Provider software sends the symmetric key to the key server where it is encrypted with an asymmetric key. The encrypted database key is then stored locally on disk in the SQL Server context.

When you start a SQL Server instance the SQL Server database calls the EKM Provider software to decrypt the database symmetric key so that it can be used for encryption and decryption operations. The decrypted database key is stored in protected memory space and used by the database. The encrypted version of the database key remains on disk. In the event the system terminates abnormally, the only version of the database key is the encrypted version on disk.

Starting the SQL Server Instance
During normal operation of SQL Server there is no invocation of the EKM Provider software and therefore no communication with an external key manager. Every normal restart of the SQL Server database instance will cause the EKM Provider software to be called to unlock the database key on the key server.

It should be noted that it is the responsibility of the EKM Provider software to handle network or key server failure conditions. SQL Server itself has no visibility on the connection to an encryption key management solution. If the EKM Provider software is unable to retrieve an encryption key, the SQL Server start request will fail. We will discuss business continuity issues in more detail later in this series.

Protecting Database Logs
SQL Server logs may contain sensitive data and therefore must also be encrypted. Transparent Database Encryption addresses this by fully encrypting database logs along with the database itself. It is important to remember that encryption of the logs will only start after TDE is activated AND after you stop and restart the database log. If you neglect to restart logging sensitive data may be exposed in the SQL Server log files.

Table and Index Scanning
Certain SQL operations on indexes require that the SQL Server database have visibility on the entire index of a column. An example of a SELECT statement would be something like this:

SELECT Customer_Name, Customer_Address FROM Orders WHERE Credit_Card=’4111111111111111’;

To satisfy this SQL query the database must inspect every row in the table Orders. With TDE this means that the column Credit_Card must be decrypted in every row. Similar operations with the ORDERBY clause can cause table or index scans.

Performance Considerations
Transparent Data Encryption is very optimized for encryption and decryption tasks and will perform well for the majority of database implementations. Microsoft estimates the performance impact of TDE of 2% to 4% and we find this accurate for most of our customers. However, Microsoft SQL Server customers with very large SQL Server databases should use caution when implementing TDE. Be sure that you fully understand the impact of TDE on your application use of large tables. It is always recommended that you perform a proof-of-concept project on very large databases to fully assess the performance impact of encryption.

In the next part of this series we will look at the other option for SQL Server encryption - Cell Level Encryption, also called column level encryption.

Patrick

Encryption and key management for SQL Server

Topics: SQL Server, Transparent Data Encryption (TDE)

Microsoft SQL Server Encryption - An Introduction

Posted by Patrick Townsend on Feb 6, 2017 2:33:16 PM

In 2008 the Payment Card Industry Data Security Standard (PCI-DSS) was gaining serious traction and Microsoft released SQL Server 2008 with built-in support for encryption. This was no coincidence. In addition to the PCI standard which mandated encryption of credit card numbers, numerous states in the US had also adopted data breach notification laws with strong recommendations for encryption. The compliance environment was changing dramatically and the SQL Server group at Microsoft provided a path to meet those new compliance regulations. This was a prescient and crucially important enhancement for Microsoft customers - the security threats have increased over time and compliance regulations have become more stringent.

Download the Webinar - Just Click! In this multi-part series I want to talk about how Microsoft implemented encryption in SQL Server, how you can leverage this capability to achieve better security and compliance, and the critical issues involved in getting encryption right with SQL Server. I hope you will find this series helpful as you negotiate your SQL Server encryption projects.

Architecture
Many Microsoft applications and services implement a “Provider” interface. This is the term that Microsoft uses to describe a standardardized, pluggable architecture for third party software companies to integrate and extend the capabilities of Microsoft solutions. With Provider architectures Microsoft enables a method for third parties to register their software to the Microsoft application, and the Microsoft application will then call that software as needed. The third party software must obey rules about the data interface and behavior of their applications. If done correctly the Provider interface provides powerful extensions to Microsoft applications.

Starting with SQL Server 2008 the database implements a Provider interface for encryption and key management. This is named the “Extensible Key Management” Provider interface, or the “EKM Provider”. EKM Provider software performs encryption and key management tasks as an extension to the SQL Server database. The EKM Provider architecture opened the door for third party key management vendors to extend encryption to include proper encryption key management.

From a high level point of view the EKM architecture looks like this:

SQL-EKM.png

Every version of SQL Server since 2008 has fully implemented the EKM Provider architecture. This has provided a stable and predictable interface for Microsoft customers and encryption key management vendors.

EKM Architecture - Column and Database Encryption
The EKM Provider architecture supports two different methods of database encryption:

  • Cell Level Encryption
  • Transparent Database Encryption

Cell level encryption is also known as column level encryption. As its name implies it encrypts data in a column in a table. When a new row is inserted into a table, or when a column in a row is updated, the SQL Server database calls the EKM Provider software to perform encryption. When a column is retrieved from the database through a SQL SELECT or other statement the EKM Provider software is called to perform decryption. The EKM Provider software is responsible for both encryption and key management activity. Implementing cell level encryption requires minor changes to the SQL column definition.

Transparent Database Encryption, or TDE, provides encryption for the entire database and associated log files. All tables and views in the database are fully encrypted. Data is encrypted and decrypted as information is inserted, updated, and retrieved by users and applications. As its name implies, transparent data encryption requires no changes to applications, SQL definitions, or queries. The database works seamlessly after encryption is enabled.

Transparent Data Encryption is the easiest of the two encryption methods to implement. In a following part of these series I will discuss when it makes sense to use TDE and when Cell Level Encryption is a better choice.

Activating the EKM Provider
After installing the EKM Provider software from a third party, the SQL Server database administrator uses the SQL Server management console to activate the EKM Provider and place the database or columns under encryption control. The activation of the EKM Provider software causes the database to be immediately encrypted and all further data operations on the database will invoke the EKM Provider software.

Microsoft EKM Provider for Locally Stored Encryption Keys
Recognizing that some SQL Server customers wanted to encrypt data but did not have the resources or time to implement a key management solution, Microsoft provided a built-in EKM Provider that performs encryption but which stores encryption keys locally in the SQL Server context. Understanding that this was not a security best practice, Microsoft recommends that customers use a proper encryption key management solution that separates encryption keys from the SQL Server database. That was good advice - locally stored encryption keys can be recovered by cyber criminals and the use of external key management systems provides better security and compliance.

EKM Provider Software
EKM Provider software is usually provided by your encryption key management vendor. This means that the features and functions of the EKM Provider software can vary a great deal from one vendor to another. Be sure that you fully understand the architecture and capabilities of the EKM Provider before you deploy SQL Server encryption.

SQL Server Versions That Support EKM
EKM Provider support is available in all Enterprise editions of SQL Server including Data Warehouse and Business Intelligence editions. EKM provider support is not available in Standard, Web, or Express editions of SQL Server.

In the following series I will go into more detail on the EKM Provider interface, transparent data encryption, cell level encryption, business continuity, compliance, and other topics.

Patrick

Encryption and key management for SQL Server

Topics: Encryption, SQL Server

Fixing the TDE Key Management Problem in Microsoft SQL Server

Posted by Patrick Townsend on Jan 10, 2017 7:31:56 AM

Many Microsoft SQL Server users have taken the first step to protect sensitive data such as Personally Identifiable Information (PII), Protected Health Information (PHI), Primary Account numbers (PAN) and Non-Public Information (NPI) by encrypting their databases with Transparent Data Encryption (TDE). It is extremely easy to implement TDE encryption as it does not require program changes.

Download the Webinar - Just Click! A common cause of audit failures might not be so obvious and that is the failure to properly protect the SQL Server key encryption key once you activate encryption in SQL Server. With Transparent Data Encryption you have the choice of storing the service master key within the SQL Server context itself, or protecting the master key with a key management system using the SQL Server Extensible Key Management (EKM) interface. Why is it important to do this?

It turns out that it is easy for cyber criminals to recover the SQL Server master key when it is stored within SQL Server itself. (Examples: https://blog.netspi.com/decrypting-mssql-credential-passwords/ and http://simonmcauliffe.com/technology/tde/#hardware)

Simon McAuliffe provides the clearest explanation I’ve seen on the insecurity of locally stored TDE keys in SQL Server. I don’t agree with him on the question of using a key manager to improve security. Given that there is no perfect security, I believe that you can get significant security advantages through a properly implemented key management interface.

If your TDE keys are stored locally, don’t panic. It turns out to be very easy to migrate to a key management solution. Assuming you’ve installed our SQL Server EKM Provider called Key Connection on your SQL Server instance, here are the steps to migrate your Service Master Key to key management protection using our Alliance Key Manager solution. You don’t even need to bring down SQL server to do this (from the Alliance Key Manager Key Connection manual):

Protecting an existing TDE key with Alliance Key Manager

First create a new asymmetric key pair within the AKM Administrative Console using the “Create EKM Key” and the “Enable Key for EKM” commands.

Then return to SQL Server and call the following command to create the asymmetric key alias for the new KEK that you created on the AKM server:

use master;

create asymmetric key my_new_kek from provider KeyConnection with provider_key_name = ’NEW_TDE_KEK’, creation_disposition = open_existing;

In this example, NEW_TDE_KEK is the name of the new key on AKM, and my_new_kek is the key alias.

Then use the ALTER DATABASE statement to re-encrypt the DEK with the new KEK alias assigned in the previous statement:

ALTER DATABASE ENCRYPTION KEY

ENCRYPTION BY SERVER

   {  ASYMMETRIC KEY my_new_kek}

Note that you do not have to take the database offline to perform this action.

Of course, there are other steps that you should take to secure your environment, but I wanted to demonstrate how easy it is to make the change.

The SQL Server DBA and the network administrator will have lots of other considerations in relation to SQL Server encryption. This includes support for clustering and high availability, automatic failover to secondary key servers, adequate support for separation of duties (SOD) and compliance, and the security of the credentials needed to validate SQL Server to the key manager. All of these concerns need to be addressed in a key management deployment.

For SQL Server users who deploy within a VMware or cloud infrastructure (AWS, Azure), Alliance Key Manager can run natively in your environment, too. It does not require a hardware security module (HSM) to achieve good key management with SQL Server. You have lots of choices in how you deploy your key management solution.

It turns out not to be difficult at all to address your SQL Server encryption key insecurities!

Patrick

Encryption and key management for SQL Server

Topics: SQL Server, Transparent Data Encryption (TDE)

OpenSSH on the IBM i and Your Security

Posted by Patrick Townsend on Jan 3, 2017 7:45:48 AM

Lately I’ve seen some criticism of the OpenSSH implementation on the IBM i platform which seems to imply that using a third-party implementation of the Secure Shell (SSH) file transfer application is better without the IBM no-charge licensed OpenSSH implementation. I disagree with that opinion and think there are good security and implementation reasons to stick with the IBM OpenSSH implementation.

Here are some reasons why I like OpenSSH:

OpenSSH is supported by an global open source community

Tatu Ylönen founded SSH Communications in 1995 and produced the first versions of an open source SSH implementation. Since 1999 the OpenSSH application has been maintained by the OpenBSD Project which is funded by the OpenBSD foundation and managed by Theo de Raadt. OpenSSH is available on a wide variety of operating systems including the IBM i where it is deployed as a no-charge licensed product and maintained by IBM.  OpenSSH continues to be actively developed and new encryption algorithms have been added recently.

OpenSSH is a widely used by large and small organizations

By some estimates the OpenSSH implementation of the SSH protocol and applications commands a 97 percent market share for SSH implementations. This means that OpenSSH is in wide use by large and small organizations to securely manage their eCommerce needs. This also means that OpenSSH receives a lot of scrutiny by compliance and security experts. Widely deployed solutions tend to get more scrutiny from security experts, and this is true for OpenSSH.

OpenSSH is secure

No application is immune to security challenges. However, OpenBSD and the OpenSSH application in particular have a stellar record for security. With security products, deep expertise and commitment matter. OpenSSH started with security as a leading goal by its developers and it shows. Over the last few years there have been fewer than a dozen security issues, and most were unlikely to be exploited and all were patched rapidly through updates by IBM. The OpenBSD set of applications that include OpenSSH have a great record on security. If you think the IBM i platform has a good security record, take a look at OpenSSH.

IBM provides technical support for OpenSSH

We have all developed a deep appreciation for IBM’s commitment to security over the years. It is one of great values of the IBM i platform. As new vulnerabilities are discovered you need to have a reliable and timely source of patches and enhancements and IBM has stood behind this critical application. Security notifications are managed by IBM so that you know when you need to do an update. By making OpenSSH a no-charge licensed program IBM i customers get patches through the normal PTF update process. Do you know any third-party IBM i vendor with an equal commitment to notification, maintenance and patching? IBM has earned our trust through this process.

OpenSSH is PCI compliant

PCI Qualified Security Assessors (QSAs) like Coalfire, TrustWave and others recognize that a properly patched implementation of OpenSSH meets PCI Data Security Standards (PCI-DSS) compliance, and IBM also tracks OpenSSH for PCI compliance. This again reflects IBM’s and OpenBSD’s commitment to security. If you are using a third-party IBM i solution for SSH how well is it tracked by the PCI audit community?

SSH is a complex protocol

Bruce Schneier said “Complexity is the enemy of security.” SSH is a complex protocol and this means that extra care needs to be taken in its development, deployment and maintenance. No third-party SSH solution rises to the level of care taken by the OpenSSH community and by IBM. Almost every business depends on secure file transfer for daily business operations. Deploying the most secure SSH solution is a critical security step.

OpenSSH does not use OpenSSL or Java JSSE

We’ve read a lot over the last few months about security issues in OpenSSL and Java. Many IBM i customers are confused about the relationship between OpenSSH and OpenSSL. In fact, OpenSSH does not use the OpenSSL library for communications. This means that OpenSSH was not subject to the HeartBleed and other OpenSSL vulnerabilities. We are all also now painfully aware of the security issues in Java. Most browsers no longer allow Java plugins for this reason. Third-party SSH products may or may not use OpenSSL or Java for communications. If you are running a third-party IBM i SSH solution, do you know if it uses OpenSSL or Java?

Third-party SSH solutions provide no significant advantage over OpenSSH

OpenSSH is a secure, reliable, and resilient implementation of SSH for secure data transfer that is backed by IBM and a worldwide community of users and developers. Our Alliance FTP Manager solution fully integrates with the IBM i OpenSSH application for secure, automated and managed file transfer. Our solution automates the OpenSSH transfer of hundreds of thousands of file transfers every day without compromising security.

My opinion? You probably don’t need more IT risk in your life. Stick with OpenSSH for your security needs. You will be in good company.

Patrick

Webinar: Secure Managed File Transfer on IBM i

Topics: IBM i, Secure Managed File Transfer

New York Department of Financial Services (NYDFS) and Encryption - 8 Things to Do Now

Posted by Patrick Townsend on Dec 12, 2016 10:27:38 AM

The New York Department of Financial Services (NYDFS) surprised the financial services industry by fast tracking new cybersecurity regulations in September of 2016. Due to go into effect in January of 2017 with a one-year transition period, it takes a very prescriptive approach to cybersecurity which includes a mandate to encrypt data at rest. The financial sector is broadly defined as banks, insurance companies, consumer lenders, money transmitters, and others. The law is formally known as 23 NYCRR 500 and you can get it here.

encryption-key-management-simplified There isn’t much wiggle room on the requirement for encrypting sensitive data. You can use compensating controls if you can show that encryption is “infeasible”. But I am not sure how you would show that. All modern database systems used by financial applications support encryption. It would be hard to imagine a financial database where encryption would not be feasible. Don’t plan on that being an excuse to delay encrypting data at rest!

The time frame is short for implementing the encryption mandate. One year seems like a long time, but it is extremely aggressive given the development backlog I see in most banks.

Here are some things you should start doing right now:

1) Inventory All of Your Financial Systems

This seems like a no-brainer, but you might be surprised how many organizations have no formal inventory of their IT systems that contain financial data. This is a top-of-the list item on any cybersecurity list of recommendations, so making or updating this list will have a lot of benefits.

2) Document Storage of All Sensitive Information (Non-Public Information, or NPI)

For each system in your inventory (see above) document every database and storage mechanism that stores NPI. For database systems identify all tables and columns that contain NPI. You will need this documentation to meet the NYDFS requirements, and it is a roadmap to meeting the encryption requirements.

3) Prioritize Your Encryption Projects

You won’t be able to do everything at once. Following all modern cybersecurity recommendations, prioritize the systems and applications that should be addressed using a risk model. Here are a few factors that can help you prioritize:

  • Sensitivity of data
  • Amount of data at risk
  • Exposure risk of the systems and data
  • Compliance risk
  • Operational impact of loss

It is OK to be practical about how you prioritize the systems, but avoid assigning a high priority to a system because it might be easiest. It is better to tackle the biggest risks first.

4) Establish Encryption Standards

Be careful which encryption algorithms you use to protect sensitive data. In the event of a loss you won’t want to be using home-grown or non-standard encryption. Protect data at rest with NIST compliant, 256-bit AES encryption. This will give you the most defensible encryption strategy and is readily available in all major operating systems such as Windows, Linux, and IBM enterprise systems.

5) Establish Key Management Standards

Protecting encryption keys is the most important part of your encryption strategy and the one area where many organizations fail. Encryption keys should be stored away from the encrypted financial data in a security device specifically designed for this task. There are a number of commercial key management systems to choose from. Be sure your system is FIPS 140-2 compliant and implements the industry standard Key Management Interoperability Protocol (KMIP).

Hint: Don’t fall into the project-killing trap of trying to find a key management system that can meet every key management need you have in the organization. The industry just isn’t there yet. Pick a small number of key management vendors with best-of-breed solutions.

With encryption standards well defined and an encryption key management strategy in hand you are ready to get started with your encryption projects.

6) Analyze Performance and Operational Impacts

Encryption will naturally involve some performance and operational impacts. Encryption is a CPU intensive task, so plan on doing some performance analysis of your application in real-world scenarios. If you don’t have test environments that support this analysis, get started now to create them. They will be invaluable as you move forward. Modern encryption is highly optimized, and you can implement encryption without degrading the user experience. Just be prepared to do this analysis before you go live.

There are also operational impacts when you start encrypting data. Your backups may take a bit more storage and take longer to execute. So be sure to analyze this as a part of your proof-of-concept. Encrypted data does not compress as well as unencrypted data and this is the main cause of operational slow-downs. For most organizations this will not be a major impact, but be sure to test this before you deploy encryption.

8) Get Started

Oddly (to me at least) many organizations just fail to start their encryption projects even when they have done the initial planning. A lack of commitment by senior management, lack of IT resources, competing business objectives, and other barriers can delay a project. Don’t let your organization fall into this trap. Do your first project, get it into production, and analyze the project to determine how to do it better as you move forward.

Fortunately we have a lot of resources available to us today that were not available 10 years ago. Good encryption solutions are available and affordable for traditional on-premise environments, for VMware infrastructure, and for cloud applications.

You can meet the NYDFS requirements and timelines if you start now. But don’t put this one off.

Patrick

 

Resources:

New York Department of Financial Services:

http://www.dfs.ny.gov/legal/regulations/proposed/propdfs.htm

 

Harvard Law School analysis of NYDFS:

https://corpgov.law.harvard.edu/2016/09/24/nydfs-proposed-cybersecurity-regulation-for-financial-services-companies/

Encryption Key Management Simplified eBook

 

Topics: Compliance, Encryption

IBM i Security Architecture & Active Monitoring

Posted by Patrick Townsend on Dec 6, 2016 7:30:42 AM

Excerpt from the eBook "IBM i Security: Event Logging & Active Monitoring - A Step By Step Guide."


IBM i Security: Event Logging & Active Monitoring Active monitoring (sometimes referred to as Continuous Monitoring) is a critical security control for all organizations. And it is one of the most effective security controls you can deploy. The large majority of security breaches occur on systems that have been compromised days, weeks, or even months before sensitive data is lost. A recent Verizon Data Breach Investigations Report1 indicates that a full 84 percent of all breaches were detected in system logs. This is why the Center for Internet Security includes active monitoring as a Critical Security Control (Control number 6).

There are several elements of a truly effective Active Monitoring strategy:

Central Collection and Repository of All Events
Attackers almost never start with your core IBM i server directly. They attack a web application or infect a user PC and work their way into the IBM i server. A defensible active monitoring strategy has to collect events from across the entire organization. By the time they show up on your IBM i server they have probably compromised a number of intermediate systems and an opportunity to prevent the breach has been missed. Collect all events across your entire IT infrastructure to gain the best early detection opportunities.

Real Time Event Collection
Data breaches are happening much faster than in the past. In some cases the loss of data happens just minutes after the initial breach. This means that you must collect security events in real time. Good active monitoring solutions are able to digest threat information
in real time and give you the chance to deter them. Avoid batch event collection – you can collect IBM i security audit journal information in real time and you should.

Event Correlation
Event correlation is key to an effect active monitoring solution. This is typically accomplished through the use of special software implemented in Security Information and Event Management (SIEM) solutions. Highly automated SIEM solutions have the ability to correlate events across a large number of systems and automatically identify potential problems. They do exactly what we want computer systems to do – handle large amounts of data and apply intelligent interpretation of the data.

Anomaly Detection
Anomaly detection is another aspect of active monitoring. That unusual system login at 3:00am on a Sunday morning would probably escape the attention of our human IT team members, but good active monitoring solutions can see that anomalous event and report on it.

Alerting and Resolution Management
When a problem is discovered we need to know about it as soon as possible. A good active monitoring solution will inform us through a variety of alerting channels. Emails, text, dashboards and other mechanisms can be deployed to bring attention to the problem - and we need to be able to track the resolution of the event! We are all processing too much information and it is easy to forget or misplace a problem. 

Forensic Tools
Forensic tools are critical to an active monitoring solution as they enable the rapid analysis of an attacker’s footprints in our system. The key tool is an effective and easy-to-use query application. Log data can include millions of events and be impossible to inspect without a good query tool. The ability to save queries and use them at a later time should also be a core feature of your forensic toolset.

IBM i


Topics: System Logging, IBM i

Dangers of Encryption on the IBM i (AS/400, iSeries): Avoid These Pitfalls

Posted by Patrick Townsend on Nov 14, 2016 8:35:13 AM

IBM has done a good job of implementing security from the ground up on the IBM i platform. But that doesn’t mean that it is immune from data breaches. All of the PCs and servers on your network with the IBM i server are potential attack points for a data breach. And make no mistake, cyber criminals know that the IBM i server is a rich target. Implementing encryption in IBM i DB2 is an essential part of a defense in depth strategy. But there are lots of pitfalls to avoid. Let’s take a look at some of them (I am shamelessly plugging our Alliance AES/400 solution, too):

Key Management for IBM i - Audit Failures Locally Stored Encryption Keys and Key Management

One of the surest ways to defeat your encryption strategy is to store encryption keys on the same system that stores sensitive data. The IBM i server is no exception. Compliance regulations and security best practices require that you store encryption keys away from the IBM i server in an encryption key vault designed for this purpose. Why is this a security best practice? Cybercriminals are often able to achieve privilege escalation on a compromised IBM i server and then get access to locally stored keys. Storing encryption keys off of the IBM i server makes the compromise of the sensitive data much harder.

How Townsend Security Can Help

The Townsend Security Alliance AES/400 product integrates seamlessly with Townsend Security's Alliance Key Manager solution for protection of encryption keys and key management best practices. Alliance Key Manager stores encryption keys in a hardware security module or VMware instance that is attached to the IBM i server by a secure, authenticated TLS connection. As a FIPS 140-2 compliant key management solution, Alliance AES/400 with Alliance Key Manager solves the key management problem!

High Availability and Key Mirroring

Encryption key management is a part of your critical infrastructure. The loss of an encryption key means the loss of your data! Your encryption key management solution should implement real-time key mirroring and real-time security policy mirroring. In the event the key manager is unavailable due to a hardware or network failure, the failover to a secondary key server should be automatic without business interruption. A key management solution based on the IBM i master key facility cannot achieve real-time mirroring and protection from these failures.

How Townsend Security Can Help

Alliance Key Manager implements real-time key mirroring to one or more backup key servers. The mirroring implementation is active-active meaning that any changes you make to keys or access policies on the secondary server will be mirrored to the production server when it comes back online. This perfectly matches your IBM i high availability failover strategy if you use MIMIX, iTera, Vision, or IBM DataMirror.

Encryption and Insider Threats

Insider threats include both intentional and unintentional access to and loss of sensitive information. Unintentional losses of data represents the largest insider threat. Accidentally copying data to a PC or development environment can lead to a reportable data breach event. This is especially true when access controls to sensitive data are only controlled by native IBM i object level security. You should certainly use native IBM i security mechanisms, but access to decrypted sensitive data should also be controlled using a “whitelist” approach. This will help minimize the intentional and unintentional access by security administrators. Note that it is not only the security profile QSECOFR that has all access to sensitive data: all users with All Object (*ALLOBJ) authority or who adopt this level of authority through a group profile or supplemental group are at risk for intentional or unintentional loss of sensitive data.

How Townsend Security Can Help

Alliance AES/400 implements a whitelist approach for controlled access to decrypted sensitive data. All configuration changes to security policies are logged to the IBM security audit journal QAUDJRN. You can achieve effective Separation of Duties between managers of the encryption keys and security administrators on the IBM i platform.

Poorly Performing Encryption Libraries

Encryption can also present an operational risk to IBM i customers. In order to meet service level expectations of end users encryption and decryption operations must be efficient. Unfortunately for IBM i customers the native AES encryption software libraries provided in the operating system may not provide an adequate level of performance. Even with the new IBM i POWER8 servers with on-chip encryption, the performance of AES encryption and decryption tasks is poor. It is important to assess the size of your protected databases and the nature of batch operations that require access to unencrypted data in order to avoid negative impacts to both interactive and batch applications.

How Townsend Security Can Help

Alliance AES/400 uses the Townsend Security NIST-validated AES encryption library for encryption and decryption tasks. This optimized AES encryption library is more than 100x faster that native IBM i encryption libraries on POWER7 processors, and more than 50x faster on POWER8 processors.

Encrypted Indexes

Many IBM i customers are surprised to learn that their RPG applications will not work correctly with DB2 FieldProc for encrypted indexes (key fields). FieldProc is IBM’s automatic column level encryption feature implemented at the DB2 database level. FieldProc is attractive to IBM i customers because it does not require modifications to applications. While native SQL applications can easily handle encrypted indexes, RPG applications do not use the native SQL Query Engine (SQE) and will not work properly with encrypted indexes. Most IBM i customers exclusively use RPG or have a mix of RPG and SQL applications. The issue with RPG and encrypted indexes represents a major roadblock to encryption. Be sure that your encryption strategy can support encrypted indexes, or be prepared to modernize RPG applications to use native the SQL Query Engine.

How Townsend Security Can Help

Townsend Security tackled the problem of encrypted indexes and offers a solution to the RPG challenge through its Open Access for RPG SQL library. Changing one line of code in your RPG application can automatically use the native SQL Query Engine for database access. This eliminates the challenges of encrypted indexes with FieldProc encryption.

Data Masking

Compliance regulations such as PCI Data Security Standard (PCI-DSS) and security best practices require that we only allow authorized users access to fully decrypted sensitive data. But other users must have access to our database applications. This means that intelligent data masking should be built into your IBM i applications. As noted above, data masking should be based on a whitelist approach and not purely based on object or database level authority. You should have the ability to define masking rules (mask all but the last 4 characters, etc.) and you should be able to define a default masking rule that applies to all unauthorized users. While Row and Column Access Controls (RCAC) can provide some data masking capability, you must manage individual user level authorities to implement this control.

How Townsend Security Can Help

Alliance AES/400 fully implements data masking using a whitelist approach and provides protections from users with All Object (*ALLOBJ) or Security Administrator (*SECADM) privileges. Data is masked in the internal decryption routines and fully exposed data is never visible in the application program.

System Audit Logs

No security policy or solution can be effective on a stand-alone basis and this includes encryption and key management. A good encryption and key management strategy involves monitoring all access to sensitive data, monitoring changes to encryption and key management configurations, monitoring all use of encryption keys, and storing audit logs for future forensic reference. The use of Security Information and Event Management (SIEM) solutions is highly recommended as a part of your monitoring and alerting strategy. Be sure that all access to encryption and encryption keys is fully audited and logged.

How Townsend Security Can Help

Alliance AES/400 and Alliance Key Manager implement system logging and audit for all aspects of administration, configuration and use. Alliance Key Manager implements full logging of all aspects of key management and the server it runs on, and transmits logs to a SIEM solution in real time. Alliance AES/400 fully logs all administrative operations and decryption tasks to the IBM i security audit journal QAUDJRN. The optional Alliance LogAgent solution transmits these logs as well as all IBM i security events to a SIEM solution or log collection server.

Encryption and key management don’t have to feel dangerous or scary! I hope the above points about encryption and key management for the IBM i help you develop a roadmap for successful (and safe!) encryption.

Patrick

Key Management for IBM i - Sources of Audit Failures

Topics: Encryption, Key Management, IBM i

IBM i Security Notification - OpenSSL and OpenVPN

Posted by Patrick Townsend on Nov 8, 2016 8:46:49 AM

In case you missed it, IBM just released a security notification for the IBM i platform - all versions of the operating system from 6.1 through 7.3. This one is important and you should take a look at it right away. The vulnerabilities are related to OpenSSL and to OpenVPN which uses OpenSSL. The vulnerability is called the SWEET32 Birthday attack. OpenSSL is used in several places on the IBM i platform, so patching this should be a priority.

alert.pngHere is the link to the description of the problem and a list of the PTFs that you need to apply for all currently supported platforms. Be sure to read it all the way through.

I recommend that you follow the guidance in this document to turn off Triple DES ciphers.

In addition to patching the IBM instances of OpenSSL and OpenVPN, IBM recommends that you contact your third party software vendors to determine if they have vulnerabilities.

Let me be pre-emptive here:

None of the Townsend Security solutions for the IBM i use the OpenSSL library for secure TLS sessions. Our solutions exclusively use the IBM i System TLS library. Why is that? Well, we ported OpenSSL to the IBM i platform more than 10 years ago. While we were successful with the port to the IBM i, I made a decision that we would not release it nor use it in our IBM i products. OpenSSL is a complex application with many moving parts. I agree with Bruce Schneier that complexity is the enemy of security. It is so easy to introduce a problem in a cryptographic library even if you work in this area a great deal. I felt then, and I feel now, that the maintenance and support of the TLS library should remain with IBM and that the native IBM i System TLS library is the best platform for us and our customers. I believe that this was a good decision then, and remains a good decision now. It has protected our customers from a number of security problems.

Our Managed FTP solution does use the OpenSSH solution for secure FTP sessions. OpenSSH uses the OpenSSL solution for cryptographic operations, but not for secure session establishment and it is not subject to this vulnerability. We use the IBM OpenSSH distribution and not our own port of OpenSSH for the very same reason as above.

I am NOT criticizing the OpenSSL development team. I’ve worked directly with the OpenSSL team over the years and have deep respect for them. They have a gargantuan task in maintaining one of the most widely used secure communications products in the world. Security programming will humble you and, if you are lucky, it will make you risk averse. On the IBM i platform we chose to use the IBM i System TLS library and I still think that was a wise decision.

There are multiple third-party IBM i products that do use their own version of OpenSSL. You need to be talking to them right away. Unfortunately I know of one or two that are no longer supporting the IBM i platform. So you may have some difficulty getting resolution on this issue. I wish you luck in this endeavor!

If you are interested you can read about the SWEET32 attack here.

If you are one of our IBM i customers I recommend that you sign up for our newsletter. If you’ve opted out in the past you might not be getting security notifications from us. When you opt out we honor that request and won’t opt you back in. You have to do that here. 

Patrick

Webinar: Sec

Topics: IBM i


Subscribe to Email Updates

Posts by Topic

see all