IPsec -- A Secure Deployment Option
- IPsec in Multitiered Deployments
- Deployment Scenario
- IPsec Caveats
- SSL and IPsec
- Recommendations
- IPsec Reference
- Acknowledgements
- Author Bio
- Third-Party URLs
Like this article? We recommend
Like this article? We recommend
Like this article? We recommend
In today's IT environments, it is critical to protect data traffic between disparate host systems in multitier applications. Information security teams look for security vulnerabilities and try to manage the risk of data tampering, snooping, and eavesdropping. Plain text data flow of critical information like passwords, credit card numbers, and privacy information between systems is highly vulnerable to misuse. The operating system, application code, and enterprise security policies and procedures all have a role in addressing security issues.
IP Security (IPsec) protects IP packets and defines the means with which a packet will be encrypted. IPsec is a valuable option for protecting data in transit. It is an Internet Engineering Task Force (IETF) standard, and many operating systems vendors have incorporated IPsec in their product offerings.
By using a standards-based implementation, interoperability between heterogeneous operating systems can be handled with lesser pain. However, IPsec is only one piece in the security solution space. The operating system, application code, and enterprise security policies and procedures all need to address security issues.
This article provides an overview of IPsec and how it is used to secure IP traffic between two systems. Software architects can use this information to deploy multitier applications, based on the Sun™ ONE Portal, Identity, Application, and Web servers. IPsec can also be used in setting up a virtual private network (VPN). The description of that setup is beyond the scope of this article. For more information on tunnel mode, refer to the documents listed in "IPsec Reference" on page 13.
IPsec in Multitiered Deployments
IPsec is an IETF standard for securing IP traffic (IPv4 and IPv6). The following information is taken verbatim from "IP Security (IPsec) Components" (see "IPsec Reference" on page 13). This document captures the relevant information about IPsec.
In typical multitier application production deployments, the presentation tier (for example, web servers) communicates with the business logic tier (for example, application servers, servlets, or JavaServer Pages™ software running in web containers). The business logic tier communicates with the data tier (for example, LDAP and RDBMS). The communications use TCP as the transport layer protocol. For the most part, this traffic is not encrypted, so it is subject to eavesdropping, confidential information snooping, and possible tampering. Security studies show that most incidents are caused by people within the enterprise.
You can use IPsec effectively to provide confidentiality and integrity by using interoperable, cryptography-based security.
This solution is applicable across a wide range of products from different vendors. IPsec operates at the network layer, so it is transparent to application. IPsec performs the encryption and decryption using the configured cryptographic algorithms and keys.
Packet Security
IPsec protects IP packets and defines the means with which a packet will be encrypted. There are two methods, either of which can be used to protect IP communication and works at the network layer of the TCP/IP stack:
Encapsulating security payload (ESP)
Authentication header (AH)
In addition, the security association (SA) defines the protection of the message in only one way by using various IPsec mechanisms, and the Internet Key Exchange (IKE) mechanism provides key management and management of various other components.
Encapsulating Security Payload
Encapsulating security payload (ESP) provides authentication, integrity, and confidentiality:
Authentication confirms that the packet received is actually from the sender of the packet. Only the data portion of an IP packet is authenticated. The IP header is not authenticated.
Integrity ensures that data in the packet has not changed in transit.
Confidentiality is achieved by encrypting the message (that is, data).
Only the data is encrypted, which adds to the size of the data. ESP adds its own "header" component between the IP header and other components of the packet before the TCP or UDP header.
ESP uses the following algorithms:
3DES in CBC mode
AES in CBC mode
Blowfish in CBC mode
HMAC with MD5
HMAC with SHA-1
Authentication Header
The authentication header (AH) provides authentication and integrity, as ESP does. However, AH also provides optional security against retransmission of packets by someone else. The AH header is inserted between the IP header and other components of the packet, before the UDP-TCP header. Unlike ESP, some IP header fields might be changed by someone listening to the communication on the Internet or wire. Confidentiality is also not provided.
AH uses authentication algorithms like:
HMAC-MD5
HMAC-SHA-1
Security Association
Security association (SA) defines the one-way protection of the message for a specific IPsec mechanisms. There are usually two SAs used in protecting traffic between two hosts. SA can control what to encrypt and what not to encrypt.
SA consists of the keys, algorithms, and lifetime values. The database containing this information is called the security association database (SADB). An SA is identified by the security parameter index (SPI), the destination IP address, and the protocol information (ESP or AH).
In the Solaris Operating System (Solaris OS), the elements that relate to the SADB are:
ipseckey(1M) to configure SADB
in.iked(1M) to manipulate the SADB
PF_KEY socket interface that enables ipseckey(1M) and in.iked(1M) to perform their tasks
You can implement SA by using the tunnel method or the transport method. In tunnel mode, every communication between two networks behind a host (for example, a firewall or gateway) is encrypted. In transport mode, communication between two computers (hosts) is secured. Only the data part of the packet is encrypted, apart from the IP header. Therefore, the IP header is not protected. However, in tunnel mode, IPsec encapsulates the complete packet including the IP header.
The IP header has source and destination as the IP address of the gateways or the firewalls that exchange the encrypted information. The hosts behind the gateways communicate in plain messages. The applications or hosts behind the gateway do not need to encrypt or decrypt. IPsec AH and ESP can be used in transport or tunnel mode.
Security Policy Database
The SA(s) used in transmitting and receiving packets is controlled by the security policy. The security policy database (SPD) contains this information. The SPD entries define the source, destination addresses, port numbers, and action to be taken (for instance, allow, permit, drop, or bypass).
In the Solaris OS, the elements that relate to SPD are:
ipsecconf(1M) command to set up a policy for a host
Per-socket policy
{encr,encr_auth,auth}_algs keywords to ifconfig(1M)
Internet Key Exchange
IPsec uses the Internet Key Exchange (IKE) protocol to automate the SA setup and to exchange keys between the two ends (that is, the hosts and gateway). This ensures that, by using keys, the sender and receiver only get the cipher text. IKE can also recreate the keys and exchange again to make sure that even if someone breaks the keys, the keys are already recreated before they are broken again. IKE basically manages this process of updating and recreating the keys.
In IKE, the SAs are set up first. Then, the traffic can be secured. The two hosts or gateways finalize on the encryption and authentication mechanisms. The two gateways then authenticate each other by a mechanism already known to each other. The Diffie-Hellman (DH) public-key algorithm then generates the shared master key. The same master key is then used to define the IPsec keys for SAs.
The two gateways decide on the encryption and authentication mechanism to be used in SAs. The master key is used to get the IPsec keys for the SA. Hence, secure communication is established.
The same master key can be used for the IPsec keys for SAs, or a new DH can be used. If the new Diffie-Hellman is used, the property of perfect forward secrecy (PFS) is preserved.
For detailed information on setting up IKE with pre-shared keys and public keys, refer to the IPsec and IKE Administration Guide at:
http://docs.sun.com/db/doc/816-7264?q=IPSEC
Trade-Offs
The nature and value of business data drive the need for securing data traffic inside a corporate network. Some examples of drivers for enhanced enterprise security policies include legal mandates, a need to protect against unauthorized disclosures, and demands or requirements of application owners and users.
To secure data in transit, vendors provide proprietary protocols and tools. For example, sqlnet can be configured with an encryption option to protect data flow between an ORACLE® client and server. Application security methods, though not public and possibly not based on industry standards, might still offer the needed security. If multiple applications run on a single server system, each might have its own data encryption and protection for the data in transit. The advantages of IPsec are that it is standards-based, available across different operating systems, and totally transparent to applications.
The Secure Sockets Layer (SSL) is an application layer protocol that can also be used to encrypt data flow. The client and server components must be SSL aware. They must be built with SSL libraries and with proper versions to interoperate. SSL is used to protect the web browser and web server data flow (for example, HTTPS protocol). Most of the SSL implementations require certificates for the server and, in some cases, the clients too. IPsec can use certificates, but it can be implemented without certificates by using manual-key information.
In a typical corporate applications environment, multiple software products from different vendors run on different operation systems. It is more appropriate to use a technology that is transparent to applications to secure data flow. Some software applications cannot use SSL. Any kind of encryption of data flow impacts performance. This also applies to IPsec.
Hardware crypto-accelerators are useful for performance tuning. For example, you can use the Sun™ Crypto Accelerator 4000 board. The application load characteristics, performance characteristics, and service-level agreement (SLA) requirements should be considered with this security-to-performance evaluation.
A specific trade off in deploying IPsec is the keys used for encryption and decryption. IKE-based key management might not be available in the target IPsec implementation. If you manually generate keys, be careful to keep them secure, and change them on a regular schedule.