Achieving Unified CCE system security requires an effective security policy that accurately defines access, connection requirements, and systems management within your contact center. Once you have a good security policy, you can use many state-of-the-art Cisco technologies and products to protect your data center resources from internal and external threats and to ensure data privacy, integrity, and system availability.
As one of those applications in the Cisco Unified Communications network, Unified CCE security considerations at a high level are not very different than those of other applications making up a Cisco Unified Communications solution. Deployments of Unified CCE vary greatly and often call for complex network designs that require competence in all areas of Layer 2 and Layer 3 networking as well as voice, VPN, QoS, Microsoft Windows Active Directory, and so forth. While this chapter provides some guidance that may touch on these various areas, it is not meant to be an all-inclusive guide for deploying a secure Unified CCE network.
Along with the Unified Communications Security Solution portal, use
other Cisco solution reference network design guides (SRNDs) in addition to this document to answer many design and deployment questions. The SRNDs provide proven best practices for building a network infrastructure for Cisco Unified Communications. Among the SRNDs at this site are the following relevant documents relating to security and Cisco Unified Communications, so refer to them to successfully deploy a Unified CCE network:
Cisco Unified Communications SRND Based on Cisco Unified Communications Manager
Data Center Networking: Server Farm Security SRNDv2
Site-to-Site IPSec VPN SRND
Voice and Video Enabled IPSec VPN (V3PN) SRND
Business Ready Teleworker SRND
Updates and additions to these documents are posted periodically, so visit the SRND web site frequently.
This chapter provides limited guidance on the intricacies of designing and deploying a Windows Active Directory. Additional information is available from Microsoft on designing a new Active Directory logical structure, deploying Active Directory for the first time, upgrading an existing Windows environment to Windows Server 2008 R2 Active Directory, and restructuring your current environment to a Windows Active Directory environment. In particular, the Designing and Deploying Directory and Security Services section of the Microsoft Windows Server 2008R2 Deployment Kit can assist you in meeting all of the Active Directory design and deployment goals for your organization. This development kit and its related documentation are available from Microsoft.
An adequately secure Unified CCE deployment requires a multilayered approach to protecting systems and networks from targeted attacks and the propagation of viruses, among other threats. The goal of this chapter is to stress the various areas pertinent to securing a Unified CCE deployment, but it does not delve into the details of each area. Specific details can be found in the relevant product documentation.
Implement the following security layers and establish policies around them:
Physical Security You must ensure that the servers hosting the Cisco contact center applications are physically secure. They must be located in data centers to which only authorized personnel have access. The cabling plant, routers, and switches also have controlled access. Implementing a strong physical-layer network security plan also includes utilizing such things as port security on data switches.
Perimeter Security While this document does not delve into the details on how to design and deploy a secure data network, it does provide references to resources that can aid in establishing an effective secure environment for your contact center applications.
Data Security To ensure an increased level of protection from eavesdropping for customer-sensitive information, Unified CCE provides support for Transport Layer Security (TLS) on the CTI OS and Cisco Agent Desktops. It also supports IPSec to secure communication channels between servers.
Host-Based Firewall Users wishing to take advantage of the Windows Firewall to protect from malicious users and programs that use unsolicited incoming traffic to attack servers can use the Windows Firewall Configuration Utility on servers or the Agent Desktop Installers to integrate with the firewall component of Windows Server 2008 R2, and Windows XP SP3, respectively.
Patch Management A system is typically not connected to a live network until all security updates have been applied. It is important for all hosts to be kept up-to-date with Microsoft (Windows, SQL Server, Internet Explorer, and so forth) and other third-party security patches.
For most of these security layers, the Unified CCE solution supports a number of capabilities. However, what Cisco cannot control or enforce is your enterprise policies and procedures for deploying and maintaining a secure Unified CCE solution.
Before discussing how to design the various security layers required for a Unified CCE network, this section introduces the differences that are inherent in the applications making up the Unified CCE solution.
The Unified CCE solution consists of a number of application servers that are managed differently. The primary servers, those with the most focus in this document, are the Routers, Loggers (also known as Central Controllers), Peripheral Gateways, Administration & Data Servers, and so forth. These application servers can be installed only on a standard (default) operating system installation. All installations can be done on Windows Server 2008 R2 Standard or Enterprise Edition. The maintenance of this operating system in terms of device drivers, security updates, and so forth, is the responsibility of the customer, as is acquiring the necessary software from the appropriate vendors. This category of application servers is the primary focus of this topic.
The secondary group of servers, those running applications that are part of the solution but that are deployed differently, are Cisco Unified Communications Manager (Unified CM), Cisco Unified IP IVR, and so forth. Customers are required to obtain all relevant patches and updates to this operating system from Cisco. The security hardening specifications for this operating system can be found in the Cisco Unified Communications Solution Reference Network Design (SRND) guide and other Unified CM product documentation.
The approach to securing the Unified CCE solution as it pertains to the various layers listed above differs from one group of servers to another. It is useful to keep this in mind as you design, deploy, and maintain these servers in your environment. Cisco is constantly enhancing its Unified Communications products with the eventual goal of having them all support the same customized operating system, antivirus applications, and security path management techniques. Some examples of these enhancements include the support of Cisco's host-based intrusion prevention software (Cisco Security Agent) and default server hardening provided by the customized operating system or applications.
Security Best Practices
As part of the Unified CCE 9.0 documentation set, Cisco has released a best-practices guide for the primary group of servers, which covers a number of areas pertaining to the new implementation in the release along with some general guidance for securing a Unified CCE deployment. The best-practices guide includes the following topics:
The guidelines contained in this guide are based in part on hardening guidelines published by Microsoft, such as those found in the Windows Server 2008 Security Guide, as well as other third-party vendors’ hardening guidelines. It also serves as a reference point for most of the security functionality in the product. The guide is also the installation guide for the Automated OS and SQL Security Hardening bundled with the application installer, Windows Firewall Configuration Utility, the SSL Configuration Utility, the Network Isolation IPSec Utility, and the Unified CC Security Wizard.
There are several important factors to consider when deploying firewalls in a Unified CCE network. The application servers making up a Unified CCE solution are not meant to reside in a demilitarized zone (DMZ) and must be segmented from any externally visible networks and internal corporate networks. The application servers must be placed in data centers, and the applicable firewalls or routers must be configured with access control lists (ACL) to control the traffic that is targeted to the servers, thereby allowing only designated network traffic to pass through.
Deploying the application in an environment in which firewalls are in place requires the network administrator to be knowledgeable about which TCP/UDP IP ports are used, firewall deployment and topology considerations, and impact of Network Address Translation (NAT).
To aid in firewall configuration, these guides list the protocols and ports used for agent desktop-to-server communication, application administration, and reporting. They also provide a listing of the ports used for intra-server communication.
The deployment in
Figure 1 represents the placement of firewalls and other network infrastructure components in a Unified CCE deployment. The design model in Figure 116 incorporates a parent Unified ICM system with legacy peripheral hosts and a child Cisco Unified Contact Center Enterprise with a Unified CM cluster. The following best practices apply to this type of deployment:
Block the following ports at the enterprise perimeter firewall:
UDP ports 135, 137, 138, and 445
TCP ports 135, 139, 445, and 593
Deploy Layer-3 and Layer-4 ACLs that are configured as described in the port guides.
Isolate database and web services by installing dedicated historical data servers.
Minimize the number of Administration & Data Servers (ADS) and make use of Administration Clients (no database required) and Internet script editor clients.
Use the same deployment guidelines when the parent Unified ICM or child Unified CCE central controllers are geographically distributed.
Deploy Windows IPSec (ESP) to encrypt intra-server communications. The use of hardware off-load network cards is required to minimize the impact of encryption on the main CPU and to sustain the load level (including number of agents and call rate) that is supported with the Unified CCE system. See IPSec Deployment for a more detailed diagram and further information.
Use Cisco IOS IPSec for site-to-site VPNs between geographically distributed sites, remote branch sites, or outsourced sites.
Network Address Translation
Network Address Translation (NAT) is a feature that resides on a network router and permits the use of private IP addressing. A private IP address is an IP address that cannot be routed on the Internet. When NAT is enabled, users on the private IP network can access devices on the public network through the NAT router.
When an IP packet reaches the NAT-enabled router, the router replaces the private IP address with a public IP address. For applications such as HTTP or Telnet, NAT does not cause problems. However, applications that exchange IP addresses in the payload of an IP packet experience problems because the IP address that is transmitted in the payload of the IP packet is not replaced. Only the IP address in the IP header is replaced.
To overcome this problem, Cisco IOS-based routers and PIX/ASA firewalls implement fix-ups for a variety of protocols and applications including SCCP and CTIQBE (TAPI/JTAPI). The fix-up allows the router to look at the entire packet and replace the necessary addresses when performing the NAT operation. For this process to work, the version of Cisco IOS or PIX/ASA must be compatible with the Unified CM version.
While Unified ICM and Unified CCE systems may still be deployed in a dedicated Windows Active Directory domain, it is not a requirement. What makes this possible is the capability of the software security principals to be installed in Organizational Units. This closer integration with AD and the power of security delegation means that corporate AD directories can be used to house application servers (for domain membership), user and service accounts, and groups.
The deployment of parent/child systems can be done on the same AD Domain or Forest, but they may also be deployed in totally disparate AD environments. The scenario where this deployment is common occurs when the child Unified CCE system is housed at an outsourced contact center site. In this case, the Gateway PG that is a parent node is a member of the parent AD domain. (Do not use Workgroup membership due to the administration limitations.) This type of deployment is common today for having remote branch offices with PGs that are added as members of the central site's domain to which the Routers, Loggers, and Distributors are members.
The topology shown in Figure 1 attempts to represent the AD Boundaries for each of the two AD domains involved in this deployment and to which domain the application servers are joined. The parent AD Domain Boundary is extended beyond the central data center site to include the Unified ICM Central Controllers and accompanying servers as well as the ACD PG (at the legacy site) and Gateway PG at the child Unified CCE site. The child Unified CCE site and its AD Boundary have the Unified CCE servers as members. This may or may not be as part of an outsourcer’s corporate AD environment. Of course, it may also be a dedicated AD domain for Unified CCE.
AD Site Topology
In a geographically distributed deployment of Unified ICM or Unified CCE, redundant domain controllers must be located at each of the sites, and properly configured Inter-Site Replication Connections must be established with a Global Catalog at each site. The Unified CCE application is designed to communicate with the AD servers that are in their site, but this requires an adequately implemented site topology in accordance with Microsoft guidelines.
The installation of Unified ICM or Unified CCE software requires that the AD Domain in which the servers are members must be in Native Mode. The installation will add a number of OU objects, containers, users, and groups that are necessary for the operation of the software. Adding these objects can be done only in an Organizational Unit in AD over which the user running the install program has been delegated control. The OU can be located anywhere in the domain hierarchy, and the AD Administrator determines how deeply nested the Unified ICM/Unified CCE OU hierarchy is created and populated.
Local Server Accounts and groups are not created on the application servers. All created groups are Domain Local Security Groups, and all user accounts are domain accounts. The Service Logon domain account is added to the Local Administrators' group of the application servers.
Unified ICM and Unified CCE software installation is integrated with a Domain Manager tool that can be used standalone for pre-installing the OU hierarchies and objects required by the software, or can be used when the Setup program is invoked to create the same objects in AD. The AD/OU creation can be done on the domain in which the running server is a member or on a trusted domain.
Do not confuse the creation of AD objects with Group Policy Objects (GPO). The Automated Security Hardening, which is provided and follows the standard Microsoft Security Template format, is not added to AD as part of the software installation through the configuration of a GPO. The security policy provided by this customized template (for Unified ICM/Unified CCE applications) is applied locally when a user chooses to apply hardening, or it can be pushed down through a GPO through manual AD configuration using the provided policy file, CiscoICM_Security_Template.inf.
AD Administrator Created
As mentioned, there are certain AD objects that may be created by an administrator. The primary example in Figure 116 is represented by an OU container, Unified CCE Servers, which is manually added to contain the servers that are members of a given domain. These servers must be moved to this OU once they are joined to the domain. This ensures that some segregation is applied to control who can or cannot administer the servers (delegation of control) and, most importantly, which AD Domain Security Policy can or cannot be inherited by these application servers that are in the OU.
As noted before, Unified ICM/Unified CCE servers ship with a customized security policy. This policy can be applied at this server OU level through a Group Policy Object (GPO), but any differing policies must be blocked from being inherited at the Unified ICM/Unified CCE Servers' OU. Keep in mind that blocking inheritance, a configuration option at the OU object level, can be overridden when the Enforced/No Override option is selected at a higher hierarchy level. The application of group policies must follow a very well thought-out design that starts with the most common denominator, and those policies must be restrictive only at the appropriate level in the hierarchy. For a more in-depth explanation on how to deploy group policies properly, see the Windows Server 2008 R2 Security Guide.
Figure 1. Active Directory and Firewall Deployment Topology
The following notes apply to the figure above:
Cisco_ICM organizational unit object hierarchies are created by the application setup.
Unified ICM Servers and Unified CCE Servers organizational unit objects must be created by the AD administrators to separately apply custom Cisco Unified ICM Security Policies through a GPO if required.
Flexible Single Master Operation servers must be distributed across Domain Controllers in the appropriate sites according to Microsoft guidelines.
The Unified CCE solution relies on Microsoft Windows IPSec and/or Cisco IOS IPSec to secure critical links between application servers and sites. The solution can be secured either by deploying peer-to-peer IPSec tunnels between the servers and sites, or by deploying a more restrictive and preconfigured Network Isolation IPSec policy, or by using a combination of both. The peer-to-peer IPSec deployment requires manual configuration for each communication path that needs to be secured, using the tools provided by Microsoft. However, the Network Isolation IPSec policy can be deployed automatically on each Server By using the Network Isolation IPSec utility, and it secures all communication paths to or from that server unless an exception is made. The Network Isolation IPSec utility is installed by default on all Unified CCE 8.0 servers.
This guide not only lists the supported paths, but also information to help users deploy Windows IPSec, including appropriate settings and much more.
Figure 1 shows a number of connection paths where IPSec is supported. The figure below illustrates the guidelines provided in this chapter and shows the various server interconnections that must be secured with either Windows IPSec or Cisco IOS IPSec. The diagram also shows a number of paths that support TLS. More information about TLS support can be found in Endpoint Security.
Figure 2. IPSec Deployment Example
By providing host firewall protection on the innermost layer of your network, Windows Firewall can be an effective part of your defense-in-depth security strategy. Unified CCE supports the deployment of Windows Firewall on the application servers. The Security Best Practices Guide for Cisco Unified ICM/Contact Center Enterprise & Hosted Release 9.x(y) contains a chapter on the implementation and configuration of this feature.
The configuration of the exceptions and the opening of the ports required by the application will still be done locally using the Windows Firewall Configuration Utility, which is included with the Unified CCE application.
The Windows Firewall is set up during Unified CCE installation, during which required ports are opened.
The Unified Contact Center Security Wizard allows easy configuration of the security features defined above, namely, SQL Server Hardening, Windows Firewall configuration, and Network Isolation IPSec policy deployment. The Security Wizard encapsulates the functionality of these four utilities in an easy-to-use wizard-like interface that guides the user with the steps involved in configuring the security feature. (This is particularly helpful when deploying the Network Isolation IPSec policy.) The Security Wizard is installed by default with Unified CCE 8.0 and later. The Security Best Practices Guide for Cisco Unified ICM/Contact Center Enterprise & Hosted Release 9.x(y) contains a chapter explaining the Security Wizard in detail.
A number of third-party antivirus applications are supported for the Unified CCE system. For a list of applications and versions supported on your particular release of the Unified CCE software, see the Hardware & System Software Specification (Bill of Materials) for Cisco Unified ICM/Contact Center Enterprise & Hosted, Release 9.0(x)) and the Hardware and System Software Specification for Cisco Unified Customer Voice Portal (Unified CVP), as well as the Cisco Unified CCX and Unified CM product documentation for the applications supported. These documents are available on cisco.com.
Deploy only the supported applications for your environment, otherwise a software conflict might arise.
Antivirus applications have numerous configuration options that allow very granular control of what and how data must be scanned on a server.
With any antivirus product, configuration is a balance of scanning versus the performance of the server. The more you choose to scan, the greater the potential performance overhead. The role of the system administrator is to determine what the optimal configuration requirements for installing an antivirus application within a particular environment. See the Security Best Practices Guide for Cisco Unified ICM/Contact Center Enterprise & Hosted, Release 9.x(y) and your particular antivirus product documentation for more detailed configuration information about a Unified ICM environment.
The following list highlights some general best practices:
Upgrade to the latest supported version of the third-party antivirus application. Newer versions improve scanning speed over previous versions, resulting in lower overhead on servers.
Avoid scanning of any files accessed from remote drives (such as network mappings or UNC connections). Where possible, each of these remote machines must have its own antivirus software installed, thus keeping all scanning local. With a multitiered antivirus strategy, scanning across the network and adding to the network load might not be required.
Due to the higher scanning overhead of heuristics scanning over traditional antivirus scanning, use this advanced scanning option only at key points of data entry from untrusted networks (such as email and Internet gateways).
Real-time or on-access scanning can be enabled, but only on incoming files (when writing to disk). This is the default setting for most antivirus applications. Implementing on-access scanning on file reads will yield a higher impact on system resources than necessary in a high-performance application environment.
While on-demand and real-time scanning of all files gives optimum protection, this configuration does have the overhead of scanning those files that cannot support malicious code (for example, ASCII text files). Exclude files or directories of files in all scanning modes that are known to present no risk to the system. Also, follow the guidelines for which specific Unified CCE files to exclude in Unified CCE implementation, as provided in the Security Best Practices Guide for Cisco Unified ICM/Contact Center Enterprise & Hosted, Release 9.x(y).
Schedule regular disk scans only during low usage times and at times when application activity is lowest. To determine when application purge activity is scheduled, see the Security Best Practices Guide for Cisco Unified ICM/Contact Center Enterprise & Hosted, Release 9.x(y) listed in the previous item.
Cisco does not test or support intrusion prevention products by vendors such as Sygate, McAfee, and so on. Such products are capable of blocking legitimate application functionality if they incorrectly identify that application as a security threat. These products must be configured to allow legitimate operations to execute.
This process applies to the application servers running the standard Windows Operating System, not the customized Cisco Unified Communications operating system (CIPT OS).
Follow Microsoft guidelines regarding when and how to apply updates. All Contact Center customers must separately assess all security patches released by Microsoft and install those deemed appropriate for their environments. For information about tracking Cisco-supported operating system files, SQL Server, and security files, see Cisco IP Telephony Operating System, SQL Server, Security Updates.
The Security Patch and Hotfix Policy for Unified CM specifies that any applicable patch deemed Severity 1 or Critical must be tested and posted to cisco.com within 24 hours as Hotfixes. All applicable patches are consolidated and posted once per month as incremental Service Releases.
Automated Patch Management
Unified CCE servers (except for the applications installed on the CIPT OS) support integration with Microsoft's Windows Server Update Services, whereby customers control which patches can be deployed to those servers and when the patches can be deployed.
Selectively approve updates and determine when they get deployed on production servers. The Windows Automatic Update Client (installed by default on all Windows hosts) can be configured to retrieve updates by polling a server that is running Microsoft Window Update Services in place of the default Windows Update Web site.
The Cisco Unified Communications Operating System configuration and patch process does not currently allow for an automated patch management process.
The CTI OS (C++/COM toolkit) and CAD agent desktops both support TLS encryption to the server. This encryption protects agent login and CTI data from snooping. A mutual authentication mechanism was implemented for the CTI OS Server And client to agree on a cipher suite used for authentication, key exchange, and stream encryption. The cipher suite used is as follows:
Key exchange: DH
Encryption: AES (128)
Message digest algorithm: SHA1
The following figure shows the encryption implementation's use of X.509 certificates on the agent desktops as well as on the servers. The implementation supports the integration with a Public Key Infrastructure (PKI) for the most secure deployment. By default, the application will install and rely on a self-signed certificate authority (CA) used to sign client and server requests. However, Cisco supports integration with a third-party CA. This is the preferred method due to the increased security provided by a corporate-managed CA or external authority such as Verisign.
The following figure shows the Certificate Authority enrollment procedure to generate certificates used by the agent and the servers. The agent desktop certificate enrollment process is manual, requiring the creation of certificate signing requests (CSRs) at each endpoint, which are then transferred to the certificate authority responsible for signing and generating the certificates.
Cisco Finesse Release 9.1(1) and later supports HTTPS for the Administration Console and Agent and Supervisor Desktops. HTTPS is not supported for Agent and Supervisor Desktops in large deployments (over 1000 agents).
Unified IP Phone Device Authentication
When designing a Unified CCE solution based on Unified CM Release 7.x or 8.0, customers may choose to implement device authentication for the Cisco Unified IP Phones. Unified CCE 9.0 supports Unified CM’s Authenticated Device Security Mode, which ensures the following:
Device Identity — Mutual authentication using X.509 certificates
Signaling Integrity — SCCP/SIP messages authenticated using HMAC-SHA-1
Signaling Privacy — SCCP/SIP message content encrypted using AES-128-CBC
Unified IP Phone Media Encryption
Media Encryption may be used with Unified CCE; however, it prevents the use of the Silent Monitoring feature. Also, if you are deploying a recording system, contact the recording system vendor to verify support for recording in an environment with Secure Real-Time Transport Protocol (SRTP).
IP Phone Hardening
The IP phone device configuration in Unified CM provides the ability to disable a number of phone features to harden the phones, such as disabling the phone's PC port or restricting a PC from accessing the voice VLAN. Changing some of these settings can disable the monitoring/recording feature of the Unified CCE solution. The settings are defined as follows:
PC Voice VLAN Access
Indicates whether the phone will allow a device attached to the PC port to access the Voice VLAN. Disabling Voice VLAN Access will prevent the attached PC from sending and receiving data on the Voice VLAN. It will also prevent the PC from receiving data sent and received by the phone. Disabling this feature will disable desktop-based monitoring and recording.
Setting: Enabled (default)
Span to PC Port
Indicates whether the phone will forward packets transmitted and received on the Phone Port to the PC Port. To use this feature, PC Voice VLAN access must be enabled. Disabling this feature will disable desktop-based monitoring and recording.
Disable the following setting to prevent man-in-the-middle (MITM) attacks unless the third-party monitoring and/or recording application deployed uses this mechanism for capturing voice streams. The CTI OS Silent Monitoring feature and CAD Silent Monitoring and Recording do not depend on Gratuitous ARP.
Indicates whether the phone will learn MAC addresses from Gratuitous ARP responses.