Exchange Network Port Reference

Applies to: Exchange Server 2010

This topic provides information about ports, authentication, and encryption for all data paths used by Microsoft Exchange Server 2010. The Notes sections following each table clarify or define non-standard authentication or encryption methods.

Transport Servers

Exchange 2010 includes two server roles that perform message transport functionality: Hub Transport server and Edge Transport server.

The following table provides information about ports, authentication, and encryption for data paths between these transport servers and other Exchange 2010 servers and services.

Transport server data paths

Notes on Transport Servers

• All traffic between Hub Transport servers is encrypted by using TLS with self-signed certificates that are installed by Exchange 2010 Setup.

  Note

  In Exchange 2010, TLS can be disabled on Hub Transport servers for internal SMTP communication with other Hub Transport servers in the same Exchange organization. We don't recommend doing this unless absolutely required. For more information, see Disabling TLS Between Active Directory Sites to Support WAN Optimization.

• All traffic between Edge Transport servers and Hub Transport servers is authenticated and encrypted. Mutual TLS is the underlying mechanism for authentication and encryption. Instead of using X.509 validation, Exchange 2010 uses direct trust to authenticate the certificates. Direct trust means that the presence of the certificate in Active Directory or Active Directory Lightweight Directory Services (AD LDS) acts as validation for the certificate. Active Directory is considered a trusted storage mechanism. When direct trust is used, it doesn't matter if the certificate is self-signed or signed by a certification authority (CA). When you subscribe an Edge Transport server to the Exchange organization, the Edge Subscription publishes the Edge Transport server certificate in Active Directory for the Hub Transport servers to validate. The Microsoft Exchange EdgeSync service updates AD LDS with the set of Hub Transport server certificates for the Edge Transport server to validate.

• EdgeSync uses a secure LDAP connection from the Hub Transport server to subscribed Edge Transport servers over TCP 50636. AD LDS also listens on TCP 50389. Connections to this port don't use SSL. You can use LDAP utilities to connect to the port and check AD LDS data.

• By default, traffic between Edge Transport servers in two different organizations is encrypted. Exchange 2010 Setup creates a self-signed certificate, and TLS is enabled by default. This allows any sending system to encrypt the inbound SMTP session to Exchange. By default, Exchange 2010 also tries TLS for all remote connections.

• Authentication methods for traffic between Hub Transport servers and Mailbox servers differ when the Hub Transport server roles and Mailbox server roles are installed on the same computer. When mail submission is local, Kerberos authentication is used. When mail submission is remote, NTLM authentication is used.

• Exchange 2010 also supports Domain Security. Domain Security refers to the functionality in Exchange 2010 and Microsoft Outlook 2010 that provides a low-cost alternative to S/MIME or other message-level over-the-Internet, security solutions. Domain Security provides you with a way to manage secure message paths between domains over the Internet. After these secure message paths are configured, messages that have successfully traveled over the secure path from an authenticated sender are displayed to Outlook and Outlook Web Access users as "Domain Secured". For more information, see Understanding Domain Security.

• Many agents can run on Hub Transport servers and Edge Transport servers. Generally, anti-spam agents rely on information that's local to the computer on which the agents run. Therefore, little communication with remote computers is required. Recipient filtering is the exception. Recipient filtering requires calls to either AD LDS or Active Directory. As a best practice, run recipient filtering on the Edge Transport server. In this case, the AD LDS directory is on the same computer as the Edge Transport server and no remote communication is required. When recipient filtering has been installed and configured on the Hub Transport server, recipient filtering accesses Active Directory.

• The Protocol Analysis agent is used by the Sender Reputation feature in Exchange 2010. This agent also makes various connections to outside proxy servers to determine inbound message paths for suspect connections.

• All other anti-spam functionality uses data gathered, stored, and accessed only on the local computer. Frequently, the data, such as safelist aggregation or recipient data for recipient filtering, is pushed to the local AD LDS directory by using the Microsoft Exchange EdgeSync service.

• Information Rights Management (IRM) agents on Hub Transport servers make connections to Active Directory Rights Management Services (AD RMS) servers in the organization. AD RMS is a Web service that's secured by using SSL as a best practice. Communication with AD RMS servers occurs by using HTTPS, and Kerberos or NTLM is used for authentication, depending on the AD RMS server configuration.

• Journal rules, transport rules, and message classifications are stored in Active Directory and accessed by the Journaling agent and the Transport Rules agent on Hub Transport servers.

Mailbox Servers

Whether NTLM or Kerberos authentication is used for Mailbox servers depends on the user or process context that the Exchange Business Logic layer consumer is running under. In this context, the consumer is any application or process that uses the Exchange Business Logic layer. As a result, many entries in the Default Authentication column of the Mailbox server data paths table are listed as NTLM/Kerberos.

The Exchange Business Logic layer is used to access and communicate with the Exchange store. The Exchange Business Logic layer is also called from the Exchange store to communicate with external applications and processes.

If the Exchange Business Logic layer consumer is running as Local System, the authentication method is always Kerberos from the consumer to the Exchange store. Kerberos is used because the consumer must be authenticated by using the Local System computer account, and a two-way authenticated trust must exist.

If the Exchange Business Logic layer consumer isn't running as Local System, the authentication method is NTLM. For example, NTLM is used when you run an Exchange Management Shell cmdlet that uses the Exchange Business Logic layer.

The RPC traffic is always encrypted.

The following table provides information about ports, authentication, and encryption for data paths to and from Mailbox servers

Mailbox server data paths

Notes on Mailbox Servers

• The Clustering data path listed in the preceding table uses dynamic RPC over TCP to communicate cluster status and activity between the different cluster nodes. The Cluster service (ClusSvc.exe) also uses UDP/3343 and randomly allocated high TCP ports to communicate between cluster nodes.
• For intra-node communications, cluster nodes communicate over User Datagram Protocol (UDP) port 3343. Each node in the cluster periodically exchanges sequenced, unicast UDP datagrams with every other node in the cluster. The purpose of this exchange is to determine whether all nodes are running correctly and to monitor the health of network links.
• Port 64327/TCP is the default port used for log shipping. Administrators can specify a different port for log shipping.
• For HTTP authentication where Negotiate is listed, Kerberos is tried first, and then NTLM.

Client Access Servers

Unless noted, client access technologies, such as Outlook Web App, POP3, or IMAP4, are described by the authentication and encryption from the client application to the Client Access server.

The following table provides information about port, authentication, and encryption for data paths between Client Access servers and other servers and clients.

Client Access server data paths

Notes on Client Access Servers

• In Exchange 2010, MAPI clients such as Microsoft Outlook connect to Client Access servers.

• The Client Access servers use many ports to communicate with Mailbox servers. With some exceptions, those ports are determined by the RPC service and aren't fixed.

• For HTTP authentication where Negotiate is listed, Kerberos is tried first, and then NTLM.

• When an Exchange 2010 Client Access server communicates with a Mailbox server running Exchange Server 2003, it's a best practice to use Kerberos and disable NTLM authentication and Basic authentication. Additionally, it's a best practice to configure Outlook Web App to use forms-based authentication with a trusted certificate. For Exchange ActiveSync clients to communicate through the Exchange 2010 Client Access server to the Exchange 2003 back-end server, Windows Integrated Authentication must be enabled on the Microsoft-Server-ActiveSync virtual directory on the Exchange 2003 back-end server. To use Exchange System Manager on an Exchange 2003 server to manage authentication on an Exchange 2003 virtual directory, download and install the hot fix referenced in Microsoft Knowledge Base article 937031, Event ID 1036 is logged on an Exchange 2007 server that is running the CAS role when mobile devices connect to the Exchange 2007 server to access mailboxes on an Exchange 2003 back-end server.

  Note

  Although the Knowledge Base article is specific to Exchange 2007, it's also applicable to Exchange 2010.

• When a Client Access server proxies POP3 requests to another Client Access server, the communication occurs over port 995/TCP, regardless of whether the connecting client uses POP3 and requests TLS (on port 110/TCP) or connects on port 995/TCP using SSL. Similarly, for IMAP4 connections, port 993/TCP is used to proxy requests regardless of whether the connecting client uses IMAP4 and requests TLS (on port 443/TCP) or connects on port 995 using IMAP4 with SSL encryption

Unified Messaging Servers

IP gateways and IP PBXs support only certificate-based authentication that uses mutual TLS for encrypting SIP traffic and IP-based authentication for Session Initiation Protocol (SIP)/TCP connections. IP gateways don't support either NTLM or Kerberos authentication. Therefore, when you use IP-based authentication, the connecting IP address or addresses are used to provide authentication mechanism for unencrypted (TCP) connections. When IP-based authentication is used in Unified Messaging (UM), the UM server verifies that the IP address is allowed to connect. The IP address is configured on the IP gateway or IP PBX.

IP gateways and IP PBXs support mutual TLS for encrypting SIP traffic. After you successfully import and export the required trusted certificates, the IP gateway or IP PBX will request a certificate from the UM server, and then it will request a certificate from the IP gateway or IP PBX. Exchanging the trusted certificate between the IP gateway or IP PBX and the UM server enables the IP gateway or IP PBX and UM server to communicate over an encrypted connection by using mutual TLS.

The following table provides information about port, authentication, and encryption for data paths between UM servers and other servers.

Unified Messaging server data paths

Notes on Unified Messaging Servers

• When you create a UM IP gateway object in Active Directory, you must define the IP address of the physical IP gateway or IP PBX (Private Branch eXchange). When you define the IP address on the UM IP gateway object, the IP address is added to a list of valid IP gateways or IP PBXs (also called SIP peers) that the UM server is allowed to communicate with. When you create the UM IP gateway, you can associate it with a UM dial plan. Associating the UM IP gateway with a dial plan allows the Unified Messaging servers that are associated with the dial plan to use IP-based authentication to communicate with the IP gateway. If the UM IP gateway has not been created or it isn't configured to use the correct IP address, authentication fails and the UM servers don't accept connections from that IP gateway's IP address. Also, when you implement mutual TLS and IP gateway or IP PBX and UM servers, the UM IP gateway must be configured to use the FQDN. After you configure the UM IP gateway with an FQDN, you must also add a host record to the DNS forward lookup zone for the UM IP gateway.
• In Exchange 2010, a UM server can either communicate on port 5060/TCP (unsecured) or on port 5061/TCP (secured), and can be configured to use both.

For more information, see Understanding Unified Messaging VoIP Security and Understanding Protocols, Ports, and Services in Unified Messaging.

Windows Firewall Rules Created by Exchange 2010 Setup

Windows Firewall with Advanced Security is a stateful, host-based firewall that filters inbound and outbound traffic based on firewall rules. Exchange 2010 Setup creates Windows Firewall rules to open the ports required for server and client communication on each server role. Therefore, you no longer need to use the Security Configuration Wizard (SCW) to configure these settings. To learn more about Windows Firewall with Advanced Security, see Windows Firewall with Advanced Security and IPsec.

This table lists the Windows Firewall rules created by Exchange Setup, including the ports opened on each server role. You can view these rules using the Windows Firewall with Advanced Security MMC snap-in.

Notes on Windows Firewall Rules Created by Exchange 2010 Setup

• On servers that have Internet Information Services (IIS) installed, Windows opens the HTTP (port 80, TCP) and HTTPS (port 443, TCP) ports. Exchange 2010 Setup doesn't open these ports. Therefore, these ports don't appear in the preceding table.
• On Windows Server 2008 and Windows Server 2008 R2, Windows Firewall with Advanced Security allows you to specify the process or service for which a port is opened. This is more secure because it restricts usage of the port to the process or service specified in the rule. Exchange Setup creates firewall rules with the process name specified. In some cases, an additional rule that isn't restricted to the process is also created for compatibility purposes. You can disable or remove the rules that aren't restricted to the processes and keep the corresponding rules restricted to processes if your deployment supports them. The rules not restricted to processes are distinguished by the word (GFW) in the rule name.
• A number of Exchange services use remote procedure calls (RPCs) for communication. Server processes that use RPCs contact the RPC Endpoint Mapper to receive dynamic endpoints and register those endpoints in the Endpoint Mapper database. RPC clients contact the RPC Endpoint Mapper to determine the endpoints used by the server process. By default, the RPC Endpoint Mapper listens on port 135 (TCP). When configuring the Windows Firewall for a process that uses RPCs, Exchange 2010 Setup creates two firewall rules for the process. One rule allows communication with the RPC Endpoint Mapper, and the other rule allows communication with the dynamically assigned endpoint. To learn more about RPCs, see How RPC Works. For more information about creating Windows Firewall rules for dynamic RPC, see Allowing Inbound Network Traffic that Uses Dynamic RPC.

For more information, see Microsoft Knowledge Base article 179442, How to configure a firewall for domains and trusts.