Exchange 2000 Server Frequently Asked Questions

  • Article
Archived content. No warranty is made as to technical accuracy. Content may contain URLs that were valid when originally published, but now link to sites or pages that no longer exist.

Updated: September 1, 2002

The following are some commonly asked questions about a variety of Exchange 2000 Server issues. You'll find answers to questions on the following topics:

  • Directory Integration

  • Deployment

  • Storage

  • Backup and Restore

  • Instant Messaging

  • SMTP

  • Interoperability

  • Routing

  • Routing Service

On This Page

Directory Integration

Directory Integration

What is the difference between a primary and a non-primary connection agreement?

A primary connection agreement replicates existing directory objects. It also creates and replicates new directory objects in the destination directory. A non-primary connection agreement only replicates information in pre-existing objects.

A connection agreement type has two check boxes selected by default, even if a connection agreement already exists. These are "This is a primary connection agreement for the connected Exchange organization" and "This is a primary connection agreement for the connected Windows domain."

If you are using more than one connection agreement to replicate Microsoft Windows® 2000 user accounts for a single Exchange Server 5.5 organization, there should be only one primary connection agreement. Using multiple primary connection agreements to replicate the same Exchange Server 5.5 organization will result in creating duplicate objects.

What is the name-matching rule, and how do I set it?

You can customize directory object–matching rules on the From Exchange tab and the From Windows tab. The name-matching rule should be set to its default setting. You should change this only when the Microsoft Active Directory™ and the Exchange Server 5.5 directory have several common objects, for example, when inter-forest replication is in place. Matching rules should be changed so that object attributes in each of the directories have different values, for example, a Simple Mail Transfer Protocol (SMTP) address or a security identifier (SID).

Note: Attributes you select affect all connection agreements. If you clear the attributes for Exchange Server 5.5, you clear the same attributes for Windows 2000.

What is the function of the Site Replication Service?

The Site Replication Service (SRS) was designed to provide directory interoperability between Exchange Server 5.5 and Exchange 2000 Server. SRS runs on Exchange 2000 Server and serves as a modified Exchange Server 5.5 directory. SRS uses Lightweight Directory Access Protocol (LDAP) to communicate to both the Active Directory and the Exchange Server 5.5 directory. To an Exchange 5.5 server, the SRS looks like another Exchange Server 5.5 configuration/recipients replication partner.

Deployment

Where is additional information on setting up Exchange 2000 Server available?

Get an extensive tour of Exchange 2000 Server deployment in the Exchange 2000 in Six Steps article. This downloadable article provides useful tips gained from the Microsoft Early Adopter beta testing program. It offers a detailed view of an Exchange 2000 deployment in a mixed-mode Windows 2000 environment. Addressing specific deployment scenarios that can be used as a basis for your own deployment process, this article provides you with a clear picture of how deployment works, placing special emphasis on the most successful order in which to carry out deployment.

Storage

What are Storage Groups, and what is the relationship between them and multiple databases?

A Storage Group is a virtual container for multiple databases (information stores). The recommendation for database capacity planning on non-clustered Exchange 2000 Servers is a maximum of four storage groups that contain a maximum of five databases for each storage group. This results in a maximum of 20 databases for each Exchange 2000 server. This limit is enforced inside the Exchange System Manger snap-in, so administrators cannot create more storage groups than the system enables.

See XADM: Maximum Number of Storage Groups in Exchange 2000 for Database Capacity Planning 251123 for details.

Backup and Restore

Does Exchange 2000 RC2 support Single-Mailbox Restore?

Exchange 2000 does not support Single-Mailbox Restore with tools and products from Microsoft. You can find several third-party backup programs that support Single-Mailbox Restore, but the Ntbackup.exe tool cannot perform this function. Exchange 2000 does provide Mailbox Retention, a feature that enables you to retain a deleted mailbox for a specified period of time before permanently deleting it.

Instant Messaging

What is Instant Messaging?

Instant Messaging is a new and fundamentally unique medium of communication. This technology gives Exchange 2000 users the ability to communicate with other Instant Messaging users in an immediate, interactive environment that conveys "presence" and "status" information.

SMTP

How many recipients can be on an SMTP message?

The maximum number of recipients is 5,000 by default. When you send a message from one server to another with 5,000 recipients, you want that message body to be carried across the wire only once. The Windows 2000 SMTP server allows the administrator to specify the maximum number of recipients per message. The intention of having a low number is to make it harder for people to send junk mail to many recipients at once. The SMTP standard specifies that messages with more than 100 recipients should be broken into multiple messages.

Note: SMTP standards specify that servers must be able to handle at least 100 recipients.

Is there any authentication performed when one Exchange server talks to another through SMTP?

In Exchange 5.5, server-to-server communication is authenticated and encrypted using system-level Remote Procedure Call (RPC). With Exchange 2000, each server uses SMTP authentication with Kerberos. Encryption is not done by default. There are two options for encryption—Internet Protocol Security (IPsec), which is built into Windows 2000, and Transport Layer Security (TLS), built into the SMTP service and used by Exchange 2000. TLS is also known as secure sockets layer (SSL).

Isn't SMTP less secure than the X.400-based RPC that Exchange 5.5 had?

Many people think that SMTP is not secure because it has a clear-text submission protocol. Exchange 2000 does several things to increase the security of data over SMTP:

  • Server-to-server communication is always authenticated. The default state of each server will not accept unauthenticated SMTP traffic. Each message is checked to see that the From: field in the submitted message is really the person who authenticated.

  • With IPsec or TLS, encryption of data between servers is as good or better than the encrypted RPC of Exchange 5.5.

  • Much of the intra-organization server-to-server mail traffic is actually somewhat obscured. Messages that originated from MAPI clients or the Web client are a set of MAPI properties that need to be carried from server-to-server. MAPI properties are carried in a Transport-Neutral Encapsulation Format (TNEF) binary large object (BLOB). This is encoded using a publicly available, unencrypted format, but it is not readable. There will be no useful information available from a message in transit. Even if a tool is used to parse a BLOB, data could be decoded. However, it would be extremely difficult to easily look at mail traffic.

Is there any way to compress data with Exchange 2000 Server before sending it to another server?

At this time, SMTP servers do not have compression for mail. The specification for mail servers, however, includes a standard for implementing compression. The TLS extension provides message security through both compression and encryption. Encryption is usually more secure if the data is not plain text, and to make compression unpredictable, you should compress before encryption. Exchange Server supports the TLS extension.

Our transport events technology also makes it very easy for Microsoft or a third-party software vendor to release an extension to Exchange 2000 that would automatically compress and decompress messages as they come into or go out of Exchange. In most cases, compression overhead taxes the CPU of the Exchange server. This reduces performance, often offsetting any network bandwidth you gained through compression. Thus, it's probably better to build more functionality into the client, where you may have idle CPU cycles to spare, than to tax the server with compression.

Note: Current TLS implementations do not use any compression algorithms.

Can Exchange 2000 run on top of a different Microsoft Windows SMTP Server from Microsoft Internet Information Server (IIS) 5.0?

No, Exchange 2000 requires and works in concert with the server events extensibility that are built into the SMTP server. This ships as part of IIS 5.0 in Windows 2000.

How is the host name of an internal or external server resolved?

The SMTP Service takes a name, call it "REMOTE," which might be a server's internal fully qualified domain name (FQDN) or an external FQDN of an e-mail domain. For example, user@domain.com looks up "domain.com" and resolves it. The following steps should be taken to accomplish this:

  • Check the domain name system (DNS) for the mail exchanger (MX) record for REMOTE.

    • If DNS returns >0 entries, connect to port 25 on each one, in lowest priority order first.

    • If DNS returns "Authoritative Host Not Found [1]," non-delivery report (NDR) the message immediately. This is returned if the name server accesses the root (.) node of DNS and does not find a record for the domain name.

  • If DNS returns any other error, or returns no MX entries, then fall through to step 2 and call gethostbyname() for REMOTE. This results in both an A record search as well as WINS lookup.

Note: By default, Windows 2000 DNS ships with the IP addresses of the InterNIC root name servers pre-populated in its configuration. This means that a request for a domain that is not defined in a zone on the DNS server will be forwarded to one of those servers. If your server is behind a firewall and cannot reach these servers, you will not get "Authoritative Host Not Found," but rather "Server Failed."

Interoperability

What is the definition of site, administrative group, and routing group in a mixed organization?

An Exchange site is a server grouping for both administrative and topological purposes. In a mixed organization, the servers running Exchange 5.5 recognize sites, while the servers running Exchange 2000 recognize both administrative and routing groups. The Active Directory Connector automatically replicates each Exchange 5.5 site to Exchange 2000 as both an administrative group with a routing group of the same name.

How do legacy servers interact with Exchange 2000 connectors?

Within a pure Exchange 5.5 site, one server is designated as the routing calculation server that is responsible for keeping the gateway address resolution table (GWART) up to date and consistent across all servers within the site. When an Exchange 2000 server is installed into an Exchange 5.5 site, the original Exchange 5.5 server resumes it routing role. Although Exchange 2000 uses a very different routing mechanism to Exchange 5.5, it will create an Exchange 5.5–compatible GWART that will be replicated into the Exchange 5.5 environment through the Active Directory Connector. This information will then be merged with the GWART that the routing calculation server generates. The net result is that users on Exchange 5.5 servers will be able to take advantage of connectors installed on Exchange 2000 servers.

How does an Exchange 5.5 site relate to an Exchange 2000 administrative group?

In a mixed or native Exchange 2000/Exchange 5.5 topology, these are mapped 1:1. The administrative group is mainly for permissions mapping, although the administrative group is used to create the legacy-distinguished name (DN).

How do messages get from an Exchange 2000 server to an Exchange 5.5 server in the same site/routing group?

An Exchange 2000 server evaluates whether the server is in the same routing group or not. If it is, then the server sends the message through the Message Transfer Agent (MTA), which creates a direct local area network (LAN), MTA, RPC connection. If it is not, the server routes the message to the routing group of the destination server through connectors.

How do messages get from an Exchange 2000 server to another Exchange 2000 server in a mixed routing group?

Exchange 2000 servers, whether in a mixed or pure routing group, always use SMTP to send messages from one server to another. The SMTP Service will open a direct connection to the destination server. However, Exchange 2000 servers will route based on routing groups, not administrative groups.

Routing

How does an Exchange Server 5.5 site relate to an Exchange 2000 routing group?

In a mixed Exchange 2000/Exchange 5.5 topology, a site is represented as an administrative group and a routing group. An Exchange 2000 routing group, contained within the administrative group, is more analogous to a sub-site than to a site because it is for purely topological groupings. However, it is advisable that servers belonging to sites in Exchange 5.5 also be applied in this case. Members belonging to the same routing group are expected to have very high bandwidth and availability.

How does an Exchange 5.5 site relate to an Exchange 2000 administrative group?

In a mixed Exchange 2000/Exchange 5.5 topology, these are mapped 1:1. The administrative group is mainly for permissions mapping, although in a mixed-mode of Exchange 2000, the administrative group is used to create the legacy distinguished name (DN). Therefore, it is not as easy to move servers between administrative groups.

How does a Windows 2000 site relate to an Exchange 2000 organization?

There is no relationship. A Windows 2000 site is defined as a group of resources (computers, servers, etc.) that have high-connectivity to one another. An Exchange organization encompasses the entire forest and bears no relationship to the topological site structure that the Active Directory administrator defines.

How does a Windows 2000 domain relate to an Exchange 2000 organization?

There is no relationship. All configuration information for Exchange 2000 is stored in the Active Directory configuration naming context. This is replicated to every domain controller to each domain in the forest. Therefore, Exchange Organization information is available for read/write in every domain.

How does a Windows 2000 site relate to an Exchange 2000 routing group?

An Exchange routing group is a collection of Exchange 2000 servers with high-availability to one another, but not necessarily high bandwidth. Although the concept of the Windows 2000 site and the Exchange routing group are quite similar, there are no alignment prerequisites for deployment. Routing groups are defined in the configuration naming context of the Active Directory.

How does a Windows 2000 domain relate to an Exchange 2000 routing group?

There is no relationship. An Active Directory domain contains users and computer information for those that reside in that domain. An Exchange routing group contains information about Exchange 2000 servers that have high-availability to one another.

How does a Windows 2000 forest relate to an Exchange organization?

In Exchange 2000, there is a limitation of exactly one Exchange organization per Windows 2000 Active Directory forest. Conversely, every server within a given Exchange organization must be in the same Active Directory forest.

Routing Service

What is the purpose of a routing group?

The routing group is the smallest unit of servers likely to be connected to one another at all times. The routing group is one node on the graph of connector paths with multiple possible connectors between routing groups.

Within a routing group, or before routing has been configured by the creation of a routing group, mail from one server to another goes point-to-point using SMTP.

If you wish to have direct point-to-point routing between a collection of Exchange 2000 servers, you can place them into the same routing group. In general, you design your routing group boundaries based upon connectivity and availability of the network. Between routing groups, you can define connectors that route messages between these routing group collections. It is common practice to use a routing group connector (RGC) to accomplish this.

What does it mean for a connector to go down?

If the source bridgehead cannot contact the destination bridgehead, then the system, by default, retries for 10 minutes. After 10 minutes, the bridgehead is marked unavailable. If there are other target bridgeheads on the connector, those are tried instead. Once all target bridgeheads on the connector are tagged as unavailable, then the whole connector is marked down and other routes are evaluated. If there are other available routes, message(s) are rerouted. If there are no other routes available, the message will sit in the local queue until the connector comes back up.

What does the routing service do when a local connector is down?

When the SMTP Service or X.400 Service notices that a connector is down, it notifies the routing service of this. The routing service marks the connection as down in its routing state graph.

How does the connector get designated as up again?

The SMTP Service creates a special connection that has zero messages, but tries the remote side of the connector according to the retry interval for the virtual server. When the connection succeeds, the service updates routing with the new information that the connector is back up.

What exactly does a routing master do?

The routing master coordinates changes to link state that are learned by servers within its routing group. When one single server coordinates changes, it is possible to treat a routing group as a single entity and to compute a least-cost path between routing groups. All servers in the routing group advertise and act upon the same information.

What happens when it goes down?

All servers in the routing group continue to operate on the same information that they had at the time they lost contact with the master. This cannot cause mail to loop, because all servers continue to operate on loop-free information.

When the master comes back up, it starts with all servers and connectors marked up. As it learns about down servers, it reconstructs the link state information and passes it around.

Does having a single routing master introduce a single point of failure?

No. Exchange 2000 may send mail to a server whose link is down, but mail will continue to flow, since Exchange will automatically switch to sub-optimal routing if a routing master fails. Exchange 2000 allows the administrator to manually change the routing master role from one server to another.

How do SMTP and X.400 servers communicate link state information within a routing group?

Each server communicates with the master through a TCP-based Link State Algorithm (LSA) protocol developed in the transport core development team. Each server, including the master, is on TCP listening port 691 and registered with Internet Assigned Numbers Authority (IANA) for this purpose. The master broadcasts changes only to all servers in its routing group.

How do servers (both SMTP and X.400) communicate link state information between routing groups?

When two servers communicate through SMTP, Exchange 2000 uses a version of LSA protocol that works as an extension to SMTP through the SMTP Service Extensions (ESMTP) framework. Exchange 2000 servers advertise X-LINK2STATE support during the EHLO. When one Exchange 2000 server sees another advertising that, it attempts to trade routing information. Routing information will only be traded if the two servers are in the same organization (a DIGEST string is compared). This only occurs in the event of per-routing-group differences in transferred information.

Between routing groups, when servers communicate through X.400, Exchange 2000 uses a version of LSA. The MTA constructs a "dummy" X.400 message to transfer this information.

How often do servers that connect between routing groups communicate link state updates? Are messages used?

In the case of link state updates tunneled through SMTP, messages are not used. Instead, when there is an update, a connection is created to the neighboring routing group. During the course of that connection, the link state information is transferred. In fact, even if there is no new information on the source side, during each SMTP transmission between two Exchange 2000 servers in the same organization, they will exchange link state information.

In the case of link state updates through X.400 between two Exchange 2000 servers, a "dummy message" is created that includes the link state update information.

Why have all of this routing?

Network routers use the Open Shortest Path First (OSPF) protocol to route packets optimally between servers. The single-source, shortest-path algorithm, used by the Exchange routing service, is very similar to the OSPF internal routing protocol used by many enterprise networks, except that Exchange provides more information than simply IP source and destination. Exchange can route messages according to destination, message size sender, and message priority.

Note: The similarity between OSPF, and the routing algorithm used by Exchange 2000, is that they are both derived from Dijkstra's algorithm. Using the same type of algorithms is where this similarity ends. You do not have to deploy OSPF before deploying Exchange 2000.

Another reason to route messages through logical connectors is to optimize message bandwidth. If a single message is destined for recipients on five different servers in a remote location, point-to-point communication causes the message body to be sent five times. By funneling that through a messaging bridgehead, the message body is only sent once, which makes a significant difference with large messages.

Note: Certain connectors may be limited as to what size messages they will take. This is not referring to the IP address of the sender, but rather the actual e-mail address of the sender. Certain connectors may be limited by who may use them.