When
transferring data between Microsoft SQL Server Compact 3.5 clients and
SQL Server, the data passes through Microsoft Internet Information
Services (IIS). For your network, you might configure a single server to
host both IIS and SQL Server, or you might have multiple servers to
enhance performance and security and multiple firewalls to additionally
enhance security. The network topology you choose to implement is based
on your security and performance needs.
Although
there are many possible network topologies that you can use for
transferring data between SQL Server Compact 3.5 clients and SQL Server,
there are three common network topologies:
Single-Server Topology
In
this topology, IIS, the SQL Server Publisher, and the SQL Server
Distributor all reside on a single server. This server may or may not be
located behind a firewall. If it is, you only have to open port 80 to
Internet traffic.
Two-Server Topology
In
this topology, IIS is configured on one server, and the SQL Server
Publisher and the SQL Server Distributor are configured on a separate
server. The server running IIS may be separated from the rest of the
network (including the computer that is running SQL Server) by a
firewall, and may also be protected from the Internet by an additional
firewall. The firewall between the client and IIS must have port 80
open, while the firewall between IIS and SQL Server must have two ports
open: port 80 and the port assigned to the instance of SQL Server. This
second port is typically 1443.
Enterprise Topology
If
you are supporting very large numbers of SQL Server Compact 3.5
Subscribers that synchronize at the same time, you can partition the
work across multiple computers that are running IIS. This reduces data
transfer delays.
If
additional load balancing is required on the computer that is running
SQL Server, you can create a republishing hierarchy on multiple
computers. The top-level Publisher publishes data to SQL Server
Subscribers. They in turn republish the data to the SQL Server Compact
3.5 Subscribers. Load balancing is static in the sense that each SQL
Server Compact 3.5 Subscriber is assigned to one of the republishers and
always synchronizes with that republisher.
As
with a two-server topology, firewalls can be used between the clients
and IIS, and between the IIS servers and the republishers.
If
additional load balancing is required on the computer that is running
SQL Server, you can create a republishing hierarchy on multiple
computers. The top-level Publisher publishes data to SQL Server
Subscribers. They in turn republish the data to the SQL Server Compact
3.5 Subscribers. Load balancing is static in the sense that each SQL
Server Compact 3.5 Subscriber is assigned to one of the republishers and
always synchronizes with that republisher. | SQL Server Compact 3.5 does not support alternative synchronization partners. |
SQL
Server Compact 3.5 replication can be used on a Microsoft Windows
cluster that contains a farm of Web servers running IIS. However, the
SQL Server Compact 3.5 Server Agent must maintain session state across
client calls; therefore, you must use load-balancing hardware or
software that can maintain session state between the SQL Server Compact
3.5 Client Agent and the SQL Server Compact 3.5 Server Agent.
Load-balancing systems typically work as follows: Clients access the cluster using a common domain name with a single virtual IP address; the load-balancing hardware or software intercepts the incoming HTTP traffic and directs it to one of the servers in the cluster.
SQL Server Compact 3.5 can be used with a variety of load-balancing systems, for example Windows Network Load Balancing (NLB). Windows NLB is included as part of Windows 2003 Advanced Server and Windows 2000 Datacenter Server. NLB can be used with SQL Server Compact 3.5 because NLB supports client affinity. This is an option that ensures that the client is always directed to the same server running IIS within the cluster. Note that if a server or network failure occurs during a client session, a new logon is required to re-authenticate the client and reestablish session state. Also, adding a new server running IIS to the cluster redirects some client traffic to the new server. This can affect existing sessions.
For more information, see "Windows Clustering Technologies" in the technical resources for Microsoft Windows Server 2003 at the Microsoft Windows Server 2003 Web site.
Load-balancing systems typically work as follows: Clients access the cluster using a common domain name with a single virtual IP address; the load-balancing hardware or software intercepts the incoming HTTP traffic and directs it to one of the servers in the cluster.
SQL Server Compact 3.5 can be used with a variety of load-balancing systems, for example Windows Network Load Balancing (NLB). Windows NLB is included as part of Windows 2003 Advanced Server and Windows 2000 Datacenter Server. NLB can be used with SQL Server Compact 3.5 because NLB supports client affinity. This is an option that ensures that the client is always directed to the same server running IIS within the cluster. Note that if a server or network failure occurs during a client session, a new logon is required to re-authenticate the client and reestablish session state. Also, adding a new server running IIS to the cluster redirects some client traffic to the new server. This can affect existing sessions.
For more information, see "Windows Clustering Technologies" in the technical resources for Microsoft Windows Server 2003 at the Microsoft Windows Server 2003 Web site.
No comments:
Post a Comment