When looking at the table below please remember that Named Connections are only available on TCP/IP protocol, and are not available on legacy operating systems.
| Configuration(s) | Capabilities | Benefits | Limitations |
|---|---|---|---|
|
Simple Client Only Networking Named Connection configuration |
Each connection is limited to databases on one server, but where available C/C++ programs can create multiple connections to transfer data between servers. The Resilience API is not used in these configurations. Applications are responsible for handling of, and recovery from, network errors. |
There are no significant differences in the operation of DP4 compared with standalone operation (other than unavoidable performance degradation), so no changes are required to applications originally designed for stand-alone operation unless access to multiple servers is now required. On Windows platforms connections to different machines are completely independent of one another, so a slow connection to one server will not slow down programs running against other servers. Named connections (also available to client only machines) allow a client to connect to any machine running as a DP4 server anywhere accessible to the machine via TCP/IP. Named Connection configuration supports "reverse" connection allowing most efficient scheme for two way transfer of data between machines. |
Connecting to multiple servers is not transparent to application - so any kind of cross server activity requires explicit code in an application. Currently No support for distributed transactions |
| Local Database Resilience Client | Each connection is limited to databases on the local machine and one other server, but where available C/C++ programs can create multiple connections to transfer data between servers, as for Simple Clent Only and Named Connection configurations. Network error handling is determined by the Resilience command tail supplied to the database manager, unless applications take responsibility for handling and recovering from network errors. Tables can be statically configure to be local or remote. |
Applications originally designed for stand-alone operation do not usually require modification. Using local/remote facility allows even some cross server activity to be transparent to application. Supports transactions distributed betweeen local and remote database. |
Databases on client machines are not accessible to other machines Connections to different machines are NOT completely independent of one another, so a slow connection to one server will slow down programs running against another server. Limited API support for controlling whether any particular table is accessed locally or remotely - so writing programs to synchronise data may be more complicated than with other configurations. |
| Multiple Resilience Configurations | A single connection can access databases on up to 99 machines, and update any or all of them in a single transaction. Programs can use the DP4 Resilience APIs to select which machines a particular database access uses. |
Offers most sophisticated scheme for controlling cross-server activity transparent to applications. Supports Distributed transactions. |
Programs can only access machines in a pre-determined list of 99 servers. Connections to different machines are NOT completely independent of one another, so a slow connection to one server will slow down programs running against another server. Multiple Resilience causes changes to the operation of DP4 which may require you to make significant changes to applications and databases designed for stand-alone operation. Incompatible with DP4 Thin Client |