An insight into the benefits, but also in the complexity of implementing advanced replication.
In addition, some general rules for the selection of appropriate replication methods to maximize the performance, taking into account the requirements for remote access.
Oracle replication has been around for a while and has become a mature environment with a large function scale, which all the requirements of a distributed computing fulfilled.
Replication was originally thought of as a way to Oracle tables or subsets of tables in a highly distributed database servers locally available. This was achieved by using snapshots (ie, at a certain time copies) of the desired tables from a master server to one or multiple remote slave servers have been copied.
This snapshot method was particularly effective for relatively static tables that do not require frequent updates to the master tables remain synchronized. Applications with an exclusive reading to benefit from the use of snapshots as time-consuming transfers in WAN superfluous, which significantly improved the performance.
Snapshots are now more likely under the term Materialized views known, and the creation of remote Materialized Views of master tables is still a common use of replication, but since then, the technology has developed and supports a much wider range of database objects. What follows is the first snapshot method and then describes an overview of the methods are more advanced.
A whole series of factors for the decision tables in Oracle replication. Particularly important are the size of the table and change the frequency of the data table (Figure A). Small Static tables are ideal candidates for the snapshot replication for remote applications, the read-only access the data, whereas extensive dynamic master tables with many inserts, updates and deletions require frequent updates, the systems and network strongly incriminate oneself. Snapshots are available for such extensive dynamic master tables is not a good solution, why on more advanced procedures (as defined below) should fall back.
In view of the system and network performance must keep in mind what size the created snapshots and how often they are created or updated. As Figure A shows, you can always re-create a snapshot or a full upgrade. You can also automatically carry out periodic updates or certain triggers that will ensure that changes to a master table of the Slave snapshots acceptable. The following rules of thumb can help to make the most appropriate method.
|