Contents Index Sybase replication technologies Replication technology characteristics summary

Introducing SQL Anywhere Studio
   Replication Technologies
    Sybase replication technologies

Choosing a replication technology


Each Sybase replication technology lends itself to particular applications. The following descriptions differentiate the technologies and let you select the one best suited to your needs.

You should consider which of the following elements are important in your application

Your consolidated database system 

In a typical replication environment, a large database serves as a central repository for information. Sometimes you can choose a database system that suits your needs. Other times, a central database already exists and you must adapt the replication system to work with it.

MobiLink can work with many popular database servers, including Adaptive Server Anywhere, Sybase Adaptive Server Enterprise, Oracle, Microsoft SQL Server, and IBM DB2.

In a SQL Remote installation, the central database must be either Adaptive Server Anywhere or Sybase Adaptive Server Enterprise.

Your remote database system 

Sybase replication technologies also differ in the types of remote databases that they can support. MobiLink allows your remote database to be either Adaptive Server Anywhere or UltraLite.

SQL Remote supports only Adaptive Server Anywhere remote databases.

Network characteristics 

MobiLink and SQL Remote are both well suited to occasionally-connected environments, where remote sites must operate for hours or days in isolation, although more frequent synchronization is possible whenever a network connection is available. In contrast, Replication Server is designed for a continuous connection to allow large amounts of data to be replicated promptly.

MobiLink is session based. A real-time connection is required during synchronization. If this connection is interrupted before synchronization is complete, the process will not complete until the next synchronization. In contrast, SQL Remote relays information via messages, which can be sent or received asynchronously. These messages may take the form of files on a hard disk, or e-mail messages. These messages can be processed whenever they are received, allowing replication to occur incrementally.

Latency 

In some situations, it may be important that your information is replicated immediately. In others, replication once or twice a day may suffice. In fact, more frequent replication may be impossible when no network connection is available.

Both MobiLink and SQL Remote are primarily intended for situations where replication occurs infrequently; for example every few hours or days. MobiLink and SQL Remote can both handle more frequent synchronization, but resource and network requirements are greater. However, given sufficient resources, MobiLink synchronizations can occur every few minutes. SQL Remote, when run in continuous mode, allows replication to occur every few seconds.

Replication Server is designed for setups requiring near real-time replication.

The number of remote sites 

If you have a very large number of remote users, the best options are MobiLink or SQL Remote. The SQL Remote message-based design allows a typical installation to handle thousands of remote users. MobiLink scalability is limited only by the scalability of the consolidated database-management system. Replication Server is designed for only a few sites.

While these numbers are guidelines, there is no hard limit on the maximum number of remote sites with any of these systems. The actual number depends on the amount of information replicated, the frequency of synchronization, and the design of your application.

Transaction ordering 

SQL Remote replicates data by scanning the transaction log and preparing messages, as appropriate, for each transaction. It orders these messages and sends them to the remote or consolidated site. When processing receives messages, SQL Remote always processes them in the same order as they were applied to the other database. When necessary, it automatically delays processing a message until all earlier messages have been applied.

MobiLink, in contrast, works by grouping the results of multiple transactions on the remote database into one set of changes to be applied to the consolidated database. Since synchronization always occurs at a transaction boundary, referential integrity is preserved. The order of the individual changes made during the component transactions is not preserved. However, since uncommitted data is never synchronized, data integrity is preserved.

Achieving data consistency at a particular time 

Immediately following each MobiLink synchronization session, the data in the two databases is consistent. The ability to guarantee the consistency of the data at a remote site at a particular point in time is an advantage of MobiLink session-based replication. For example, if it is important that the data at a remote site accurately reflect the data in the consolidated database at a particular time, say 10 o'clock in the morning, this objective can be achieved by synchronizing just prior to this time. As long as the synchronization completes successfully, the currency of the data at the remote site is assured.

When changes to the data are replicated through an exchange of messages, it is difficult to guarantee that the data in a particular remote site is completely consistent with the data in the consolidated site at any particular point in time. For example, sometimes a message is lost in transit. SQL Remote automatically recognizes this fault and resends the message, but such interruptions can cause unexpected delays.


Replication technology characteristics summary
MobiLink characteristics
SQL Remote characteristics
Replication Server characteristics

Contents Index Sybase replication technologies Replication technology characteristics summary