Database Reference
In-Depth Information
Furthermore, the database mirroring session-which is running in synchronous mode-will
guarantee that the transaction log chain is unaffected by the mirroring failover and only
the valid log is restored all the time by managing the Log Sequence Number (LSN). Next,
the secondary server will continue to copy the log backups without the knowledge of any
changes that occurred due to the failover of a different server instance, which has become
the new primary server. So, if the principal server is lost for any reason after the database
is synchronized and the mirror server and witness servers continue to communicate, then
automatic failover occurs. This automatic failover causes the mirror server to assume the
principal role and brings the database online as the principal database to continue the log
shipping job schedules.
Improving Replication scalability at
multiple sites
In a distributed environment, whether it is intranet or internet based, it is crucial to keep the
data flow autonomous and available online when needed. The transaction is atomic and a
certain level of scalability is needed as the load against the database increases. In any case,
the application is dependent on more reads than writes, for such scenarios it is a compulsory
requirement to scale out the 'read' processes of the workload by caching read-only data across
the multiple database and instances where the web server connectivity is directed. If the
databases are geographically dispersed, then it is highly essential to scale out the data flow
between the sites to support read and write requests for scalability and improve availability.
In this recipe, we will go through the common requirements for replication, scalability, and
provide appropriate solutions with the type of replication to use.
Getting ready
The prerequisite for this process is to ensure the appropriate SQL Server version is used and
replication is deployed.
F
Further, the system should allow changes to be made at any server and have the
ability to perform changes replicated to all other servers
F
The replication model must consist of a minimum of one publisher and more than
two subscribers and distributors (locally or remotely)
F
The types of replication suited for this recipe are: snapshot, transactional, and merge
replication;
F
The source must be a publisher having the required data at the source that is
included in the publication
F
Each cache is a subscriber to the publication where the schema and data is received
as subscription
Search WWH ::
Custom Search