Replication Terms Defined in the VVR Context
This section defines the following replication terms in the VVR context:
Write-Order Fidelity
To use the Secondary in a disaster recovery scenario, write-order fidelity must be maintained. The term write-order fidelity means that VVR tracks writes on the Primary in the order in which they are received and applies them on the Secondary in the same order. It is important to maintain write-order fidelity to ensure that the data on the Secondary is consistent with the data on the Primary. While the data at the Secondary can be behind in time, it must be a consistent image of the Primary RVG at a known point in the past.
Without write order fidelity, there is no guarantee that a Secondary has consistent, recoverable data. VVR maintains write-order fidelity regardless of the mode of replication and across all the data volumes in an RVG. For example, in a database environment, the log and data space are typically on different volumes. On the Primary, VVR applies writes to the log and data spaces in a fixed order and maintains this fixed order when applying the writes on the Secondary. If write-order fidelity is not maintained, a database application may not recover successfully when failed over to the Secondary.
Consistent Data Versus Current or Up-to-date Data
Data is consistent if the system or application using it can be successfully restarted to a known, usable state. The data on the Secondary is consistent if it correctly reflects the data on the Primary at some point in the past. At all times, VVR maintains the data at the Secondary in a consistent state with the data at the Primary. For example, if the data being replicated is used by a database, the data is consistent if the database can be started and recovered to a usable state with zero data corruption. If the data contains a file system, the data is consistent if the file system check utility can be run and it can recover with no file system corruption.
Data is considered consistent only if it contains all updates up to a point in time and none of the updates that come after that point. For example, if it is a file system, the most recently created files may be missing when it is restarted. Or, if it is a database, one or more of the most recently committed transactions might be missing.
Data that is current or up-to-date contains the latest changes made at the Primary. For example, if you are replicating a database, the most recent transaction is available at the Secondary. Whether or not the data on the Secondary must always be current is a business decision and can be controlled by choosing between synchronous and asynchronous modes of replication. Synchronous mode guarantees that the data on the Secondary is current, at the cost of application performance; asynchronous mode does not guarantee the data is current, but has less impact on application performance and provides the ability to use more cost-effective telecommunications.
|
