Sql 2016 replication immediate updating subscription
Replication can be thought of in terms of the newspaper model - the publisher produces the newspapers and sends them to the distributor, who separates and sends them on to the subscribers (e.g.
newsagents, shops) where the newspaper is 'accessed' by purchasers.
As you can see, SALES_MASTER takes on the transactional load and replicates selected publication data to different databases located in different countries. There are two types of replication subscriptions - push subscriptions, and pull subscriptions.
Currently, Microsoft provides only one tool to monitor latency, and that is the ubiquitous 'Replication Monitor'.
Poor latency can be planned for and incorporated into an architectural plan - normally the acceptable latency will be a function of business requirements, for example in the model above, the natural latency between the UK and Japan may be 3 minutes.
The Japanese management team may require data that is a maximum of 1 hour old.
Latency can be caused by various factors, including geographical distance between nodes; network traffic or bandwidth constraints; transactional load on the publisher; and system resources.
Less common causes, but by no means rare, are problems such as stalled agents, failed jobs and unexpected locking / other resource conflicts.
The subscriber will not contact the publisher directly.