When you have multiple master databases it possible to achieve faster access to consolidated data, better uptime performance for round the clock updates, and read scalability all using commodity hardware. Say goodbye to scheduled batch jobs that temporarily synchronize data across locations from a central master. Continuous asynchronous master to master replication means one region's data is available to all regions in near real time. If a master in one region fails, users won't have long delays to enter / update data until a replacement is brought online. You can re-route the downed region's database requests to another region's database. When the replacement master re-joins the cluster, it will re-synchronize with the other masters automatically. If conflicting edits occur on the same row during replication (i.e. a uniqueness, update or delete conflict), you can choose from Earliest Timestamp, Latest Timestamp, Node Priority or manual options for determining a resolution. You can even have a standby resolution should your first choice not resolve the conflict. In addition to each of your local regions or users having faster access to integrated data, your systems will be able to handle more reads and queries. Each master will be available to handle queries from any region or any query needing consolidated data. To scale further you can build out Hot Standby replicas to each master that handle even more read loads providing you a very flexible and scalable architecture for the future.
Comments not found