ScaleArc

Replication Awareness

Ensure data consistency by monitoring replication lag and avoiding out-of-sync servers.

SQL databases are highly stateful, and applications that use them are typically very sensitive to data consistency. Since data replication is critical to any highly scalable, resilient database stack, ScaleArc has delivered a best-in-class replication monitoring and management framework that provides deep intelligence into the state of the replication frameworks behind your databases.

ScaleArc routes traffic based on your application’s data consistency requirements. If your application requires highly consistent data, ScaleArc lets you enforce traffic policies that will ensure data is sourced only from synchronous nodes. You can configure your own values for how much replication lag is tolerable to your application - all the way down to just one second. If a server falls behind the configured replication lag threshold, ScaleArc can take that server out of the load balancing pool, allowing it to catch back up and come back into service or marking it down if replication is broken and will not recover.

Replication awareness is also extremely critical to ensuring a safe and stateful failover between your SQL database servers. ScaleArc can use its replication intelligence to ensure the server we will be failing over to is in good health, and we can also ensure that the server is in sync with the last known state of the failed server before starting to use it for writes.

ScaleArc’s replication monitoring engine enables the following capabilities:

  • Replication type and role detection
    • Detect whether a database is using synchronous or asynchronous replication
    • Detect whether the server is a primary / master or a secondary / slave
  • Replication-aware load balancing
    • Reduce or completely stop load to a server based on replication lag threshold
    • Alert and mark a server’s health as down when replication is broken
  • Replication-aware failover
    • ScaleArc can ensure the server being failed to is in sync and has no pending transactions to apply in its replay log before sending write traffic to it

ScaleArc supports replication monitoring for the following databases and replication types:

  • Microsoft SQL Server 2005, 2008, 2008/R2, 2012, 2014
    • Transactional replication
    • Mirroring
    • Merge replication
    • Peer-to-peer replication
    • SQL AlwaysOn synchronous and asynchronous replication
    • GoldenGate replication
  • MySQL 4.x, 5.1-5.5, 5.6, and above
    • MySQL’s built-in transactional / binlog replication
    • GTID-based replication
    • Galera / Percona XtraDB Cluster synchronous replication
    • Multi-master or master/slave topologies
  • Oracle 11g
    • Dataguard
    • GoldenGate