In our previous post What Happened to High Availability in Oracle 19c Standard Edition?”, we talked about how the high availability option in Oracle Database Standard Edition (SE HA) isn’t all it’s cracked up to be. On the surface, it looks functional, but when you look closely, it has more holes than a slice of swiss cheese.

Since you’ll likely want to re-evaluate your high availability strategy when you upgrade to 19c, you might want to consider whether a solution that adds horizontal scaling will help extend your usage of SE. SE has always been limited to 4 CPU sockets, and even when Oracle let you use RAC with SE (pre-19c), you were still limited to 4 CPU sockets across the RAC instance.

Because a data replication solution keeps two (or more) separate instances in sync, you can actually scale horizontally in the following ways:

  • Use an HA/DR instance for real-time reporting to remove workload from the primary server
  • Create active/active architectures that allow you as many read/write instances as you’d like to build capacity or place instances closer to your users

While Oracle SE HA will meet most basic high availability needs, you may want to consider adding SharePlex® to address all the challenges that come with the SE HA solution. As we discussed in the previous post , these include:

  • The need for a single data-center
  • No disaster recovery if the data center itself is damaged
  • Single point-of-failure on the database
  • Slow failover time
  • Increased and longer outage times for rolling upgrades
  • No access to a secondary server for performance enhancement

HA and DR together at last

With SharePlex, there are two (or more) databases, a source, and one or more targets. Since the transfer of data between the source and target(s) happens over a network, your target database can be in the same data center, across town or in a data center halfway around the world. The target can serve as both a high availability solution, and, since it can be located far away, a disaster recovery solution as well.

No single point of failure

Since the source and target are separate databases, there's no longer a single point of failure. Also, since the target database is open, you always know that your data is available when you need it. SharePlex includes robust compare and repair functions, so you can prove that the source and target databases contain the same data, without having to perform an actual failover and take down your business-critical application.

Scaling Standard Edition

Since SharePlex keeps two separate databases in sync, you can transfer reporting workload to the secondary database. If you really want to scale up Standard Edition, you can use the Active/Active capabilities to scale out horizontally, placing as many read-write instances as you need to scale up and/or put instances closer to different user populations.

Heterogeneous systems? No problem

Since SharePlex performs replication at the database level, it fully supports replication between heterogeneous systems. SharePlex can replicate from Oracle 11 to Oracle 19, or from a Solaris system to a Linux system, or even from an Oracle SE database to an Enterprise Edition database.

No downtime for upgrades and migrations

With SharePlex, customers are able to upgrade or migrate instances with virtually no-downtime. Real-time replication means users can be seamlessly switched to a secondary server while the primary undergoes maintenance. Once the upgrade is complete, SharePlex synchronizes the two servers, and the workload can be switched back to the primary without users even noticing.

Simple setup and management

Above all, SharePlex is simple to install, configure and manage. The same set of binaries are installed on the source and target servers and SharePlex uses one configuration file to manage all replication functions. There's a single command line interface and a no-charge add-on component called SharePlex Manager that gives you a graphical view of your replication environment and sophisticated alerting and alarming if something goes wrong.

Request a Trial

Related Content