There has been a lot interest lately in data replication. Unfortunately, the flurry of words has caused quite a storm of confusion. And at the eye of the storm? Synchronous or asynchronous replication? That is the question.
For anyone considering synchronous data replication, the most important factor to understand is that this process can require a tremendous amount of bandwidth, depending on the amount of data involved. This is because every write to the disk subsystem is replicated immediately.
Aside from the beaucoup bandwidth needed, the process can be costly. Even if your synchronous replication is local, it still requires high-performance disks and fast local network connections because the process is only as fast as its slowest element. In addition, all hardware and networks needed in replication must be available always. Mirroring of data is the same in basic principle except it can be achieved within the same device or with two devices.
Given the extraordinary bandwidth and cost involved, why would anyone want synchronous replication? In one word: redundancy. In case you were to lose one system, business can keep on going uninterrupted. It’s the same reason businesses favor the security of clustered servers. If you have mission-critical databases and you can afford it, synchronous replication delivers reassuring peace of mind.
In contrast, asynchronous replication is much more common and affordable. Async basically makes the most sense for backing up data as part of a disaster recovery application. It provides outstanding fault tolerance for network storage by capturing changes to data at the byte level. The largest benefit of asynchronous data replication is if there’s an outage, replication can continue and all data in the replication system is secure.
What factors are driving the requirements for asynchronous replication? A primary reason is the growing number of organizations with multiple sites that want to consolidate their backups and/or achieve heightened disaster recovery capabilities.
Asynchronous replication has evolved as DR planning has become a front-and-center issue. Thanks to some more recent technologies, such as bandwidth optimization and data deduplication, async replication now is more affordable and accessible than ever. In particular, bandwidth optimization can ensure you have high-capacity connections to transmit replicated data while data deduplication can lower the time and cost of backing up data to disk.
So if you’re looking for ways to bolster your current disaster recovery plan, consider asynchronous replication as a way to ensure that all your sites—not just your data center—have the highest levels of protection possible.
___
Jeff Graham
Senior Product Manager