Replication protects against power outages and failures in processors, networks, servers, and databases, among other parts of an infrastructure where data may be housed. As such, synchronous replication—and other data duplication types—are part of what is called failover.
Failover is a mode in which a system or infrastructure can maintain all functionality in a backup system until all virtual or traditional components are able to come back online. Replication is essential to failover backup systems. As an application or software, it receives signals from a LAN or WAN that there has been an outage or other potential means of data loss, such as a fire, or severe weather system. It then kicks into gear, ensuring all the most current (and in some instances all iterations of) information, software development, and IT services are backed up. As a secondary system, replication that synchronizes is failure-tolerant—making it an absolutely critical system for use in large enterprise infrastructures, virtual storage, datacenters, and other platforms used by large-scale corporate, federal, and medical facilities.
Synchronous replication is not only used for disaster recovery, but also for applications that have low recovery time objectives (RTOs): a representation of how much downtime an application can afford without affecting enterprise operations too much. In other words, synchronous replication allows large systems to better protect themselves, reduce latency and data interruptions, and generally keep the enterprise working without any disruption.
Synchronous replication—or any replication for that matter—is not without its faults. While many IT professionals and developers labor under the misconception that replication will save all lost data through duplication, this is just not always the case. The main reason for this is that even though data, apps, software, and other information have been replicated synchronously (or otherwise), they may not necessarily have been commited to memory. This presents a huge problem with data loss, most especially for applications and software in development.
Some newer providers are now duplicating data and other information needed to keep development and IT operations (DevOps) up and running during outages by replicating individual VMs under the hypervisor. This newer method reduces the complicated setup of other replication systems, while writing (duplicating) all information in real time, rather than waiting for a disaster to occur. So, if any part of an infrastructure crashes, experiences an outage, or otherwise fails, the backups for everything are already (and have been) in place for immediate execution without interruption of any kind. The advent of this new type of replication does not necessarily mean you should stop using synchronous replication. What it does mean is that—as always—there are newer, faster, and more agile ways to create the same or better results. One thing IT professionals are recommending is the combination of existing replication systems with this newer VM-type duplication to cover every imaginable base.
When synchronizing data, the key difference between asynchronous and synchronous replication is how and where the replications are stored and maintained. With synchronous data replication, information is duplicated onto whatever the primary storage area within your WAN or LAN is. Conversely, asynchronous replication make copies of information (creates replicas) after it has been completed. This can be hazardous to developers and IT professionals, as it means a total loss of anything they have been working on that would have otherwise been replicated using a synchronous system.
But, there are some advantages of asynchronous replication. For small- to medium-sized ventures, it uses far less bandwidth. Additionally, it has the capacity to work nationally or even internationally, because replications are not happening in real time. Another upside to asynchrony is that it doesn't need bandwidth from a LAN to servers or HDDs within datacenters. Synchronous duplication of data can require extensive bandwidth, and it almost always requires additional hardware within your servers or datacenters.
For applications and software being developed or housed in cloud environments, real-time replication is crucial, and that's exactly what Tintri SyncVM brings you. With per-VM replication, you can expect RTOs of 30 seconds or less, and Tintri support can assist you with inter-data center connections as distant as 100 kilometers apart.
Tintri synchronous replication also offers synchronous and asynchronous replication within the same all-flash array simultaneously. This is available through the use of VMs as well as secondary replica VMs to ensure you never lose any iteration of any app, software, data, information, files, or anything else you have stored.
The best part? When you perform synchronous replication with Tintri, there's no need for you to purchase add-on hardware, converters, SAS switches, or new and specialized array models. A quick and simple configuration with single-click failover makes it fast and simple for your IT team to understand and maintain replication during catastrophes, or just as a day-to-day precaution for real-time backup—all on the VM level.
Check out how Tintri SyncVM works by viewing the product demo here.