Replication simply means copying data from one primary storage location to another location. This post clarifies the differences in the synchronous vs asynchronous replication of data.
Data Replication in the Cloud
In the context of cloud computing, data replication is vital for businesses that want to ensure data redundancy. The last thing you need is your primary storage server for a business-critical application going offline, affecting end users. With effective data replication in place, the end-users will never notice if a primary storage system goes down.
It’s important for IT departments to note that data replication in the cloud can be instigated by two different parties:
- Cloud service providers. As part of a cloud provider’s service level agreement (SLA), each provider typically replicates data automatically at no cost to you; the business using their cloud services. But you also have no control over this replication.
- Users of cloud services. As a business user of cloud services, you can take action to replicate your data or other important resources if you want added protection beyond what your cloud provider does. Or, as you’ll soon see, if you want a different type of replication.
It’s worth noting that replication is a disaster recovery strategy. In the event of a disaster causing the failure of a primary system, replication provides failover to a secondary system.
The scope of replication does not extend to data corruption or deleting data. If there is a fundamental problem that affects a dataset, such as accidental deletion, the problem is simply replicated to your secondary systems.
From this perspective, data replication is a tool with an important role to play in your wider business continuity strategy and disaster recovery plan.
Synchronous vs Asynchronous Replication
If you want to take action to replicate your data and important cloud resources aside from what your cloud provider does, you need to know about the different types of replication. By understanding these ways to replicate data, you can choose the best method for your use cases.
Synchronous replication happens in real-time. As you write data to your primary storage, it gets replicated to other secondary locations simultaneously. With synchronous replication, data is written to multiple locations all at the same time.
Asynchronous replication is when data is not replicated in real-time. The data is written to the primary storage location, and there is a time lag between the initial write and the copying of this data from primary to secondary sources.
By definition, each of these types of data replication in the cloud comes with its own advantages and disadvantages.
Synchronous replication is useful precisely because it happens in real-time without any delay. There is no risk of data loss caused by the failure of the primary resource between the time of writing and the time of replicating.
The disadvantage of synchronous replication is that it is highly limited by distance. Any sort of network latency above 30-40 ms between two storage locations is not conducive to effective synchronous replication. As such, this type of replication is typically only suitable within a data center or between two locations close to each other geographically.
Asynchronous replication is not limited by geographic location. You don’t need pristine network connectivity and as much bandwidth to replicate asynchronously.
The problem with asynchronous replication is that if a disaster strikes the primary data resource during a period of heavy updates with lots of write operations, these updates will not be reflected when the replication occurs.
In Closing
The main question you need to answer as a business is, “Which type of replication is better”? The answer depends on the specific priorities and budget at your company’s disposal.
In general, you should choose asynchronous replication when you want a more cost-efficient solution with minimal network burden and long-distance replication.
Your business should use synchronous replication if you can’t tolerate any level of data loss and your secondary storage locations are in close proximity, such as in your own data center.
Lastly, always remember to monitor your cloud usage, making sure that data is actually being replicated how you want it to be.