In its work to make the web simpler and more accessible, Amazon launched its Amazon Elastic File System (Amazon EFS) service in 2016, after announcing it a year earlier, in early 2015.
The company has always explained that EFS was launched to make it easier to build applications that require shared access to file data. EFS is highly durable and scalable, and offers a robust model of read-after-write consistency, they ensure.
Since the 2016 launch, Amazon has been adding new features and capabilities, including encryption of data at rest and in transit, an infrequent access storage class, and several other lower cost storage classes.
And, most importantly, they have worked to improve performance: in six years they have provided a 400% increase in read operations per second, a 100% increase in performance per client, and then an additional threefold of read performance.
Now, in the middle of 2022, Amazon claims to have managed to travel at the speed of light. Or, at least, they’re close, as latency-sensitive EFS workloads can now run twice as fast as before.
Until today, EFS latency for read operations used to be in the single digits of milliseconds. As of today, new and existing EFS file systems offer an average latency of just 600 microseconds for most operations.
Over the past few weeks, Amazon has enabled the option to improve performance for all EFS file systems at a general level, so it is possible that customers have already noticed the improvement.
The point is that to take advantage of these speeds, Amazon explains that you have to be close, very close to the server . So, for now, these spatial velocities will remain for clients that are close to the servers, as is usual for these services.