Maximise the performance of your flash storage ‘race car’

Imagine having a Formula One race car at your disposal, but no race track. Just rocky, rutted dirt roads that slash the performance potential of your car to that of an average sedan.

Imagine having a Formula One race car at your disposal, but no race track. Just rocky, rutted dirt roads that slash the performance potential of your car to that of an average sedan.  

That’s how many storage administrators felt when decreasing prices and increasing densities made flash a viable alternative to disk storage. For the first time in the history of storage networking, the storage protocol stack was the performance bottleneck. This made storage protocol stacks suddenly relevant in a way that they had never been before.  

Enter Non-Volatile Memory Express (NVMe) – a specification that dramatically increased the speed at which data moved to and from a flash module. This protocol was built explicitly for flash and designed to improve data and latency sensitive applications.  

For uses in which the movement of data was solely between the server and storage module, NVMe took flash from “viable” to “compelling.” But for enterprises that move data through a network, NVMe was just a frustrating taste of how the unprecedented performance of flash might someday be realised.  

Accelerating performance  

Most businesses today need to perform ultra-fast data transfers across large-scale networks for business critical applications.  

Better performance/throughput and lower latency are two major benefits expected from flash. But without a balanced infrastructure of compute, storage and networking elements, the potential of flash technology is minimised. With aggressive deployment of flash across the industry to increase performance, this problem will only get worse for many companies.  

This has led to the latest big hype for scaling NVMe to the data centre – NVMe over Fabrics. Announced last year, this industry standard dramatically reduces latency and eliminates the need for SCSI translation, by directly transferring NVMe commands and structures across an existing network. This makes applications run faster or scale better.  

A primary advantage is the ability to add sustainable scale without affecting performance. By delivering low latency for all-flash arrays, NVMe over Fabrics is emerging as a promising solution for large-scale enterprises and SAN infrastructure.  

Owing to its nature, NVMe can run across a variety of fabrics, including Fibre Channel and Ethernet. Choosing the right transport protocol is a crucial strategic decision that will affect storage networking efficacy for years to come.  

The right race track

For critical storage networking, Fibre Channel is the outright NVMe fabric of choice. NVMe over Fibre Channel works with flash storage to ensure you get the performance and low latency that you originally wanted, but it also provides the additional reliability and performance of a Fibre Channel network.  

In this way, running NVMe natively across a Fibre Channel network extends the benefits of flash storage, maintains flash simplicity and efficiency, while eliminating the need for translation. That means you can enjoy higher application performance, more data storage, better analytics and more personalisation of information.  

For organisations building greenfield storage networks or those that aren’t running Fibre Channel throughout their infrastructure, these reasons make NVMe over Fibre Channel a convincing choice.  

In reality, most of the shared flash deployed today already uses Fibre Channel and SCSI. For these deployments, an evolution to NVMe over Fibre Channel provides even greater value. As an incremental change, the evolution comes with far less disruption, lower risk and less cost than transitioning to a different transport protocol.  

Selecting Fibre Channel as the transport protocol for NVMe over Fabrics is an important first step in realising the performance potential of flash in storage networks.  

Gaining future traction

By investing in hardware that supports NVMe today, you can ensure your network and storage are optimised for whatever’s coming next. This will enable you to support enterprise data centres, mobile computing, high-performance computing, relational databases and other uses of the technology.  

As applications continue to advance and virtualisation permeates the data centre, flash storage will grow right along with them. Understanding how much bandwidth response your applications require — and what application response time growth you foresee — is critical for your planning.  

Rethinking your current data centre environment to accommodate flash and the applications that require it, will most likely bring you to a discussion about NVMe and NVMe over Fabrics. Understanding how communication occurs among storage, compute and network resources will greatly affect your success.  

With the right platform in place, you can implement the best solution for your business, using NVMe as the next evolutionary step for your storage network.  

Phillip Coates is a systems engineering manager for Brocade in Australia and New Zealand. He has architected and installed solutions covering WAN connectivity, data centre and campus solutions, carrier based connectivity, various forms of network security, as well as fault diagnosis and analysis. He also has extensive experience in storage area networking using traditional fibre channel, as well as next generation solutions utilising converged Ethernet with iSCSI, FCoE or network attached storage.