When it comes to storage, flash is undeniably a hero. But it is unlikely to save the day on its own. While it has an important role to play, flash is just one of a number of solutions required to address the storage pain points in the data centre. Just like The Avengers or The Fantastic Four, it needs to be part of a team to get the best results. In any case, whatever role flash does play in an organisation’s storage infrastructure, there’s an underlying problem (or villain) that needs to be tackled first.
The real issue stems from the shift from physical to virtualised workloads. While an organisation has gone virtual, its storage is still built on an architecture meant for a physical world. To gauge the extent of the trend, it’s worth noting that the percentage of virtualised workloads has grown from 2% to 75% in just ten years. This rapid adoption of virtualisation has created a disconnect in the data centre, with physical storage becoming the source of growing costs, bottlenecks and frustration.
Virtualisation brings storage pain
A recent survey of 1,000 data centre professionals found that the two most cited storage pain points were performance (50% of respondents) and manageability (41% of respondents).
That’s hardly surprising given that increasing numbers of virtual workloads are generating far more random I/O patterns that are bound to choke disk-centric storage. To try and improve performance, storage admins shuffle virtual machines from one storage LUN or volume to another but this presents them with all manner of manageability shortcomings along the way.
Buying time won’t solve the problem
One way to try and overcome performance pain is to use flash because it is low latency and can handle random I/O. It’s also a lot faster. A single commodity SSD (Solid State Drive) is 400 times quicker than a hard disk drive (HDD). To put that comparison in context, the speed of sound is "only" 250 times faster than walking!
But flash’s super speed can only buy admins time. It doesn’t have the powers required to deal with the root cause of storage pain or relieve any management burden, namely the disconnect between virtual workloads and physical-world storage. It only addresses the symptomatic pain.
Over time, data centre professionals are likely to add more virtualised workloads as they expand their footprint from virtualised desktops to servers to private cloud. To keep up with the pressure put on their infrastructure, they may need to buy more and more (high cost) flash. Not only is that bad for their budget, worse still, it won’t resolve the disconnect.
Matching storage to virtualised environments
The best way to solve the root cause of storage pain is to deploy storage specifically built for the world of virtualised workloads, in other words, storage that is VM-aware.
VM-aware storage (VAS) has none of the remnants of physical storage —no LUNs or volumes, striping or widths. It operates at the most granular level, allowing admins to take action on individual virtual machines.
Conventional storage groups virtual machines into LUN or volume ‘containers’ and applies policies at the container level — assigning an amount of performance to be shared by all the virtual machines inside. A rogue virtual machine in the container will waste performance that should be used by its neighbours.
VM-aware storage gives admins the x-ray visibility to see and act at the VM-level and assign a specific performance level to each individual virtual machine. That means they can give mission critical applications more performance while setting a cap on any rogue virtual machine. This enables admins to guarantee every workload will get the exact performance it needs.
Managing the pain away
Manageability is an important part of the equation. Most storage admins and/or virtualisation admins have to maintain a large spreadsheet to map all virtual machines to their respective LUN or volume. As the virtual machines are shuffled around, the spreadsheet must be meticulously maintained. VM-aware storage makes the spreadsheet obsolete with its x-ray like powers. Admins can login and see every individual virtual machine, drill in for full analytics or set a policy (replication, cloning, etc.) at the VM-level.
Getting the solution with the best value
A storage admin can only improve performance in the long term with VM level manageability and the ability to align performance levels to individual virtual machines. What is required is a combination of the brawn to handle large scale deployments with the brains that provide VM-level visibility, control, automation and analytics.
VM-aware storage relies on its software brain to assign tasks to flash and spinning disk storage based on the speed and strength required to handle a particular workload. Just like The Fantastic Four, The Avengers or any other super hero troupe, it’s the perfect combination of brains, brawn and agility.
With the arrival of VM-aware all-flash on the market, organisations have the choice of enhancing their hybrid storage strategy of mixing flash and spinning disk or adopting an all-flash approach where it is appropriate.
In other words, customers can decide which workloads benefit from an all-flash or hybrid-flash approach. This means VM-aware storage is smart enough to help balance workloads so organisations only buy storage when they need it; and only buy the type of storage they need.
Flash might not be able to rescue an organisation's data centre single-handedly, but it can play a prominent role in helping the cause of VM-aware storage to save the day.
About the author
Kieran Harty is the Chief Technology Officer and co-founder of Tintri. Prior to becoming CTO, Kieran served as CEO and Chairman of Tintri. Before founding Tintri, he was Executive Vice President of R&D at VMware for seven years, where he was responsible for all products. He led the delivery of the first and subsequent releases of ESX Server, Virtual Center and VMware’s desktop products. Before VMware, he was Vice President of R&D at Visigenic/Borland and Chief Scientist at TIBCO.