Virtualization and Cloud executives share their predictions for 2013.  Read them in this VMblog.com series exclusive.

Contributed article by Gridstore, co-founder and CEO, Kelly Murphy

Five Predictions for 2013

Flash crash - Too much supply, not enough demand

No, it's not the Wall Street fiasco of 2010. Flash has been the most over hyped technology in 2012. There's a ton of vendors in the market today pushing their flash hardware arrays - while consuming tons of investors' cash to stay afloat. 

But many flash vendors are quickly going to see themselves into obscurity, as the market consolidates. The reality is that few of today's enterprise workloads require this much performance, and if you really need it, you are not likely going to buy from one of the 50+ startups selling flash storage - you will likely buy from the large established vendors (EMC, IBM etc.) who you know will be around for a few years and who recently acquired flash storage startups to add to their portfolios. 

2013 will be the year of the flash crash.  A rapid consolidation in the flash storage array market after the 3 or 4 remaining larger storage vendors acquire a flash storage startup that suits them while the rest will simply crash in an over supplied market with relatively low demand for very expensive storage. 

The end of big storage is now

As we see players like Facebook, Google and Amazon no longer buying storage from the big vendors, we will begin to see a mass exodus. Traditional businesses, both small and large, will follow suit. Enterprises and SMBs alike should no longer have to buy monolithic storage built on 30 year old technology. Today's workloads do not fit these old architectures and suffer from scalability and performance as a result of trying to retrofit them. 

New architectures have emerged that move away from the limits placed on capacity and bandwidth. Today, businesses have the choice of performance, capacity and affordability all in one solution.  There's no need for overhead or complexity. No need for forklift upgrades. The days of big storage are over.

Converged Compute and Storage:  No one wants to grow servers at 60% per annum!

There has been much buzz around hardware appliances that tie compute and storage together. This concept has gained traction as a result of its simplicity, ease of use, and small footprint. While these benefits are great, the whole concept just doesn't make sense.

Since storage separated from servers in the mid-80's, storage has grown 6X faster than compute has.  When you physically put the two back together again - you will ultimately end up growing your server infrastructure at the same rate as storage - 60% CAGR.

With the exception of a handful of workloads - this makes absolutely no sense. The rise of the virtualized data center has been driven by flexibility and higher utilization of IT assets that has resulted in server consolidations of 10:1 and 20:1.  Physically converging compute and storage is the exact opposite - taking us back decades to where data center server sprawl gave rise to server virtualization.

This is just the wrong direction. In 2013, you will see software defined storage take center stage in a way that virtually converges compute and storage to deliver the benefits of simplicity, performance, ease of use - and - the ability to scale storage separately from servers to drive the maximum utilization and flexibility of the IT fabric. 

Finally RAID is dead.

This has been going on for some time now.  And it's because it really isn't sufficient any longer. Today, there are a number of challenges that change the way data protection and fault tolerance need to be delivered.  From natural disasters to simple manual misconfigurations, RAID doesn't cut it anymore. 

Microsoft, Intel, Gridstore and others are leveraging erasure coding, fault tolerance far superior to that of RAID.  The key today is to design for failure.  When you expect failure, you design resilient systems that have no single point of failure, built to withstand data loss and/or corruption.  It can protect from not only disk failures, but also network link failures and even entire storage node failures.  In today's dynamic and highly scalable data centers, this is a new mandate.

Clustered Storage will be replaced by Software Defined Storage

The reality is that today, clustered scale-out storage architectures are not designed for workloads like virtualization. Clusters add a lot of cost, complexity and wasted resources. Clustered Scale-out requires high performance backplane networks, cluster managers, and 3 way replicas - they literally take what you already have too much of (data) and multiply it by 3X. And while multiplying all this data, the inter-node communication cuts the IOPS available in half or more. You wind up with half the throughput you expect while having to power, cool, house and manage three times the amount of data you started with. When you deploy this kind of storage for your virtualization - the I/O blender problem of virtualization is multiplied by 3 and its performance literally grinds to a halt. 

In 2013, you will see Software Defined Storage emerge as a scalable, high performance storage architecture that is specifically designed for the virtualized data center. Software Defined Storage puts virtual storage controllers inside the VM/host to deliver high performance and low latency IOPS from host flash devices while being able to scale pools of storage separately and utilize these pools to dial in just the right mix of capacity, performance, utilization specific to each workload requirement. Instead of one-size fits all monolithic storage arrays (sometimes wrapped in 1970's clustering) - storage for the virtual data center will become completely virtualized and optimized to the unique requirements of each individual workload. The result is the best possible price/performance combined with flexibility and simplicity that is not possible with today's monolithic storage.

About the Author

Gridstore, cofounder and CEO, Kelly Murphy

Kelly Murphy is a serial entrepreneur with a track record of spotting and developing emerging technology markets. In 1998, Murphy founded Marrakech, the first on-demand software company that offered on-demand procurement and supply chain systems to some of the world's largest retailers, consumer food producers, packaging companies and utilities - as well as the thousands of trading partners they collaborated with.