Make Your Next Storage Array An App

Storage has been confined to hardware appliance form factors for far too long.  Over the past two decades, innovation in the storage space has transitioned from proprietry hardware controllers and processors to proprietary software running on commodity X86 hardware.  The hardware driving backup appliances, NAS systems, iSCSI arrays, and object storage systems, are often quite similar in terms of processors and components, yet despite this fact I&O professionals are still used to purchasing single purpose systems which lock customers into a technology stack. 

Over the past few years, companies such as HP (StoreVirtual VSA), Nexenta, Sanbolic and Maxta have released software-only storage offerings to complete head to head with proprietrary hardware appliances, and have found some success with cost conscious enterprises and service providers.  The software-only storage revolution is now ready for primetime with startup offerings now reaching maturity and established players such as IBM, EMC and NetApp jumping into the market.

I&O professionals should consider software only storage since: 

The storage technology acquisition process is broken.  Any storage purchase you complete today will be bound to your datacenter for the next 3 to 5 years.  When business stakeholders and clients need storage resources for emerging use cases such as object storage and flash storage, these parties often do not have the luxury of time to wait for storage teams to complete RFPs and product evaluations.  With software-only storage access to new technology can be accelerated to meet the provisioning velocity needs of customers.

Software can provide elasticity.  While many storage vendors have created elastic sales models for arrays to allow for online expansion, at this point elasticity is one sided since there is no way to make an array resource disappear after it has been activated.  Early adopters of software-only storage in the service provider space have benefited from elastic OPEX pricing, which allows the providers to only pay for the storage resources they consume, to gain cloud-like elasticity. 

Commodity hardware pricing can lower costs.  Today even the high end systems sold by market share leaders such as NetApp and EMC, are difficult to distinguish from standard servers once you get past the fancy bezels and get under the hood.  Given this market dynamic, now is the right time to start treating storage technology more like an application and leverage commodity pricing of hardware components such as flash SSDs which are often less expensive than storage media sourced from an enterprise storage vendor.

Software-only storage will be making a big splash at the upcoming VMworld 2014 show, with several vendors demonstrating their wares.  For more information on software-only storage, see Forrester's market overview report on Software-only Storage.


How About RedHat / Inktank Ceph

Thanks for the article Henry. I wonder why would you not mention RedHat/Inktank Ceph as a block/object stor solution? It is a scale-out solution, rock solid, simple to maintain and can grow up to tens if not hundreds of petabytes. I think you should look into this solution, because in a sense it is a direct competitor to EMC's ViPR, with no expensive storage backend (control + data plane in one). Shared-nothing architecture, RAIN platform.



Hi Alex Thanks for your

Hi Alex
Thanks for your comments. The market overview report actually covers RedHat, and discusses Ceph and Red Hat Storage (Gluster). I've also wrote a blog entry on Red Hat Storage after the Inktank acquisition. You can read that post here.


How About SoftNAS Cloud

Great article and right on point on the trend away from legacy storage vendor arrays. Glad to see this issue being discussed more openly like this.

Similar to Alex's comment, customers are not only moving to software-defined storage apps in droves, the new pivot point is increasingly whether to get out of the hardware business altogether and move IT infrastructure to the cloud.

I understand why you did not mention SoftNAS Cloud, as we are one of the new kids on the block, relatively speaking, providing the enterprise-grade software NAS for both traditional data centers and cloud IaaS.

What we see is this 3 to 5 year decision, precipitated by either a new storage array purchase decision or maintenance on existing arrays coming up, which is forcing the decision...

Door #1 - Renew legacy storage array maintenance for 3 to 5 years and re-commit to our own data center

Door #2 - Move away from the legacy storage arrays and onto commodity x86 servers with software-defined storage and virtual storage apps

Door #3 - Bite the bullet and migrate mission-critical data and applications to IaaS in the cloud.

Hi Rick, It is going to take

Hi Rick,

It is going to take awhile before we see a transition away from hardware systems, but I believe this may happen faster than I initially anticipated.

Most cloud storage options today are object storage or block storage like Amazon's EBS. NAS (running on an IaaS cloud) is a great market opportunity which should be good news for SoftNAS and other players like Red Hat and NetApp.


Indeed, Henry. Thanks for the

Indeed, Henry. Thanks for the quick reply.

Hope to see you at VMworld!

Great post Henry and nice job

Great post Henry and nice job calling out the SW defined disruption so clearly. SW defined Storage as well Flash Storage will bring Storage into the next era. There has never been a more exciting time in storage for innovation and new capabilities

Clear and concise

I think managing this change into large organisations is going to be the biggest challenge: a lot of storage folks have built their career up out of managing monolithic storage arrays, and they wont give them up without a struggle... it's the mainframe/open systems thing all over again ;-)