As we go to press it looks like Nirvanix – a pioneer of cloud storage – has, according to widespread reports, closed its doors in what is a huge blow for customers and the very idea of cloud storage.
Nirvanix's customers were apparently given two weeks to move all their data out of its facilities. That’s not what you want to hear if you’re a customer. If you used Nirvanix for third or fourth duplicate copies you need assurance that data will be destroyed. If you used it for primary data you need that data back, and that is no trivial task right now.
The whole scenario is clearly also a big blow to the cloud storage model, since it apparently validates fears over the risks of handing your data over to a third party.
But taking a step back, the question is begged; why did Nirvanix fail, and what are the implications of this for other cloud storage specialists? After all, the company seemed to have a lot going for it. It had first mover advantage, and this brought it plenty of funding; around $70m in total.
It also brought in an extremely experienced management team and established a good-sized customer base including some large organisations in the media, financial, healthcare and education verticals, not to mention plenty of service providers.
Finally – and crucially for any startup in terms of third part validation– it won the support of some of the industry's heavyweights including a strategic partnership with IBM Global Services in October 2011.
It seems these apparent tailwinds ultimately failed to count. We believe the seeds of Nirvanix's failure were sown much earlier. Indeed, the company may have been doomed from the start. Why is this? We think there are three primary reasons.
A capex-centric model drained the coffers
One aspect of Nirvanix's approach that made it stand out from the crowd is that it was, in effect, a service provider, offering storage-as-a-service billed on a $/GB basis at a competitive price point. The idea was that this would appeal to customers because it simplified things – one bill, only one throat to choke if things go wrong etc.
But in reality this became the albatross around Nirvanix's neck because it had to construct its own physical infrastructure at no little expense; the company built out several storage nodes worldwide for storing customer data, and also offered the option for customers to deploy one of its nodes on-site – though still managed by Nirvanix – as a private cloud.
This model, we believe, was its undoing. It raised a good amount of funding, using this cash to rent floorspace in datacentre facilities and buy hard drives, enclosures, racks and other components to assemble multiple very large storage systems. But, this is not the way to offer differentiated value; it's table stakes stuff. And in a market where prices are being continually squeezed by much larger rivals – giants such as Amazon and Microsoft – Nirvanix was always going to be at a disadvantage to these rivals' enormous economies of scale.
This perhaps wouldn't be so bad were Nirvanix operating on a relatively small scale, but Nirvanix was a big-game hunter; as far as it was concerned, the more data the better. It cited several multi-petabyte deals, including an 8.5PB digital archive with the University of Southern California, though it said other deals were substantially larger.
Storing such data volumes requires substantial capital investments, and though the inner workings of Nirvanix’s technology were never fully discussed (see point 2, below), Nirvanix’s pay-per-month business model meant it had to bear much of the up-front cost itself. This is a very difficult model to sustain at significant scale, especially for a startup.
Where was the real IP?
Allied to this was a nagging suspicion that Nirvanix had a relatively small base of software intellectual property. Though it waxed lyrical about its Cloud File System the company was reluctant to share details of its underlying storage architecture, and whether there was any significant IP here.
The success of the business would have depended in large part on its ability to scale effectively and efficiently; if it didn’t have much in the way of real smarts here then this would hamstring its efforts to remain cost competitive. It also relied fairly heavily on partners for crucial aspects of its service; for example, it partnered frequently with cloud storage gateway vendors such as Panzura to on-ramp data onto its cloud. Though it had some of its own capabilities here (ie, CloudNAS) these didn't scale to meet the requirements of many customers.
This is not to say that Nirvanix wasn't innovative. It pioneered the notion of storage capacity as a service, but without serious software smarts to optimise the infrastructure this was never going to be enough.
Limited appeal of storage-only services
Storage doesn't exist in a vacuum, and while there are undoubtedly plenty of storage-specific services, such as those mentioned above, in reality this is still a small market overall. Moreover, while storage-as-a-service for things like backup and disaster recovery is indeed booming, adoption is overwhelmingly among small and mid-sized businesses; Nirvanix, on the other hand, was primarily targeting large enterprises, because of the much larger deal sizes on offer.
Nirvanix never really succeeded in evolving beyond offering storage-as-a-service. Amazon sells bucket-loads of storage, but in large part this is because the storage is attached to some other application or service also running on AWS. Because Nirvanix could not effectively compete as a broad-based IaaS provider (see point 1), this ultimately limited the extent to which customers regarded it as a strategic cloud partner; if the customer wanted anything other than storage in the cloud, they had to look elsewhere.
Lessons to learn?
While its always tempting to view one failure as symptomatic of a wider issue, it's important to view it as just that; the failure of one relatively small operation. Nirvanix is not the first – recall Iron Mountain's decision to pull its cloud storage service a couple of years ago, and even the failure of storage service providers such as StorageNetworksInc over a decade ago? – and it probably won't be the last, though it is one of the most high-profile businesses of the cloud-era to actually fail.
Ultimately though, we think Nirvanix's problems were self-inflicted. The market for cloud storage undeniably exists, but as a small startup Nirvanix's focus was on the wrong part of the solution. Nirvanix will go down as a cautionary tale and a brutal example of that old technology maxim; innovate, or die.