A few weeks ago I had a chance to speak at length with Andres Rodriguez, the incredibly passionate founder and CEO of Nasuni. Nasuni is a highly innovative storage company providing storage infrastructure backed by the cloud. I’ve been writing a lot about caching and flash in virtual infrastructures, and went into the conversation thinking that they’d be another company improving storage performance with SSD, oh, and they had this cloud thing going on. Boy, was I wrong. After a lot of questions, I came away with a real respect for what they’re doing: attacking a number of big storage-related enterprise IT problems all at once.
Nasuni bills itself as Storage Infrastructure as a Service. Its goal, put simply, is to free IT staff from the task of managing storage. All aspects of storage. Acquisition. Day-to-day management. Backups. Replication. Monitoring. Upgrades. All of it. From a customer perspective, its systems manifest themselves as a storage controller on premise with your infrastructure. This storage controller is most typically a physical server that they supply, but it can also be a virtual machine suited for use in a cloud situation (there’s an AMI in Amazon AWS, for example). The controller, called a Nasuni Filer, is a multiprotocol device speaking various file-level protocols like SMB, NFS, and HTTPS; security protocols like Active Directory; and the block protocol iSCSI for your virtualization environments. It also speaks deduplication, compression, and encryption, encoding your data with encryption keys you supply so your data is protected from competitors and governments both in flight and at rest.
The Nasuni Filer is a gateway and a cache. The gateway is to cloud-based storage, supplied by multiple vendors that you, as a Nasuni customer, won’t have to worry about because they handle the redundancy and the contracts. The nature of this storage means that capacity is effectively infinite. But as anybody will tell you, capacity is but one aspect of storage. Performance is the other, and cloud storage doesn’t usually lend itself to performance. That’s where their caching comes in. The Filer, essentially a commodity server loaded with SSD, is sized to cache 99% of all disk accesses, both writes and reads. They do work on the front end to size the controllers correctly, and if the workload changes over time, the Nasuni operations staff will take action to correct the situation with the customer.
“This is all neat,” you might be thinking, “but there are a bunch of products out there that are cloud gateways.” Yes, but put another Nasuni Filer somewhere else in the world and all your data shows up there, too. Neat trick, to eliminate storage administration in remote offices and for disaster recovery. Doubly so when you consider that your data can appear in Amazon AWS, courtesy of their AMI, potentially removing the need for a private DR site. There is what I call a propagation delay, of course, because interacting with the cloud isn’t instantaneous. But it isn’t long, either. Usually 5 minutes or less for data consistency, and less than a minute for snapshots to become consistent worldwide.
Snapshots form the basis of one of the more surprising aspects of Nasuni’s offering. First, you can make unlimited snapshots, which is pretty useful. After all, Amazon and Microsoft can store nearly infinite objects with no performance penalty, so all Nasuni needs to do is keep track of all those objects. Because you have such a robust snapshot capability built into a file-based storage system, you don’t really need backup anymore—you have it built right in. Tape is one of those storage technologies we love to hate, usually justified as a necessary evil. Deciding that it is actually evil and throwing it out for good just got easy.
So what do you pay for all of this? Unlike pricing for typical on-premise storage, you only pay Nasuni per usable terabyte of storage. The controllers are figured in with a scheme for “hardware credits” based on capacity. Snapshots are free, too—the savings from Nasuni’s compression schemes don’t get credited to customers, but are instead given back as unlimited snapshots. That’s a big incentive to consider their methods for backup. For your money you also get 24×7 monitoring of your systems by Nasuni’s operations staff, further removing IT staff from day-to-day operations of storage. To me, that’s a welcome idea. Storage administration is complicated, delicate, and risky; hard to monitor; and fraught with terrible UIs and expensive forklifts. And there isn’t a more thankless job than backup administration. Imagine if we could be done with both of those, solve our DR problems, and manage all of our branch offices’ storage centrally, too.