Symform’s Peer Architecture may be the Answer to High Cloud Storage Prices
Symform’s Peer Architecture may be the Answer to High Cloud Storage Prices
Most vendors in the storage industry, at one time or another, have heard potential users ask the same question: “Why does storage cost so much when I can buy a 1TB disk drive for $50?” The answer is that there’s obviously a lot more to the cost of a business-class storage system than raw disk capacity. The cost of the infrastructure, controllers, networks, software and overhead are typically many times the raw cost of storage.
Now cloud storage providers are hearing the same refrain from potential users about the price of their offerings and not surprisingly, are giving the same explanation that infrastructure costs are what’s causing it. Cloud providers typically have very large upfront costs, essentially everything the private user must provide, but ‘supersized’. There’s also an aspect to cloud storage that makes these economics even worse, unpredictability.
Every IT organization has to deal with scaling storage capacity and the difficulty in predicting user demand, plus the timing of that demand. But cloud providers have it much worse. They have to build data centers in advance of potential need and put enough capacity online to handle the peak load requirements for a given customer base. The result is typically very low utilization rates for this infrastructure, which drives up the costs they must charge each user. A new cloud storage provider, Symform, has taken a different approach.
Symform has created a decentralized cloud infrastructure that taps unused capacity that its users have available. For a flat monthly fee, one that they claim is 10x less expensive than traditional online storage services, customers can upload as much data as they want to the cloud, as long as they provide the capacity to cover their usage. Here’s how it works.
The Symform client is loaded onto the Windows computer or server to be cloud connected (Linux and Mac are in the works). This system must have an always-on ‘business class’ internet connection with 512Kbps upload and 1Mbps download bandwidth available. Symform provides a spreadsheet tool to calculate how much bandwidth will be needed based in data set size and other factors.
The client monitors the designated folders and automatically uploads data to the cloud. In this process, data’s broken up into 64MB chunks, encrypted and then divided into 64 x 1MB fragments. Symform adds 32 parity fragments using a RAID algorithm and uploads the entire 96 fragments to the cloud. These are randomly distributed to 96 separate nodes in the Symform cloud network.
When devices fail or data is corrupted in transmission the system can automatically regenerate the lost fragments. This 1.5:1 ratio (96 total fragments for each 64 data fragments) provides a high degree of data protection, what they call a “Resilient Storage Architecture”.
Each client computer provides storage for the Symform cloud network equal to 1.5x the amount of data they upload to the cloud. This covers the 1.5:1 ratio outlined above. For most users this is the easy part, finding unused storage in their environments. In fact that’s one part of the appeal to the Symform technology, capturing and making use of otherwise unused resources.
Geographic data dispersal is certainly not a new concept. It’s being widely used in many cloud infrastructures today. The technology to parse data into little pieces and spread it around is stable and peer networks have been proven over the past 10-15 years. Symform is leveraging these two technologies to create a novel solution to a perennial storage complaint, high cost. In the process users are getting increased security and availability, since this encrypted data is spread across so many physical locations.
Storage Swiss Take
Once people realize they’re not trusting their backups to Napster but actually leveraging the same technology that most other cloud storage providers use Symform could really catch on. The appeal of lower cost storage is something that users will always respond to and the idea that they can actually use the TBs of storage sitting idle in their environments will generate a lot of trials. As for the competition, the overhead cloud providers must pay to cover potential peak demand will always be there, making it more difficult to match the cost structure of Symform’s highly decentralized architecture.
Friday, September 2, 2011
Eric Slack, Senior Analyst
VMworld 2011 Blog