So Crazy It Just Might Work - InformationWeek

InformationWeek is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them.Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

IoT
IoT
Software // Information Management
Commentary
4/21/2008
02:47 PM
Connect Directly
LinkedIn
Google+
Twitter
RSS
E-Mail
50%
50%

So Crazy It Just Might Work

Startup RevStor cuts your unstructured data into chunks and stores them on the PCs and servers on your network -- no NAS or SAN required. Is this plain crazy, or just crazy enough to work?

Startup RevStor cuts your unstructured data into chunks and stores them on the PCs and servers on your network -- no NAS or SAN required. Is this plain crazy, or just crazy enough to work?As I read through RevStor's Web site in preparation for an interview with CEO Russ Felker, the one thought running through my mind was "This is insane. Why would anyone buy this?"

But after speaking with Felker, I'm beginning to think there's a place for his software.

RevStor SANware is a unique approach to storage virtualization. Unlike Rainfinity or Acopia, SANware doesn't mediate between clients and NAS filers or a SAN. Instead, you deploy SANware agents on the PCs, laptops, and servers already on your network. The agents then construct a virtual storage repository consisting of all the available disk space on those machines.

When users save data to this virtual repository, the agents cut data into chunks, encrypt the chunks with 256-bit AES, and squirrel them on other computers. The agents also create a distributed file system to track the physical location of each data chunk.

Felker says there are several benefits to his approach. First is security. Because the data are chopped up and dispersed, you can't lose an entire backup tape filled with customers' Social Security or bank account numbers.

SANware agents will copy each piece of data between 2 and 20 times, depending on how much redundancy you want. The file system also ensures that no more than 33% of any file goes on one machine. The encryption provides an additional security layer.

That's true to a point, but I'm not convinced enterprises are ready to give up tape just yet. And what about sensitive information stored on laptops? It's likely that users who save sensitive files to the distributed system may also have a copy on the local drive, fully intact and unencrypted.

Second is performance. Rather than point every user to the same SAN or group of NAS filers and create a bottleneck, Felker says the distributed file system can be more efficient because multiple computers will deliver small chunks of data. He says restoring a 1-GB file might take a minute and a half to two minutes.

"It works in a similar fashion to BitTorrent, so you don't have a big network hit, because multiple systems are getting the data," says Felker.

The third benefit is financial. RevStor charges $999 per TB of data stored, not per agent. So you could deploy 10 agents or 100 agents for the same $999. You also get to capitalize on the unused disk space on computers you've already purchased. It's a lot cheaper than dedicated storage, and practically free compared with the price of a typical file virtualization product.

The distributed file system, which is a proprietary creation of RevStor, is housed by supernodes. The supernodes keep track of the actual file locations. If one supernode goes down, the others maintain the file system.

When a user wants to restore a file, the agent sends a request for it. A supernode picks the optimal download locations based on CPU and network utilization, sends those back, and the agent downloads the file from multiple systems simultaneously.

As an added layer of protection, pseudo-supernodes also maintain a copy of the file system, but don't handle actual requests to store or restore files.

The software also includes data deduplication capability to cut down on unnecessary disk use. If changes are made to an existing copy of a file, the system only copies the changes.

Convinced? Felker says he's aiming at midsize businesses up to the enterprise. I can see a smaller or midsize deployment, but the product lacks significant enterprise features. For one, there's a reason NAS and SAN can get expensive -- you get higher-quality disk than a typical PC or server.

For another, the product can't provide granular, policy-based storage options. You can set the agent to back up the contents of a hard drive, but it backs up everything regardless of business value, so financial spreadsheets get the same treatment as the MP3s on your users' desktops.

Finally there's the issue of capacity. You can add more servers or PCs as your need for space increases, and the system can alert you as you approach your limit, but it also means more PCs and servers to manage. As companies grow, a centralized storage system may make more sense from a management perspective.

That said, RevStor may be on to something. Sometimes it helps to be a little crazy.

We welcome your comments on this topic on our social media channels, or [contact us directly] with questions about the site.
Comment  | 
Print  | 
More Insights
News
COVID-19: Using Data to Map Infections, Hospital Beds, and More
Jessica Davis, Senior Editor, Enterprise Apps,  3/25/2020
Commentary
Enterprise Guide to Robotic Process Automation
Cathleen Gagne, Managing Editor, InformationWeek,  3/23/2020
Slideshows
How Startup Innovation Can Help Enterprises Face COVID-19
Joao-Pierre S. Ruth, Senior Writer,  3/24/2020
White Papers
Register for InformationWeek Newsletters
Video
Current Issue
IT Careers: Tech Drives Constant Change
Advances in information technology and management concepts mean that IT professionals must update their skill sets, even their career goals on an almost yearly basis. In this IT Trend Report, experts share advice on how IT pros can keep up with this every-changing job market. Read it today!
Slideshows
Flash Poll