Gap Between Disk and Tape Is Too Large to Ignore

Arthur Cole spoke with Jim Sherhart, director of product management, PowerFile Inc.

Cole: PowerFile has one of several new storage appliances on the market this year. Is this a sign of the recession perhaps? People looking for low-cost, easily deployed and managed storage?
Sherhart: I don’t know if it’s necessarily a function of the economy as much as it is a function of how enterprises prefer to buy storage. Customers don’t want to get hit with hidden costs, don’t want products that require a system integrator to install, and most importantly they don’t want to take unnecessary risks. When you consider the success of NetApp, it’s safe to say the appliance model is proven and I think buyers take comfort in that fact.

Cole: Your Hybrid Storage Appliance features a "distributed-performance architecture." What is that, and how does it improve functionality?
Sherhart: Our product uses a combination of cost-effective SATA drives on the front end, an array of Blu-ray media on the back end, and a virtual file system that makes it easy to manage all data in dynamic volumes. The purpose of the SATA front end is to act as a fast, dynamic cache to stage both writes coming into the system and archived information that is currently being accessed.

What we discovered on earlier systems is that the SATA would become a bottleneck during periods of both heavy read and write operations. The new architecture of the Hybrid Storage Appliance offloads many of the operations previously performed in the SATA cache down to the library controllers that have their own cache. In other words, the system essentially distributes I/O operations between the cache and the library, allowing our performance to scale as more libraries are added to the system.

Cole: Are you predicting the end of disk-based long-term backup at this point? Can it be justified economically anymore?
Sherhart: I think customers are starting to understand that the old backup tape is not an acceptable archive today. Employees expect data to be online and accessible, not to mention the need to have data more accessible for eDiscovery and litigation support purposes. In other words, tape is not an option.

Having said that, the data we typically address, such as fixed content or persistent data, is not mission/business-critical and therefore does not require millisecond response time, so a pure disk-based system is overkill. This is especially true when you add in the unnecessary operational burden -- including replication, backup, maintenance, data migrations -- of storing unchanging data that must be kept long-term on something volatile like spinning disk.

Our belief is that the gap between disk and tape is too large to ignore and that no one storage technology is going to solve the long-term archive issue. The key to success is going to be a virtual file system that can seamlessly manage multiple underlying storage technologies, and we feel very comfortable with the value proposition our system offers when compared to any other technology available today.