By Kevin Komiega
Storage software upstart Parascale has big plans for little servers. This month, the company launched a software-only storage platform called the Parascale Virtual Storage Network (VSN), which uses existing disk capacity on Linux servers to create a petabyte-scale virtual storage pool for storing digital media content, including video, images, and audio files.
The VSN software aggregates the disk storage on commodity Linux servers to provide one or more virtual file systems. The software configures one Control Node with as many as several hundred Storage Nodes, where each node is a server.
The Control Node directs traffic while all data moves directly between client applications and the Storage Nodes, allowing for massive read/write bandwidth. With I/O speeds of more than 50MBps at each Storage Node, the VSN can be tuned to deliver tens of gigabytes per second of aggregate I/O for applications with high-bandwidth requirements, such as video-on-demand (for reads) and video surveillance (for writes).
Applications talk to the Control Node, which redirects the application to the appropriate Storage Node, or nodes, to access the data. File data typically resides in multiple locations under the VSN for redundancy.
The software supports x86 servers certified for Red Hat Enterprise Linux 5, which are available from a variety of OEMs, including Hewlett-Packard, IBM, and Dell. Because client applications access the VSN via NFS, HTTP, or FTP, custom client-side software is not needed.
International Data Corp. (IDC) reports that digital content is the fastest-growing segment of the storage market, with volumes doubling annually. Richard Villars, vice president of storage systems research at IDC, says dramatic growth in the use of video, image, and business analytic applications, coupled with regulatory mandates for information retention, are changing the storage landscape and creating opportunities for new vendors such as Parascale as enterprises seek to monetize their information assets.
“Along with the emergence of new companies that want to establish gigantic ‘content depots’ for collecting, storing, and organizing rich content for consumers and businesses, this trend is driving a need for storage solutions like Parascale’s that provide both the scalability and flexibility to meet exploding information storage and access needs,” says Villars.
Some users might hesitate to cobble together a storage network using off-the-shelf PCs, but Villars says Parascale’s VSN could win over some naysayers. “A virtual storage network built on commodity hardware may seem risky, but with the right architecture it can be very reliable,” he says.
Bill Evans, Parascales’ CEO, claims that reliability is not an issue with the VSN. “Our strategy is to use software to make a collection of PCs into extremely reliable storage for digital content,” says Evans. “Nodes are going to have failures almost every day, but our software can deal with the loss of a server by replicating data across PCs.”
The Parascale software provides three ways to transparently move data among servers under the VSN. First, automatic file replication copies files across servers for data availability in the case of disk or server failure. Second, automatic file migration distributes files across multiple servers, eliminating server hotspots and improving server utilization. Third, load- balancing distributes file requests across all the servers that have a copy of the requested file, reducing the load on any one server to maximize system throughput.
The VSN allows administrators to scale file storage and bandwidth independently. Parascale delivers scalability in disk capacity because each server manages its own direct-attached storage and the CPU-intensive block management that it entails. It also delivers scalability in bandwidth because multiple applications access multiple Storage Nodes in parallel.
For example, if a new server is added to expand storage capacity, Parascale’s software will automatically migrate existing files to the new server. As a result, new files are written across all servers and disks, and the new server does not become a bottleneck in system performance.
The Parascale VSN is licensed by annual subscription. The software is priced per disk drive rather than capacity, allowing customers to keep costs down as they upgrade to higher-capacity drives. Because Parascale does not depend on RAID, users can replace failed drives with new units without any change in software license fees. Pricing for the VSN ranges from $300 to $700 per drive.