Parallel NAS software maker ParaScale Inc. is coming out of stealth with a beta version of its ParaScale Cloud...
By submitting your email address, you agree to receive emails regarding relevant topic offers from TechTarget and its partners. You can withdraw your consent at any time. Contact TechTarget at 275 Grove Street, Newton, MA.
Storage product (PCS), which can turn a group of commodity servers into a parallel NAS farm.
While ParaScale executives acknowledge their approach to network attached storage (NAS) is not for every workload, they're hoping to undercut planned systems, such as the Hewlett-Packard ExDS9100 and EMC's Hulk, on price.
Users can download a free 4 TB license of the PCS software from ParaScale's website. CEO Sajai Krishnan said the VM-based software package can be used indefinitely, though users must purchase licenses if they want to move on to the retail version or receive support.
Krishnan said ParaScale has five alpha and early beta testers, including WAN optimization and gateway vendor Blue Coat Systems. Blue Coat's internal IT department is testing a 3 TB instance of the first beta version of PCS in a disk-based backup environment. "We're a big NetApp shop, but we'd never put a NetApp filer in that type of environment – it's just too expensive," said Jonathan Bensen, Blue Coat's systems and infrastructure manager for IT.
Blue Coat might eventually put PCS into use for an internal cloud, "like an internal Facebook," where company employees can share documents and multimedia files, Bensen said. "That needs a ton of storage, but I don't know that it needs NetApp's enterprise features to store that stuff."
PCS goes one-on-one with nodesOther clustered NAS systems place control nodes between the client and the storage server, or parallel systems that stripe small chunks of data over all disks to boost performance and resiliency. ParaScale's system maintains one-to-one relationships between clients and storage nodes. If there's a "miss" – i.e., a client asks its storage node for a file it doesn't have stored locally – the file will be sent from the node that owns it and delivered to the client from its storage node's cache.
That makes PCS useful only for large sequential reads and writes, which is typical for Web 2.0 installations that need to offer streaming multimedia files. "Ours is not the kind of architecture best suited to locking files for multiple concurrent writes or small-block reads," Krishnan said.
However, automated replication of heavily accessed content across multiple nodes is on ParaScale's roadmap for its first general release sometime next year. That would boost the performance into the Tier 1 range for online data centers.
While not in the data path, the system also relies on a Control Node to direct file transfers. The nodes can be beefed up with solid-state media for added performance, but scalability beyond two nodes will also have to wait until next year.
The basic configuration runs ParaScale's proprietary software in user space, making it compatible with most hardware, and PCS uses NFS, HTTP and FTP. Theoretically, since it's based on server hardware, applications can reside with storage on the same box.
Bensen's used Symantec's NetBackup software to tune the size of backup files in order to take advantage of the ParaScale architecture.
Looking to sign up ISPsIn addition to private clouds in the enterprise, ParaScale is marketing the software to small ISPs looking to compete with Amazon or emulate Google. Illuminata analyst John Webster said ParaScale offers individual users and smaller companies "hardware defined by what's already out there as a commodity, using the economies of scale already available in the wider marketplace".
It's all part of what some industry watchers and vendors say is the beginning of a convergence between server and storage hardware, and between storage and LANs in data centers. Webster said that as unstructured data grows, interactive applications become popular, "more IT people will come to conclude, 'whatever the cloud guys know, we need to know, too.'"