As the number of high-speed processors that can be packed into a server continues to increase, many IT organizations are about to confront some significant I/O challenges. Each of those processors is capable of hosting large numbers of virtual machines and containers trying to access a common pool of block, file and object storage resources. New classes of servers based on NVMe backplanes will go a long way to solving some of those issues. But a case can also be made for new approaches to software-defined storage software.
WekaIO this week emerged out of stealth to address that issue by unveiling a distributed parallel file system capable of accessing data in a block, file or object storage format. Barbara Murphy, vice president of marketing for WekaIO, says the company’s namesake software has also been engineered in a way that makes managing all those various types of storage feasible for the average IT generalist.
“We built a distributed parallel file system from scratch,” says Murphy. “But it doesn’t require a storage specialist to manage it.”
In addition, WekaIO provides connectors to S3 and Swift object storage systems, which Murphy notes allows organizations to deploy a common file system both on-premises and in a public cloud in a way that lets them mix and match servers and cloud platforms as they see fit. Later this year, Murphy says, WekaIO will also move to unify the management of storage across a hybrid cloud deployment.
IT organizations of all sizes are soon going to discover that while the amount of compute horsepower that can be packed into a server has increased exponentially, the I/O requests all those processors generate will test the limits of legacy storage systems. Of course, distributed parallel storage systems have been around since the mainframe. The issue WekaIO is trying to solve is making such an approach to storage accessible to the average IT administrator working in a classic distributed computing environment.