CHICAGO -- In two to three years, users will not be able to distinguish between network attached storage (NAS) and storage area networks (SAN), according to Network Appliance CEO Dan Warmenhoven.
Warmenhoven, who delivered a keynote address Wednesday at Storage Decisions 2001, maintained that the lines will blur so much over the next few years that the distinction between the two models will go away.
He cited Network Appliance's newly released Snap Manager for Exchange 2000 as an early version of a future SAN/NAS hybrid. Snap Manager takes SCSI commands, encapsulates and sends them to the NAS device where the commands are de-encapsulated and processed. Although this approach was born of necessity when Microsoft decided not to support NAS devices in its latest Exchange Server, Warmenhoven said it quickly became a "tactical, and strategic" move on the part of Network Appliance.
Warmenhoven said his ideal world involves Gigabit Ethernet (or 10 GigE) as the primary interconnect technology; iSCSI or NFS/CIFS as the primary data transport modes; and most applications (along with network end-users) accessing data by name, not by the more "primitive" blocks.
Fibre Channel is a "transitional interconnect technology," said Warmenhoven, to be replaced by Gigabit Ethernet and 10 GigE installations running iSCSI or NFS/CIFS.
"I don't agree that there are inherent performance limitations in Gigabit Ethernet," he said, citing significant
As the lines blur between SAN and NAS, Warmenhoven envisions devices that will have block-level and file-level access to data. He also believes that file-level access will eventually become the preferred way of moving data.
"Block access is to storage what an IP address is to the Internet," he said.
When most people type a URL into their Internet browser, they aren't likely to type the IP address, he said. They are more likely to type the Web site's name. Block access is the same as an IP address.
"It's much easier for humans and applications to access things by name," he said. Warmenhoven views block-level access as a primitive form of data movement that will "trend over time to accessing things by name."
Although he admitted application vendors are not all jumping on the file-level access bandwagon yet, Warmenhoven sees this as an issue of an investment they will ultimately make to keep up with the evolving trend toward file-level access.
Attendee John Blackman, systems architect, Wells Fargo, Minneapolis, Minn., responded to Warmenhoven's views on the future of storage architecture by saying he sees Network Appliance "driving the (development) of technologies we will be using, like DAFS (Direct Access File System). They will be a big driver of DAFS and iSCSI. They were also a big driver of CIFS."
Blackman works for Wells Fargo's Emerging Technologies group that analyzes the merits of evolving IT technologies and helps Wells Fargo's business units make informed decisions about IT investments and deployments.
He also sees a future where storage subsystems will be smarter, more intelligent in their communications with the I/O and CPU modules on the network. "To make a call from the CPU to the disk, it needs to be done efficiently and via a direct model," he said.
Blackman has great hope that the DAFS protocol will provide this type of efficiency via an iSCSI transport. He also freely admits that DAFS and iSCSI may not be "the final solution, but they will still help us advance the technology."
FOR MORE INFORMATION: