The guarantee will be part of the contracts when the Axion 600 disk array begins shipping next month, but terms are vague and could vary from customer to customer.
The Axiom 600 disk array has twice the capacity and performance of the Axiom 500 model, which will remain on the market. Pillar disk arrays are made up of controller modules called slammers and capacity modules called bricks. A single Axiom 600 brick can support up to 24 GB of RAM, two dual-core AMD processors, four 4 Gbps Fibre Channel ports, eight Ethernet ports and up to 6 TB of storage. The Axiom 600 system can scale to 96 GB of RAM, 16 Fibre Channel ports, 32 Ethernet ports and up to 832 TB raw storage.
Pillar is also adding application-tuning features to the Axiom management software, including support for Oracle's Automated Storage Manager (ASM). Application profiles, which are also available on the Axiom 300 and Axiom 500, provide a template for the arrray to modify I/O priorities, block sizes and the layout of data on disk according to application.
Pillar claims these features improve storage utilization because they allow applications with different performance profiles to share storage without contention for system resources. Mike Workman, Pillar CEO, said that allows Pillar to guarantee better than 80% utilization, along with the lowest dollars per IOPS price in the market and five 9s of availability.
But the specifics of the guarantees are vague. The terms would be negotiated on a "customer basis," Commins said, claiming the guarantees would be "to remediate any issue, as well as financial pain." The terms of the guarantee would have to be negotiated as part of the original sale, he added.
When asked how these features differ from its current support and services agreements, Commins said the new language is a response to high-end customers with a "new-vendor risk objection. " In other words, "we haven't yet earned the right to make claims unilaterally, like maybe IBM would," he said. "This is our way of showing we stand behind our products."
Networking equipment manufacturer Foundry Networks has standardized on Pillar storage for primary and disaster recovery sites, and Pillar has hit the guaranteed percentage with approximately 30 TB out of its usable 36 TB utilized, said Foundry CIO Dave Riley. Foundry uses some of Pillar's original quality-of-service features to ease contention in Oracle and Exchange, but has not yet investigated the more granular application tuning features.
Carlos Rosa, storage and Unix systems administrator for Foundry, said the company has set Oracle and Exchange production databases at the "premium" level and uses the rest of the disk capacity on two Fibre Channel bricks for backup and archive data from both applications. Other applications, such as file shares, are stored on SATA-based bricks in Axiom 500 arrays. "Before Pillar supported Fibre Channel disks, we initially put them all on the same bricks and set the priority to high," he said. "We didn't notice any issues."
At least one analyst is uncertain that the utilization guarantee will work for all customers. "It seems like a hard one to do," said IDC analyst Rick Villars. "To run at 80%, the user has to be pretty confident about capacity needs going forward."
While happy with the utilization guaranteee, Rosa said that Pillar's array reporting features leave something to be desired. He said he'd like to see more trending and forecasting reports, plus a report on the utilization of the array's Fibre Channel ports and switches. "I want to know how much data is going through each channel," he said. Pillar offers a health check feature called Pulse, but it is designed as an overall environment-optimization service, rather than a day-to-day storage resource management tool.