Talking virtualization with a FlexPod user
Fast deployment the mantra for Mantra
Sydney’s vForum conference was the venue for the Cisco / NetApp / VMware tie-up getting one of its first major Australian outings on Wednesday October 20.
Vendors pitching into the cloud/virtualisation space – including the trio above – are increasingly creating reference architectures designed to give customers something like a “single box” way to roll out solutions.
The FlexPod solution, launched earlier this year, had its first detailed customer implementation discussion, with resort accommodation operator Mantra Group put in front of lunching journalists to talk about its experience.
As it happens, some of its outcomes sound pretty good: having outgrown its first broom-closet IT department, and then outgrown its internal proper data centre (chiefly bumping its head on power and cooling constraints), the fast-growing tourism business got its new architecture as part of a move to an external facility in a Varsity Lakes, Queensland, data centre.
Mantra’s reasons for virtualizing its infrastructure centre around its fast acquisition-based growth, having added more than 100 properties over three years to create a 16,000-room national operation.
Group general manager for IT Garry Rich said that “server sprawl”, storage constraints and backup windows were also among the “hard limits” it was encountering in its former DIY approach to IT.
He cites fast rollout and vastly reduced electricity consumption as two key benefits the company has enjoyed: “We deployed 24 hosts in a day,” he said, “and we’ve gone from 90 servers down to 24, so the energy saving is massive” – currently, power consumption is down by around 60 percent.
“The real estate, power, and cooling savings justify virtualization on their own”, he said.
In endorsing the FlexPod solution, Rich particularly highlighted the simple deployment with a “piece of tin” with one set of plugs. As well, being able to deploy the virtualization as a “stateless blade” offers very fast recovery.
Some users still need to be convinced, he noted: “People tend to host their applications on the biggest server they can buy – ‘right sizing’ the host for the application can make people nervous.”
Often, those concerns can be got around by stealth: leaving the existing server in place, but when an application needs new processing, adding the extra capacity into the virtualized environment.
There’s also a business process advantage, Rich said. Adding new capacity in Mantra’s former DIY environment was a much larger bureaucratic exercise, in which he would have to scope the new system, choose products, conduct due diligence on the technology (including its interoperability with what was already in place), create the project, estimate the staffing requirements, put forward a business case and get a sign-off. New capacity under the FlexPod regime is much more straightforward, he said.
Even the “big bang” project when Mantra first implemented its new environment was fast, he said: “the rollout window was about ten days”, he said.
The FlexPod environment, built on Cisco’s Unified Computing System, uses Cisco Nexus switches, NetApp FAS3210 storage, and VMWare virtualization. It was implemented by Cisco channel partner Alphawest.
Along the way, Rich also tossed some bouquets at the Australian government’s planned National Broadband Network, from a business point of view.
Resorts, he noted, are frequently in regional locations (the Gold Coast in Queensland and the Hunter Valley in NSW come to mind), and are consequently often in broadband “black spots”.
As a result, the company’s preference for centralized IT is constrained by connectivity. Some applications have to be deployed out to the resorts, either because they can’t get a fast enough connection, or because carriers won’t offer reasonable SLAs in distant locations.
If, for example, restoration times are measured in days rather than hours, centralizing a property management application is too great a risk, and the idea of a sizeable application running 3G backup is questionable.
“Centralising all of the IT only works if the communications will support it,” Rich said. “The NBN helps the case for centralization.”
It’s not an inconsequential consideration, Rich claimed: “the capex benefits are massive, and the operational costs can be cut by 50 percent.” ®
Sponsored: Hyper-scale data management