Pacific Aluminium was eager to set up the service desk on-premises rather than follow the outsourced model its parent company had adopted. But with a remit to separate as quickly as possible, deployment had to be be swift — that reality pushed the business towards cloud.
"We had an aggressive timeline to separate from Rio Tinto, aiming to be fully independent by end of April 2013. The possibility of an IPO, or being sold, was a key driver in the initial infrastructure design. We needed to be able to integrate into other systems and scale up or down as seamlessly as possible, thus the decision to go down the IaaS path in the first instance," says Gavin Bills, general manager of information systems and technology at Pacific Aluminium.
From a SaaS perspective, Bills and his team put together an implementation of service desk application ServiceNow in 30 days from concept to go live — the fastest implementation of their product offering to date.
ServiceNow is used internally by over 100 service desk staff serving over 5,000 Pacific Aluminium employees who generate over 400 tickets a week.
Bills says the split from Rio Tinto gave the company an opportunity to build its enterprise capacity from scratch, allowing it to design and size for its own specific requirements.
"You could almost call it a Greenfield build," he said.
Pacific Aluminium began planning for IaaS with the vendor in mid-2012 going through various configuration and pricing iterations in the process. It decided to use the cloud model for all of the company's enterprise-level systems and services. These include Active Directory, Microsoft Exchange, SAP, and all other non site-specific applications. The company chose not to go down the fully-managed path in the first instance, opting instead to take ownership "from the OS [operating system] up." According to Bills this better suited the requirements at the time.
From there on, the core infrastructure was in place in a matter of months.
"Parallel comms were built and we had the enterprise Active Directory up and running utilising IaaS around September 2012, with migrations of users and apps scheduled to commence on a site-by-site basis from October 2012. So our core systems had to be in place by then. We were migrating approx. 5000 users and 350 production servers across all sites, the infrastructure in place had to be robust enough to support and sustain these numbers and provide the ability to scale up and down as required. This was particularly important during the first phases of the migration where we required additional capacity."
Despite achieving a rapid rollout, Bills admits the project was a painful learning experience for both parties.
"Inability to supply on time, and differing views on just what IaaS incorporated, led to a high degree of frustration," he said.
"That said, given our tight timelines, we were demanding in our requirements and pushed hard to get what we needed," added Bills.
"To their credit, the customer-facing team we were dealing with worked hard to ensure we achieved the desired outcomes. When it works, it works well, and becomes a vendor supplied commodity with the internal tech team maintaining a watching brief."
Despite all the hype about the cost savings involved in cloud storage, Bills strongly suggests organisations take a pragmatic approach when assessing their data requirements on cloud projects.
"It can be a major cost if not managed," he said.
"Ensure you understand the back-up regimen and associated charging structure — 1 byte changed could result in a full back-up with resulting additional costs in storage. Ensure your chosen vendor understands the back-up regimen and get them to outline it in writing. Clearly define who has accountability for what, particularly with respect to restoration timeframes and accountabilities.
"Regularly check your processor and disk utilisation and make sure you can scale up or down as required."