Jim Ginsburgh, BP’s Vice President of Enterprise Architecture, gave a presentation on the consumerisation of IT on the LEF Executive Programme’s 2006 study tour. During that tour, he learned about Amazon’s plans for cloud computing. When Amazon announced the beta availability of its EC2 service, Jim asked the LEF to help BP join the beta programme.
With that first step, BP began a journey exploring the various ways the cloud can add value to its business. The first experiment was an effort to respond to a group that wanted to set up a new customer-facing website. When this group asked the IT organization for a server, they were told it would take 3 months. Unhappy with this response, the group was eager to be a test case for using the cloud for site development and testing. From an IT perspective, the data center staff was also happy to see the project moved outside their production environment.
The web development team’s experience building their new website was good — so good that they wanted to go to production with it. This triggered concern from BP’s security group. After much discussion with Amazon, it was agreed that BP would run a penetration test to see if they could get into Amazon’s systems. When they could not, the new customer-facing website was allowed to go live.
After this initial Amazon success, BP began a series of experiments in the cloud. One of the most important was with SAP, where BP has undertaken a substantial consolidation of systems, which has required new processes and designs. The problem, however, was a lack of available servers for the process teams to do the necessary development and testing. Queues and delays resulted, because IT saw this as a peak load problem and so did not want to commit more servers to it.
BP was intrigued by the idea that, in the cloud, it could run multiple development and test systems in parallel, greatly reducing delays. Initial efforts were not very successful. The first Amazon virtual machines were not fast enough to run SAP’s R3 and its database effectively. However, as Amazon began to introduce more capable servers, it soon became practical to run SAP in the cloud.
Thus cloud computing solved a problem which BP’s existing data center had no cost-effective way of addressing.
Along the way, BP learned a number of lessons both about the cloud and about itself. For example, it discovered that there was a very substantial cost associated with its IT procurement process. Development teams not only had to navigate the capital expenditure process, but they also had to perform due diligence in the data center, determining where the boxes would go and their requirements. A second lesson was that, although its internal server provider would provide free replacement parts in case of a technical problem, the cost to BP of configuring and installing a new physical server was far from free.
Like others, BP learned that the cloud offered little price advantage if the server was run around the clock, every day of the year (although Amazon recently announced a price cut of 30 percent for those firms willing to commit to a year’s contract instead of buying on demand). BP also learned that storage in the cloud, while having different response characteristics from local storage, could cut costs by as much as 90 percent.
— From "Digital Trust In The Cloud," by CSC's Ron Knode