How Proper Capacity Planning Can Impact your Product Launch
An iconic brand looks back at how their IT team set the stage for future product releases from the success of their initial iPhone launch. The biggest and most successful product launch in company history at the time.
When your key principles are to focus on the customer and deliver value to the bottom line, you have little room for error. The Verizon Wireless team that headed up the company’s initial launch of the Apple iPhone, which was the company’s biggest launch at the time, needed to build out, upgrade and accurately predict how their IT infrastructure would react on launch day.
Plan for Demand Peaks
Verizon Wireless must be prepared for peak selling seasons so infrastructure upgrades are completed well in advance. These peaks come during holidays such as Christmas, Mother’s Day, Valentine’s Day, Thanksgiving, and during product launches. Demand in retail stores and Customer Service centers on big retail selling days create significant spikes.
This time, the team only had six months to measure, plan, and implement, so the team started its push in July knowing that final hardware installs had to be completed by October for a December launch.
Modeling is important to simulate anticipated demand levels. “For example,” said the IT leader, “we are able to accurately predict peak utilization within 5% on critical Point of Sale systems on Black Friday and Cyber Monday.”
The team referenced historical launches as a baseline. “A baseline was created for each of our regions and broken down across the various sales channels,” he said. A series of scenarios highlighted demand on every region and channel depending on the different traffic levels. “Baselining our application servers helped us isolate the ideal number of processors to add to the database servers and enable us to deal with major shifts in traffic volume.”
Changes Made According to Findings
Based on various models, a wide range of changes were instituted to prepare for the iPhone launch and to ensure that there were no system failures or slowdowns during the launch.
For open systems, they installed and configured hundreds of logical environments on their servers. “Additional standby systems were available for rapid deployment for spikes in usage,” he said. “As the initial models were accurate, though, these standby computers were never required.”
On the storage front, four new frames were included to add 800 terabytes of extra capacity available for the launch. The company optimized firewall efficiency and configurations to permit rapid internal and external application interfaces and faster processing. Firewall and network switch system capacity was increased in all data centers to avoid bottlenecks. The IT department fortified and expanded its single sign-on systems to facilitate anticipated application access demand. This facilitates access for resellers without delay.
Mainframes were set up to deal with up to 4x the capacity from an earlier launch, if needed. More DB2 subsystems were implemented to allow 6-way data sharing, and high-intensity workloads were moved to isolated partitions.
Overall data center capacity for DCL infrastructure consumed nearly 465 KW of power. Such was the extent of the build out that launch hardware represented about 50% of their total power growth from the prior year.
The build out and upgrades in preparation for the iPhone release were completed on time — within a 60-day time frame. “We were ready for the launch well in advance,” he said.
The Business Side of the Launch
“In order to avoid long lines outside of retail outlets, we provided our customers with the opportunity to pre-order the device in advance of the launch day,” he said. “Customers would even receive them just before they were available in our stores.” Both the business and IT readied themselves for the results, feedback and sales figures.
The results were beyond their expectations. More than 60% of iPhone sales took place online, amounting to 100 times the previous online sales volumes. “Due to the expertise of our capacity planners,” he said, “we accomplished these record numbers without a performance issue or outage.
“Every single system performed within planned thresholds. No performance issues were encountered in our stores, online or by our customers when they activated their new phones. We ended up selling more phones to existing customers in the first 2 hours of the launch than any previous single day in Verizon Wireless history.”
“We avoided the long lines that plagued other device launches while demonstrating loyalty and superior service to our installed base,” he said. Customer satisfaction was high, and continues to be, for existing and new customers. He said they saw strong sales figures and comments from media and analysts.
Despite the massive ramp up in service delivery, the team provided customers with the network they wanted. Call failure during the launch was less than half of 1% and lower in larger markets such as San Francisco, New York and Chicago.
“We attribute our success to adherence to our core principles of focusing on the customer and delivering value to the bottom line — and to our modeling and capacity planning capabilities,” he said. “In fact, we use capacity management software to prove a need before we spend and to predict peak demands going forward.”
If you like this article, sign up to receive the monthly newsletter for more information about IT successes and how you can build credibility with the business through improved performance and capacity related activities.