Before and After We Adopted Server Virtualization
This year Harley-Davidson celebrates 115 years as an iconic U.S. manufacturing company. Through these 115 years, we have continuously evolved to deliver premium experiences that fulfill our customers’ dreams of personal freedom. Harley-Davidson is a global manufacturer, a robust retail and experience business, and a financial services organization that helps dealers and customers through funding and insurance. Our company’s10-year objectives to grow 2 million U.S. riders, grow international business to 50% of annual volume and launch 100 new high-impact motorcycles require an IT organization that adapts quickly to a changing business landscape and enables robust digital experiences while overcoming the limitations of our legacy workloads.
These10-year objectives required a transformation in our IT operational efficiencies. As we looked at the industry trends – infrastructure-as-a-service, software defined infrastructure, software-as-a-service and the increased use of digital platforms – we decided that the virtualization of our application and infrastructure stack was a must. We set goals to have higher system availability, simple disaster recovery capabilities, increased operation efficiencies, and cloud-ready infrastructure. Over the last several years, we have accomplished these goals and more by moving 95% of our x86 server workloads to virtualized environments using VMware, Inc.
Higher System Availability - Virtualization allows Harley-Davidson to eliminate hardware failure impacts on service recovery as we have a duplicate host ready to accept the workload. In the physical world, our recovery times for critical services for our dealers, customers or employees could include two to four hours for parts replacement, or would require implementation of complex and costly replication solutions. Virtualization has also given Harley the capability to extend the life of many of our aging assets. We were able to virtualize our older legacy workloads to run on the latest hardware without changing functionality. This virtual-to-physical conversion takes place frequently on our interconnected shop floor between manufacturing equipment, tools, applications and servers.
Simple Disaster Recovery– Image-based disaster recovery makes backups almost instantaneous from the virtual machine perspective, and with stretch clustering we can failover virtual machines between datacenters instantly. This level of “live-DR” was not possible in the past and instead required at least a reboot to failover to the DR site. This capability enables us to perform DR tests more frequently with higher success. We are also able to restore clones from the image backup for application and performance testing and use these backup images if we have a failure during regular maintenance to minimize any impact to our business.
Increased Operational Efficiencies–The move to virtualization has provided several operational efficiencies for our global IS organization. The most visible is the reduction in our physical data center. We achieve a 50:1 consolidation ratio of VMs to hosts, which means we technically have a datacenter 50 times smaller and consuming 50 times less power than it would if all servers were physical. This not only equates to space, but helps us avoid costly datacenter investments as well. We’ve also been able to keep our environments more current with less effort and disruption. Before virtualization, we updated the firmware, driver and OS on each server monthly. This required an hour per server across nearly 700 servers, in addition to the required testing. After virtualization, we update the host firmware on our large frames once or twice a year, with only OS updates occurring monthly. The operational efficiencies have improved more than our datacenter operations; they have also allowed for projects to request a virtual machine via our IT Service Catalogue, and once approved, the automated provisioning is completed within minutes on an available host. Most recently, our product development team– which uses extremely expensive graphic-intensive workstations that cost as much as a new sedan – was able to enjoy better performance that can scale as their projects do by virtualizing their environment rather than purchasing new physical workstations. The VMware horizon solution is easy to enable and manage while the session compression keeps WAN links from saturation.
Cloud-Ready Infrastructure- Being 95% virtualized in our x86 stack has made those workloads cloud ready. As more cloud providers adopt VMware capabilities, we have the opportunity to move workloads seamlessly from on-premise to cloud. This accelerator of cloud-based IAAS is how we will further transform our services to be more scalable, flexible and cost competitive as we support our changing dealer, customer and employee needs. Technology is changing at such a rapid pace that it’s not feasible to predict our business needs three and four years in advance to achieve the best ELA deal. With cloud services, our global IT will scale and adapt with our company rather than retaining and being constrained by underutilized assets.
We look forward to continuing our relationship with VMware and Dell Technologies as we continue to evolve our infrastructure in support of our company’s mission to build the next generation of riders globally.