The Benefits of the Private Cloud
January 4, 2011
The ever-increasing demand on insurance technology systems from multiple sales channels and processing operations makes the leveraging of cloud computing a logical choice for carriers. Businesses need to respond quickly to market demands and to scale resources up or down on demand, while providing customer access to those resources from anywhere at any time — all while reducing costs. In such a high-pressure, competitive marketplace, building availability, flexibility and agility into the IT infrastructure is key.
With these considerations in mind, Philadelphia Insurance Companies’ (more than $2 billion in 2009 gross written premium) IT department has developed a living road map for its private cloud initiative, the drivers of which are standardization, virtualization, distributed data management and automation into the data center.
Standardization facilitates the sharing and pooling of resources and enables faster response times, speed to market and business agility. As our infrastructure team prepared to move into a new data center in fall 2010, one of the main goals was to drive standardization in all aspects of the design architecture — server, network, storage, virtualization infrastructure, database, applications and vendors. Blade servers were chosen as the server architecture platform for their modularity, scalability and flexibility.
Today entire environments can be refreshed and made portable and available for use. Preloaded templates with the right amount of computer resources and application configuration are rapidly provisioned and deployed in a matter of minutes to respond to business needs.
Taking an evolutionary approach, the next phase was to inject automation into the data centers to meet service delivery, elasticity and scalability needs. Automated monitoring and provisioning methodologies currently are being developed to dynamically manage application delivery between data centers.
Instead of manual redirection of all traffic to a specific data center in the event of a disaster, the goal is to intelligently and automatically distribute inbound customer traffic requests to the data centers based on application performance, data center capacity and availability, or business policies. Portions of this multifaceted solution will go into production in Q1 2011.
In order to achieve dynamic scalability with continuous availability and elasticity, the Philadelphia Insurance IT organization has to radically change its way of thinking from a legacy, old-fashioned methodology to one that is suited to take advantage of the cloud. In a highly virtualized and distributed cloud environment, monolithic database servers and applications are difficult to provision, scale out and make portable. The result is decreased flexibility and agility for the business and limited or compromised failover or availability.
The new thinking calls for cloud-ready web applications and back-end databases that can be partitioned horizontally across cluster members. In that vein, the database infrastructure team will undertake in Q1 2011 a detailed architecture review of the current database design and requirements.
In a cloud computing environment with a requirement for continuously available data centers, a memory-oriented data management platform can provide clustering of the applications by dynamically replicating and partitioning data across multiple servers. The objective is to provide the same resiliency and availability at the software/application tier with dynamic scalability as the physical infrastructure.
Achieving this level of scale-out and availability combined with the mobility and portability offered by virtualization, application delivery options and automation effectively have made the concept of a dedicated disaster recovery site obsolete. It changes the paradigm from one focused on recovery to a concept focused on continuous availability and business continuity. Resources now can shift from one data center to another seamlessly without degradation of service levels or customer expectations.