David Vellante with Mark O’Gara and Daryl Molitor.
This is a summary of a discussion Wikibon led in collaboration with On Magazine, EMC's CIO publication.
This is Part one in a series of three.
- See Part two Future Proofing Converged Architectures
- See Part three The Business Impact of Convergence
Increased pressure from business lines will require IT management to build infrastructure that is simplified, more cost effective, agile and highly available. This is the conclusion of two experienced practitioners, Mark O’Gara, Vice President of Infrastructure at Highmark Blue Cross Blue Shield of Pennsylvania, and Daryl Molitor, Senior Architect at retailer JCPenney who spoke on a recent On Magazine-produced Webcast that is now available on demand.
Highmark serves 28 million customers and generates $12B in annual revenue, and Plano, Texas-based JCPenney is a $20B company with about 135,000 employees and 1,200 stores. While these organizations participate in completely different industries, they both use a combination of mainframe and open systems and, together, manage multiple petabytes of online storage.
Both of these professionals strike a similar chord when talking about infrastructure. Specifically, the following three themes emerged from the call:
- Virtualization of server and storage resources is a mainspring of efficiency, and both firms are driving toward higher levels of abstraction to cut costs and simplify IT.
- Business lines increasingly demand high availability and extreme agility from IT, pushing organizations to deliver mainframe-like recovery on open platforms (e.g. Windows, Unix, Linux).
- In a further effort to simplify, organizations are planning for substantial changes to fundamental networks, specifically moving toward 10 Gigabit Ethernet and Fibre Channel over Ethernet (FCoE). However this transition will occur slowly and require planning actions that organizations need to consider.
Virtualization
The importance of virtualization at these and other organizations is clear. Both JCPenney and Highmark are implementing virtualization wherever possible to create what O’Gara calls “interlocking capabilities” across IT applications. By this, O’Gara means introducing higher levels of abstraction to achieve common goals for disparate infrastructure (e.g. interoperability, flexibility and manageability).
Molitor underscores the importance of having common goals across the organization, stressing; “You can’t have individual teams or groups going down the wrong path. It [virtualization] has to start from the top and be a common goal that everybody’s moving towards.”
O’Gara and Molitor agree that IT practitioners should try to understand efficiencies for the different parts of the infrastructure (e.g. servers, storage, networks) and push hard where there is opportunity for improvement. O’Gara further advises that small teams can innovate better than larger ones, and they should be encouraged to try different virtualization methods and processes. Examples include applying mainframe-like methodologies to a UNIX platform to increase efficiencies and re-architecting backup to accommodate virtualized infrastructure.
A Service vs. Application Orientation
The economic downturn is forcing businesses to be more responsive, and a “do more with less” mantra has been the theme of 2009. But for JCPenney and Highmark the drivers are different. As a retailer, JCPenney was hit hard by the downturn, whereas the healthcare industry has been somewhat insulated from the economic challenges facing firms in other industries. However, looming healthcare reform is forcing changes to which firms such as Highmark must respond. As well, the two firms’ approaches to delivering flexible infrastructure are quite different.
Specifically, JCPenney is seeing demands for high availability moving beyond mission-critical applications into business-critical and more mainstream back-office systems. The company is moving toward greater resiliency by designing availability into the application layer. Going forward, this will simplify infrastructure by building redundancy directly into applications as opposed to having redundant hardware across the organization.
Highmark, on the other hand, is approaching its main challenges of making applications more facile and flexible by modernizing legacy mainframe-based code, making it more open and service-oriented. Maintaining mainframe-class performance and availability in a distributed world is a challenge, especially from a skills perspective. Highmark’s methodology is to create re-usable Web services that can be invoked by applications as required. The key benefits of this services-oriented approach are that infrastructure is simplified and Web services can be re-used to support multiple applications.
Network Convergence
The past decade has seen steady movement toward data-center consolidation, and virtualization has furthered this theme, allowing a separation of physical and logical resources. In 2009, thanks to the drive to do more with less, we are seeing consolidation in the form of virtualization being extended to more applications. Furthermore, virtualization supports the next wave of consolidation, namely network convergence, which is coming about as a result of the ubiquity of high-speed networks and the maturity of network protocols and in particular 10 Gigabit Ethernet. Cisco’s UCS announcement has also heightened awareness of the potential to converge networks that support separate client, storage, and compute resources.
Why does IT management want to further consolidate and combine different types of network resources? The answer is to simplify infrastructure in support of a much more capable and network-centric computing environment across the broader Internet. Because today’s networks are built with unique infrastructure (i.e. adapters/switches/cables) and management systems, they are complicated and costly to maintain. By combining networks into a single ‘pipe,’ IT can dramatically cut management costs and more easily enable connections across networks and importantly, to external networks including the cloud.
Two key discussion points between O’Gara and Molitor focused on the evolutionary nature of the converged network and the organizational aspects of converged networks.
The conclusion of the practitioners is that convergence will not occur overnight but will evolve slowly, driven by the need for high-bandwidth 10Gig Ethernet in the data center. Also, certain organizational issues remain to be reconciled. Storage and networking professionals in many large organizations report to different parts of the organization. CIOs must rationalize the reporting structure and set common goals for cross-functional teams, such that these functions balance the flexibility requirements of the business with the mandate of end-to-end quality-of-service (QoS) typified by storage networks.
Action Item: IT organizations (ITOs) need to simplify infrastructure to support the demand for agility from business lines. Two viable approaches to managing complexity are to build resiliency into applications or take more of a services-oriented approach, allowing applications to access a menu of services as required. Both methods are effective. However, each brings certain trade-offs that ITOs must consider before moving toward wide-scale adoption.
Footnotes: