Shedding the Silo Mentality

Isolated silos of technology may have cut the mustard back in the '90s data center, but not anymore

February 25, 2005

3 Min Read
Network Computing logo

Over the years, the use of data center applications and services has curved upwards at an ever increasing rate. Traditional data center networking infrastructure models were built in "silos" of applications and services (referred to as tower, cage, or cell network architecture), designed to be stable and to not react quickly to changing business strategies and processes. It was easy: Just add more hardware to solve immediate tactical issues and to support the implementation of a new service or application.

The concept of being able to turn on additional services without performance degradation or service disruption, the need for high availability, and the goals for meeting different lines of business or new services requirements, all worked very well together in the data center model of the late nineties. Part of the reason traditional data center systems were able to deliver predictable results was that the number of services were themselves predictable and infrastructure resources were relatively inexpensive (e.g., one application per system was acceptable because resources were relatively cheap).

However, in the last couple of years this approach has led to rapid consumption of precious data center rack space, complicated wiring infrastructures, major network scaleability and stability issues, and complex security as organizations cope with continual threats.

As data center managers continued purchasing isolated infrastructure (routers, switches, firewalls, servers, storage, etc.) to scale their networks silos, they spent most of their time trying to reduce operation and capital expenditures. They also had to resolve operational issues and find fixes for the same problem over and over again. Data center complexity, stability issues, and high total cost of ownership (TCO) became the data center manager’s constant companion and worst nightmare.

At the end of the day, the increasing number of applications/services and the stringent security/management requirements of next-generation services, mean that data center managers cannot go on creating an isolated infrastructure.So, how will they reduce complexity and lower costs, while optimizing the delivery of those applications and services that have the greatest impact on their organizations?

The new data center architecture must provide an infrastructure model that is open, flexible, pre-designed, standardized, virtualized, and highly automated, but at the same time highly secure and reliable. This model will provide the opportunity to implement advanced networking technology that enables virtualization through a Service Oriented Architecture (SOA) strategy. Specifically, the utility model stages are standardization, consolidation, integration, virtualization, and automation.

We’ve already seen server and storage consolidation and virtualization happening by combining multiple application and service environments into a single system by utilizing virtualization technology. In the future, this trend should be extended to the network; but how can this be achieved? By consolidating the network to provide a much higher degree of integration between all elements of the infrastructure (especially between network and security devices). Over time, this will include virtualization and automation of these infrastructures, leading to a service-oriented computing environment that shares application components across many service environments.


Are we moving from a data center architecture model that supports the silo infrastructure to a shared infrastructure model? The answer is no. We are moving from a physically segmented network architecture model to a logically segmented and integrated system-based approach to networking (with the ultimate goal of a virtual infrastructure). This will substantially reduce complexity and ease operation, while increasing resilience and functionality of the infrastructure.Data center managers should be aware that the migration to a full-utility networked data center model is incremental, and will require a service-oriented architecture strategy to achieve highly integrated infrastructure, implement dynamic resource sharing, and simplify management.

— Amr Ahmed, Internetworking Practice Director, and Vincent Hayden, Principal Engineer, Greenwich Technology Partners

Stay informed! Sign up to get expert advice and insight delivered direct to your inbox

You May Also Like

More Insights