Implementing a cloud computing architecture ultimately is the next generation in cost management -- a shifting
of traditional IT platforms to a resource-efficient, dynamic, hosted framework. Although this cost-based view of the cloud dominates the dialogue on cloud adoption, it falls short of the complete picture.
Beyond cost, line departments are looking to implement a cloud computing architecture to enhance worker productivity. To these departments, the biggest benefit of implementing a cloud computing model is the model's ability to apply IT tools faster and more flexibly. Line departments want flexibility in presenting application services to users and in assigning applications to resources based on cost and other metrics. This means looking at the overall IT architecture -- specifically the network -- in a whole new way.
Cloud computing model efficiency calls for flexibility in applications and resource pools
Both forms of flexibility require an evolution in two areas of technology. One is the evolution of applications to a simple application programming interface (API) integration model, where each application service is presented as a URL and composed into worker screens as needed. The other is the evolution of virtualization from static partitions in a data center to dynamic pools of resources that can be hosted anywhere and by anyone. In an effective cloud computing model, individual workers would get the data they need in an optimal form using Software-as-a-Service (SaaS)-like delivery, and each application would run where cost-benefit relationships are optimal.
Read more on cloud computing architecture
Networker's guide to private cloud orchestration
Nicira network virtualization architecture: The VMware of networking?
The application tier of a software-defined networking architecture
With such trends as the virtual desktop, browser-based access, bring-your-own-device (BYOD) programs and service-oriented architecture, or SOA, most applications can be served up with SaaS-like flexibility. At the resource level, flexibility isn't as simple, however. Resource flexibility places complex challenges on the network. The fact that the private cloud is likely to be hybridized with public-cloud hosting only expands the issue of dynamism.
As resource assignments to applications become dynamic, traditional construction and optimization strategies are challenged. We are just now learning how things will work with distributed responsibility involving cooperation between the data center and the outsourced cloud host. We're also just now learning how creating and sustaining the resource pool and keeping it connected to the users of applications will affect the network.
Three cloud computing architecture components: Access, resource pools and address mapping
All cloud computing models have three key components: access, resource pools and address mapping. The access component lets users connect with the applications they need. Resource pools support the servers and storage that users can draw on to run those applications. The address mapping component links elastic resource locations with such references as URLs; these allow users to access applications no matter where they run.
Read more on cloud computing architecture from
Agile cloud networks: Key to application access and performance
Five challenges in private cloud networking
Access networks typically are built on routing and VPNs. Resource pools typically are supported on data center networks built on the Ethernet and virtual local area networks, or VLANs. The technologies of these two areas will expand as the private cloud is built, but the real change will be in address mapping. This is what will connect users to the applications that now are running in a dynamic resource pool. Address mapping demands a level of network flexibility that's beyond the typical needs of static internal IT hosting or even Internet hosting. Without network flexibility, a cloud computing model's dynamism is lost.
In fact, it's the network that builds the cloud. Enterprise networks include data center LANs; storage area networks; Internet tunnels; and WANs built on switching, routing and a VPN or Virtual Private LAN Service. These network components are more costly than cloud software stacks, and making a mistake in the network part of cloud-building could be absolutely fatal to security and availability.
What we call the private cloud computing model is the one that will guide all future IT investment. It's the first model that recognizes the fusion of business and IT, public and private resources, networks and software. It's a model that's very different from the Internet or from the current enterprise data center, but it's also a model that can be built from current infrastructure components and can provide both immediate and sustainable benefits in IT return on investment and worker productivity.
Tom Nolle is president of CIMI Corporation, a strategic consulting firm specializing in telecommunications and data communications since 1982. He is the publisher of Netwatcher, a journal addressing advanced telecommunications strategy issues. Check out his SearchTelecom.com networking blog, Uncommon Wisdom.