The virtualised data centre of the future

Enterprise data centres are being transformed into strategic assets of the future, courtesy of the virtualised network architecture.

Virtual data centres will be based on networks that can monitor, anticipate, and detect congestion – then dynamically adjust bandwidth and fabric resources to meet application service levels. As a result, fabric intelligence can monitor resource usage at the application level rather than just at the port level.

As IT organisations increasingly deploy virtual servers and storage to consolidate resources and improve utilisation, they must be able to dynamically provision applications on virtual servers and move them non-disruptively as workloads change.

However, this increased mobility of application workloads can sometimes create unpredictable congestion in the networks connecting server-to-storage traffic, server-to- server clusters, and storage-to-storage replication.

An innovative approach to networking helps simplify fabric management and improve resource utilisation while enabling superior performance, higher availability, and non-disruptive scalability – critical requirements for virtual data centre environments. It can assist reaching goals that include:

• Extending physical connectivity to devices using intelligent virtual connections to pools of shared resources
• Implementing virtual partitions to provide flexible management
• Integrating application service levels within the fabric for more flexible provisioning, higher resource utilisation, and adaptive networking applications
• Extending data management policies into the fabric to unify, control, and scale data mobility applications (such as replication, migration, and copy)
• Consolidating networks and protocols onto a single fabric capable of integrating emerging technologies such as Fibre Channel over Ethernet (FCoE) and Data Centre Ethernet (DCE)

This adaptive networking approach can support each of these goals and simplify key tasks such as provisioning, configuration management, capacity planning, security management, and fault isolation for evolving virtual data centres. The ultimate benefits of this approach are reduced management, lower operating costs, and a dynamic infrastructure that can evolve at the same speed as the business.

Application-aware fabrics needed for the future data centre

Replacing static, dedicated connections between physical servers and storage with dynamic connections between virtual machines and storage can create unexpected congestion in the fabric. To address this challenge, the data centre fabric must be able to assign application workloads to an appropriate service level and flexibly allocate resources as these workloads move within the server resource pool.

Virtual channel technology can create virtual connections within physical links that logically connect virtual servers to virtual storage.

Sets of virtual channels are assigned a specific fabric Quality of Service (QoS) priority. Fabric bandwidth and resources are assigned to each of three fabric QoS priorities (low, medium, and high), helping to ensure consistent delivery as workloads vary. Fabric QoS improves network utilisation, simplifying virtual server and storage provisioning, and also reduces network configuration tasks as application workloads move between virtual servers.

Extending intelligence into the data centre

Adaptive networking technology can extend fabric intelligence to the application, enabling fabric-wide application service level monitoring and management that automatically reacts to changes in virtual server workloads.

This approach enables the fabric to dynamically allocate shared resources as changes occur in the data flows between virtual servers and virtual storage. If congestion occurs (or is predicted), the fabric can adjust bandwidth and other resources according to defined service levels – helping to ensure that higher-priority workloads receive the resources they need.

“Adaptive networking technology can extend fabric intelligence to the application, enabling fabric-wide application service level monitoring and management”

The adaptive networking approach for data centres introduces new services including:

• Fabric QoS: Enables granular allocation of fabric resources to applications based on the relative importance of the application as defined by the assigned QoS.
• Traffic management services.
• Fabric dynamic profiling services: Provide end-to-end analysis of individual application data flows and resource usage.
• Policy management services: Prevent buffer credit exhaustion (buffer credits provide fabric flow control) and detect under-utilised shared physical resources, reclaiming them or reallocating them to optimise application flow according to defined policies.

As more of the world’s data centres become virtualised, today’s networks must be able to handle the additional workload.

Virtualisation will help organisations create more dynamic application workloads, and fabric intelligence must provide more in-depth knowledge about the state of fabric-wide resource utilisation. It must also be able to associate resource usage with application workloads (data flows) across all virtual devices to meet the data centre needs of the future. – Graham Schultz, Brocade

Image courtesy Bob Mical under CC

https://www.flickr.com/photos/small_realm/11189803153

https://creativecommons.org/licenses/by/2.0/