Is your Network Delivering the Performance your Business Relies on?
Read our latest ideas on trends affecting today’s communications network and how to ensure yours delivers all that your business requires.
The business issues we address
Your operational productivity and efficiency depends on a network that is always on and always accessible. We understand that networks must be flexible, tailored to meet your communication needs, compliant with regulation and corporate governance, and also aligned with industry standards. Is your network delivering the performance your business relies on?
• Does your network support your adoption of unified communications and collaboration and other businessenabling applications, to drive improved employee productivity?
• Is your network enabling your enterprise mobility strategy to drive workforce productivity through full access to all corporate resources, from any location?
• Does your network architecture support a virtualised operating environment, helping you to reduce IT footprint and drive efficiency?
• Are you managing your network assets to reduce operational costs and complexity, and improve processes for manageability and security?
• Do you have the right balance of sourcing options, giving you flexibility in procurement, investment and ongoing management of your network?
• Does your network infrastructure support your environmental sustainability initiatives, so reducing costs, driving efficiency and managing business risk?
Three network tiers, or two, or perhaps just one?
Traditional networks comprise three tiers (access, aggregation and core) and the
majority of organisations’ networks have been built in line with this principle. These designs were approapriate for traditional network traffic patterns, however, the introduction of virtualisation, automation and orchestration technologies has resulted in a significant increase in server to server and server to storage network traffic (east-west traffic). East-west traffic currently accounts for up to 75% of traffic running across a data centre network, much of it very sensitive to latency and poor performance. The more simple the network architecture is and the fewer tiers it comprises, the more efficiently it will cope with the increasing amounts of eastwest traffic. In addition, there are choices to be made between ‘top-of-rack’ (ToR) and ‘end-ofrow’ (EoR) network switching architectures. ToR architectures mean that switches are deployed into the top of each rack of the data centre, or specific section of data centre, while EoR switching architectures call for larger switches deployed at the end of each data centre row. ToR allows for far fewer long cable runs as servers are cabled inside the rack while EoR allows for fewer switches to be deployed. Both have their place and should be considered as part of the desired network architecture.
Networks that support virtualisation
Virtualisation technology is being adopted across the entire IT environment and the virtual machine (VM) has emerged as the new compute building block. Server virtualisation has a dramatic affect on the network:
• Having multiple VMs on each physical server increases server utilisation rates which in turn significantly increases the amount of traffic flowing to the network
• Applications running on VMs are mobile and can move from one physical server to another on-demand or when a fault is detected. To allow for this migration, virtualised systems need flat Layer 2 networks, where all the involved servers share a single broadcast domain and IP subnet.
• Additional VMs can be provisioned very quickly when the application needs more computing resources. The network and associated network services cannot be so quickly provisioned
• The network edge traditionally stops at the physical server network interface card, which means that the network does not have visibility of traffic flowing to and from the different VMs. This means that the network port and its associated security policy cannot be applied to a specific VM and therefore cannot move with that server during a VM migration. To resolve this requires extending the network edge into the hypervisor of the virtualised server. This ensures that the network has complete visibility of VM network traffic and that security and other network policy (quality of service, acceleration, etc) can be applied directly to VM traffic. This also allows end to end network management to be consolidated for improved management. The result is a network that supports each VM in the same way that is does a physical server.
Conclusion
The business and IT worlds are changing and driving the evolution of data centres and computing models, and the network, which is the foundation of the data centre, is a critical enabler of this evolution. There have been many innovations in networking technology and network architectures in recent years and organisations should consider working with partners who have significant depth of network and data centre knowledge to assist the to navigate their way forward. The result should be a solid data centre network platform in the data centre which will allow for the successful modernisation of data centres and the move to private and public cloud computing models.