Network convergence is increasingly becoming a reality for IT departments as IT leaders are looking to get the best performance and efficiency out of their network environments. Johan Ragmo, Director of Data Solutions for North Europe at Alcatel-Lucent Enterprise, explains the factors driving network convergence and examines why 2012 will be the year of the Ethernet fabric.
Network convergence – the merging of data and storage networks onto a common Ethernet infrastructure – has, in the past, often been a theoretical curiosity rather than deployment reality for a lot of IT departments.
However the benefits of convergence lie in simplicity and efficiency: having just one network to build, maintain and manage is ultimately better than two. Whilst one of the key drivers in moving to a converged network is cost, most organisations that are pursuing such technology are also doing so in order to improve network reliability, performance and agility.
Enterprises are increasingly having to provide high quality application delivery that can adjust according to the user, application and device in use across the enterprise. This means that they are starting to look to virtualised networks which can improve productivity and significantly increase network efficiency.
Ethernet enables Virtualisation
So there is a huge shift in the market towards virtualisation, and in order to enable this within the enterprise, IT departments must look at their fabric first.
They need a fabric which is designed to be resilient and easy to operate – a low-latency architecture, such as the Ethernet fabric. This way, data centre networks can adapt to the higher bandwidth requirements of media-rich applications such as video while supporting server and desktop virtualisation, and take the next important steps in providing true application fluency across the network.
According to a recent report1, server virtualisation has dominated the strategies of most IT organisations over the past few years and is particularly common among those organisations with converged network implementations or plans – it found that 79% of this group had already deployed virtual servers in production or in test labs.
And enabling a virtualised IT infrastructure requires a new approach to networking, which differs from the traditional strategy of simply throwing bandwidth at the problem. IT departments instead need to opt for a flatter, converged Ethernet fabric which can deal with the increasing array of devices as they appear on the network.
Ethernet vs. the Fibre Channel
Having disparate data and storage networks, as with the Fibre Channel approach, creates both equipment and administrative redundancy: with two sets of switching hardware, two sets of network management software and two sets of network administrators. These inefficiencies can be eliminated by consolidating data and storage networks onto a single Ethernet backbone.
It is true that the Fibre Channel has previously been extremely important for storage across the data centre, but it’s vital that we can now open up Fibre Channel over Ethernet – and this means more performance across the enterprise.
10 – 40Gig Ethernet
Performance plays a huge part in the Ethernet fabric. Now that the era of 10Gig Ethernet is reaching middle age, with upgrades to 40Gig becoming common on the latest servers, there’s plenty of bandwidth to go around – and the standards and technology path is already laid out for that capacity to only increase over the next few years.
According to the 3 Paths to Network Convergence report, when asked whether their deployment of Fibre Channel over Ethernet and 10Gig Ethernet allowed them to eliminate the Fibre Channel from their environment, a huge 71% of respondents in the research said that they had either eliminated Fibre Channel or were planning to eliminate it.
And when questioned about the factors impacting their move to 10Gig Ethernet, respondents highlighted ‘better performance for network-based backup or data replication’ and ‘faster inter-server communications for other applications’ as the most important factors.
So performance plays a huge part in the Ethernet fabric. Alcatel-Lucent Enterprise decided not to invest in Fibre Channel technology, instead developing its 40GigE and Pod and Mesh data centre technology, which enables enterprises to build just one infrastructure for storage and communication and can start to evolve Fibre Channel over Ethernet. As the industry increasingly upgrades from 1Gig to 10Gig and from 10Gig to 40Gig, networks need improved performance, and this switching technology enables just that.
By extending high-performance 40Gig Ethernet links throughout the Mesh, the speed, responsiveness and flexibility of enterprise data centre networks can be significantly increased, making them better able to support bandwidth-hungry applications such as video collaboration, virtual desktop and cloud services. This move, coupled with the use of Shortest Path Bridging, enables enterprises to essentially transform data centre networks into private clouds, giving them the flexibility to reconfigure data centre resources quickly and easily.
Just in Time Data Centre Performance
By implementing 40GigE technology, enterprises can achieve additional benefits by being able to manage the traffic going into the data centre by analysing the performance of the pod within the fabric. Enterprises are able to monitor the performance of the network and can analyse at what stage virtual servers might need additional performance, enabling organisations to plan upgrades and move virtual applications.
It is also possible to predict the performance of the network and monitor pressure on the switches based on which applications are being used, such as video, instant messaging, unified communications etc. and prioritise accordingly. This enables enterprises to offer JIT performance benefits across its applications delivery.
How does this work in practice? Take a customer service agent on a call with a customer where there is a delay in finding the correct information due to network performance issues. This could have a detrimental effect on the customer experience. If the enterprise implements a ‘Just in Time’ approach to the data centre, applications can be prioritised easily and appropriately.
Low-latency, loss-less fabric
One of the main drivers for network convergence and the implementation of an Ethernet fabric is that of cost savings. The key question IT managers need to be asking themselves is does their fabric currently have the right latency to enable virtualisation?
By implementing a low-latency, loss-less Ethernet fabric, enterprises can take significantly increase network performance, prioritise applications based on user, device and task, as well as implementing virtualisation across the network.