Hybrid cloud networking: achieving the best of both worlds?

The reality of cloud computing today is that we are in a phase of hybrid delivery. Enterprises need uninterrupted business continuity for critical applications delivered over their global networks, but face an explosion of bandwidth requirements...

Share

The reality of cloud computing today is that we are in a phase of hybrid delivery. Enterprises need uninterrupted business continuity for critical applications delivered over their global networks, but face an explosion of bandwidth requirements from less critical applications.

The majority of businesses won’t trust business critical applications such as SAP or EPOS to public cloud providers, but are prepared to move applications such as email, SharePoint, file servers and even VoIP to the public cloud. This situation can lead to a spaghetti junction of different network types, to support each model.

Multiple networks, delivering ever more applications, create the need for increased intelligence within the network. With traditional MPLS-based VPNs, performance is high and guaranteed by providers’ service level agreements (SLAs). There is often an instant feeling of confidence when we realise an MPLS network is being used to deliver an application. It’s fast, secure, tried and tested.

However, the Internet is deployed everywhere and provides large amounts of bandwidth at cost-effective prices, which is becoming increasingly attractive to enterprise CIOs who are consistently being asked to do more with less.

Internet networks are being deployed by enterprises today, but they don’t typically guarantee performance or availability. Companies are often presented with choices: either a fast or large network? Business-grade or standard-grade performance? Value-based or low-cost?

If you’re going to combine the cost advantages of internet networks and the reliability of private MPLS networks for core applications, you’ll need to take steps to unify the operation of those networks through Hybrid Network Unification technology.

To treat these networks as separate dumb pipes is a mistake that could impact application performance and undo a lot of the cost efficiency achieved by a hybrid strategy. If the two network types operate in isolation, it’s likely more bandwidth will be required to guarantee user experience and application performance, precisely the cost businesses were looking to reduce in the first place.

To avoid this situation, it’s becoming essential that enterprises build greater intelligence into their WANs so they are able to dynamically select the correct path for traffic and assign network availability appropriately for specific types of traffic, depending on its business priority.

This dynamic decision making needs to happen in real-time, based on a holistic view of demands being made across the entire network, from a variety of users. This ‘sense and respond’ approach means enterprises can work with hybrid networks and rest assured that traffic is flowing in the most appropriate way.

Posted by Thierry Grenot, CTO of Ipanema

Find your next job with computerworld UK jobs