- Béatrice PIQUER DURAND
- Ipanema Technologies
(Sponsor-Contributed Paper)

In today's increasingly mobile and connected world, IT departments are experiencing change like never before. Companies are tapping into the benefits brought by unified communications, personalized devices, SaaS applications, private and public cloud, social media, virtual desktops..., extending the requirements for hybrid networking where companies combine traditional MPLS networks with Internet accesses. Then comes the next question: how to select the right network, so business applications perform well, all the time, wherever? We touch again at a fundamental (should i say 'philosophical' ?) difference between policy-based and objective-based solutions.
Static or Policy-based Path Selection
- Manual: IT managers explicitly decide what are the primary and the back-up link for each application
- Semi-static: the decision is to use the primary link unless it does not work so back-up must be used
- Local: the decision applies whatever happens at the far end (i.e. independently from the impact on IT users)
- Coarse: the decision is taken for all connections of a same application
- Typical implementations: Cisco Policy Based Routing (PBR) or Riverbed Path Selection
Static WAN selection methods are renowned for their huge maintenance workload throughout the network life cycle. In addition, while static WAN selection methods provide acceptable availability guarantee thanks to their fallback behavior, they face strong limitations as a) they cannot optimize the usage over multiple lines (there is no link between the users' traffic demand and the WAN selection decision) and b) they cannot guarantee the impact on the end-to-end user experience as they just work from a local point of view.
Dynamic Path Selection
- Automatic: selection of the WAN does not require any manual configuration
- Dynamic: the WAN path is selected according to instantaneous situation (WAN access load, network quality, application performance, balance between applications, etc.)
- End-to-end: the WAN path is chosen according to what is happening end-to-end (locally, in the network and at the far end) to guarantee the business critical application performance
- Fine-grained: WAN selection is performed per user connection to guarantee the experience whatever the IT and network activity
- Typical implementation: Ipanema Dynamic WAN Selection based on Autonomic Networking and WAN governance principles.

Dynamic path selection methods are simple to operate as they automatically adapt to the constantly changing application traffic conditions. New applications, number of users, employees' real-time activity, application criticality, network performance: all these parameters evolve all the time, so does dynamic solutions.
Your enterprise has a dynamic and mixed application portfolio (like ERP, Unified Communications, VDI and cloud apps)? Aligning IT to business requirements matters for your organization and you need to enforce application SLAs over your global infrastructure? Look at dynamic solutions working at global level, implementing a closed-loop feedback mechanism that can guarantee the delivered quality across large and meshed networks. Look closer... you get Autonomic Networking.

Trending Discussions