Application Availability is the Real Value of Active-Active Architectures

Software defined WANs (SD-WANs) have gained market momentum so quickly because their value proposition is multi-faceted. Some enterprises have looked to SD-WAN as a way to dramatically lower network transport costs, while others are building SD-WANs to automate network operations. One of the more common use-cases I have seen is to shift toward an “active-active” architecture.

Active-Active WAN Architecture

Historically, WANs are built on the concept of “active-passive”, where a branch can be connected using two or more links, but only the primary link is active and passing traffic.  In this scenario, the backup connection only becomes active in the event the primary connection fails. While this might seem sensible, it’s highly inefficient as enterprises are paying for far more bandwidth than they are actually leveraging. This inefficient architectural design is driving increased interest in active-active configurations.

In actuality, the interest in active-active has always been there, but it was difficult to configure and expensive to implement. The emergence of SD-WAN technologies with multi-path capabilities combined with the ubiquity of broadband has made active-active highly attractive and something any business can deploy and manage quickly and easily.

However, businesses that are looking to transition to an active-active architecture should do so, not only for capacity purposes, but to improve application availability. Conceptually, improved capacity seems like the “low hanging fruit” for active-active configurations, but there are other, more cost-effective ways to increase network throughput. The best way to solve bandwidth capacity challenges is to optimize a single link with the use of compression or other WAN optimization technologies.

Managing Additional Complexity

Geographically distributed enterprises seeking to leverage a pure, dual broadband WAN for increasing bandwidth are not likely not realize the ROI they are seeking, as there are some hidden costs that weren’t there before. The first issue is that managing multiple broadband providers can be complex, tedious and time consuming. There is no single, global or even nationwide broadband provider, so large organizations will likely need to engage dozens of different service providers to fully address their connectivity requirements. Enterprises can engage bandwidth brokers or agents who offer services to manage this process, but that comes with added expense. Also, not all broadband types are available in all areas, so it’s likely the WAN will leverage a mix of DSL, cable, satellite and cellular (3G/4G) connectivity.

Using multiple broadband providers and having to “mix and match” transport types adds additional management overhead that must be taken into consideration when going through the planning and assessment phases of the deployment. Also, many wireless broadband connections come with monthly usage caps that limit the ability for enterprises to run applications across them continuously. Wireless networks are great for backup and replication, but don’t often make for great primary connections.

Another issue with active-active is that it can be difficult to configure the networks with path and infrastructure diversity without spending a bunch of money on build out costs to bring new networks into the building. In this case, having two connections from the same provider might protect against a CPE issue, but any disruption, brownout or outage on the service provider side of it will more than likely result in both connections going down simultaneously.

The Benefits of Active-Active Configurations

Considering the complexity and added overheard involved with active-active configurations, they still offer a strong use case to improve application availability in a hybrid configuration where MPLS is complimented with broadband. In this scenario, enterprises are wise to run their mission critical or real-time applications on MPLS and then offload all the best effort traffic to broadband. In the event of an MPLS failure, the critical traffic can fail over to the broadband link to eliminate downtime of important applications.

An alternative configuration would be to run all traffic on MPLS with traffic allocated by class of service to ensure the best possible performance for the critical and real time applications. In the event of a failure, the important applications must have the ability to quickly fail over to the second active connection without impacting the user experience.

And yet another alternative configuration would be to deploy an advanced SD-WAN solution that enables the creation of virtual overlays that bond multiple underlay links together. Virtual overlays enable enterprises to classify and categorize applications based on their unique service level requirements and provide fast-failover should an underlay link experience congestion or a brownout or outage.

Active-active has certainly become central to any discussion of SD-WAN, but large distributed enterprises need to make sure they are moving to this model for the right reason. While capacity optimization is certainly something businesses need to be concerned with, there are better ways to improve the efficiency of bandwidth. Ultimately the true value of active-active configurations is to ensure the availability of critical and real-time applications.