At Cisco’s recent annual Partner Summit, the company announced a grandiose vision of a world of interconnected, federated clouds. The event kicked off on Monday, March 24th with Cisco announcing it would invest a billion dollars over the next two years to build a Global Intercloud. The concept behind Intercloud is that cloud will shift from being a bunch of random, individual cloud services to a federated, connected set of clouds — an “Internet of Clouds” may be the best way to think about it.
This would enable customers to move workloads, applications, and data between a private cloud and any number of public cloud services while maintaining security and management policies. If successful, Intercloud should create a “rising tide” that lifts the entire cloud industry as more businesses get comfortable with shifting part or all of their IT infrastructure to the cloud.
Cisco has a very comprehensive set of cloud services planned, including SAP HANA, collaboration, PaaS, IaaS, storage, energy management, security, network management, and other services that can be sold by its massive base of about 65,000 worldwide distributors. In essence, Intercloud opens the door for all of Cisco’s resellers to jump into the cloud game.
Let’s assume that Cisco is successful here and that businesses jump on board the bandwagon faster than ever before. It certainly sounds great on paper, but are there any unforeseen ‘gotchas’ that may come up along the way? One possible challenge for network managers is the impact to the enterprise WAN.
Current WAN design was never meant for the volumes of Internet traffic that could be created by something along the lines of Intercloud. A legacy WAN is designed for the majority of traffic to be of the client-server variety with traffic moving to and from the data center. Cloud, on the other hand, drives a significant amount of Internet traffic. With a traditional WAN, Internet traffic passes through a single ingress point, into the data center, and then to the branch. This is often known as a “trombone” effect, as Internet traffic effectively passes over the network twice. When Internet traffic is the minority of WAN utilization, this inefficiency of a traditional WAN isn’t ideal, but most network managers can live with it. However, the shift to cloud could make the overwhelming majority of traffic Internet based.
I believe the majority of companies will leverage some sort of hybrid cloud model where applications, data, and workloads are continually being moved, first from the company’s private data center to the cloud, then to a branch office, then back to the cloud. For the WAN, this means what was once a highly predictable set of traffic patterns moves to a chaotic environment that’s hard to manage and nearly impossible to bring any predictability to.
The title of this blog asks: Will the growth in the cloud break the enterprise WAN? In my opinion, the answer to this is a most resounding “yes!”, and network managers should look at re-architecting the WAN before the wave of cloud hits. As network managers go through the process of architecting the WAN for the cloud, the following should be considered:
- Direct Internet access for branch offices
- In-region cloud providers to move the data closer to the cloud service
- The use of lower-cost broadband for at least the backup connection, if not the primary as well
- Path selection technology to optimize the utilization of network connections
- WAN optimization in every location — consider a virtual product to keep costs down
- Network visibility tools so network managers can proactively manage the WAN links
The cloud is coming, and coming in a big way. I strongly urge companies to take a look at the WAN and get out in front of this trend instead of having to react to it.