We recently posted a blog about the merits of using private connections to public clouds. We’re going to follow that up by encouraging you to read Willam Norton’s recent article about the same subject.
While the Internet (public transport) is cheap and ubiquitous, it has a number of potential downsides. Access products like SD-WAN attempt to mitigate some of these, like encrypting to overcome the lack of security and suggesting you use multiple provider solutions and embedded performance measurement to route around variability in performance or availability issues. These mitigations have their own limitations, and none of them address the financial consequences of “cheap and ubiquitous” access.
As more and more enterprises adjust their WAN architectures to the reality of pervasive hybrid work, cloud-delivered IT and digitization, the “ubiquity” of the internet will still have a role in the low-end branch and work-from-home location connectivity-to-cloud scenarios. Here the utility of “ubiquity” may compensate for the long term cost in application performance and cloud egress data fees.
Enter the higher utilization cloud core: a rapidly evolving enterprise hybrid cloud architecture with edge clouds that may exist in large enterprise production facilities interconnected with private cloud in multiple DCO provider footprints and public cloud resources from more than one source – that operates on “industrial” time scales (around the clock). Here, the egress data costs of public transport quickly dwarf any “ubiquity” benefit and a majority of the potential locations involved are potentially served by multiple private connectivity providers.
As we pointed out in the prior blog, the calculations of cost when comparing public to private transport options can be complex and require a lot of reasoning about your use cases and architecture, both near and long term. Norton’s voice has added to that conversation, providing another attempt to cut through the confusion to deliver a detailed analysis with similar findings.
The simple upshot?
“The paper’s summary is that it is provably less expensive to interact with your corporate cloud presence over a private network when you have more than 25Mbps of traffic. Beyond this, the cost of the private network is entirely paid for by the cost savings from the reduction in cloud Internet egress fees.”
Norton’s analysis options are pared down (NAT, optimal routing, etc.) to declutter the argument. The result is, again, that It only takes a very low amount of sustained network utilization to reach the breakeven point of using private connections to public clouds (with all the attendant benefits).
Our own analysis had postulated a break-even for a majority of the permutations of public access to public cloud between 10-20% utilization during an “office-month” usage period. That’s between 25 and 50Mbps of average utilization on a 1Gbps link.
Regardless of whose analysis you use or which one clarifies the argument best, the differences in result are insignificant and the point is the same: using private transport to access the public cloud has a really compelling business case that you should consider.
Schedule a call with our team here to learn more.