Cloud Connectivity for a multi-cloud strategy with IaaS in 2021, you have probably asked, how to connect AWS to Azure or Google Cloud Compute (GCP) or Oracle Cloud Infrastructure (OCI)?  Many cloud teams are looking to diversify their IaaS strategy with AWS utilizing additional services with other clouds.  

Connecting AWS to other clouds is the more common ask, mostly because organizations have deployed some form it in their IaaS posture.  We have seen an uptick in AWS to OCI for a Database tier as another common topology since Oracle has delivered a solid option in the marketplace especially for organizations moving off premise.  

You do not have to be a networking expert with Cisco certifications to design the most optimal and fault-tolerant topology.  However, connecting AWS Direct Connect to support this multi-cloud strategy must be reviewed with deep understanding of networking from Layer 1-3.  We have a good video on Multi-cloud connectivity to AWS, GCP, OCI, IBM and Azure and how the global internet backbone supports this new endeavor.

We’ve pulled out some quick tips we begin conversations with customers to narrow down how to connect multi-cloud which should help you.

Tip #1:  Cloud Egress

Know your egress charges when leaving a cloud.  Some cloud providers charge customers depending on the method of taking traffic off your workload environment to another cloud.  AWS Direct Connect has two billing elements: Port Hours and Outbound Data Transfer. Port hour pricing is determined by connection type – Dedicated Connection or Hosted Connection – and capacity. Data transfer out over AWS Direct Connect is charged per GB.

AWS Direct Connect for example has two billing components as the Data transfer leaving AWS is charged per Gig:

  1. Outbound Data Transfer
  2. Port Hours

How does AWS charge for egress on AWS Direct Connect?  

AWS Pricing across Direct & Hosted options is updated often, and it will depend on your uses case.  Search here for a full list of AWS Direct Connect Pricing, however we’ve broken a quick snapshot below

We have summarized a Dedicated Connection port hour pricing is consistent across all AWS Direct Connect locations globally with the exception of Japan from their website. The table below lists the port hour price by Dedicated Connection capacity selected. 

Capacity Port-Hour rate (All AWS Direct Connect locations except in Japan) Port-hour rate in Japan
1G $0.30/hour $0.285/hour
10G $2.25/hour $2.142/hour
100G $22.50/hour $22.50/hour

AWS Direct Connect Hosted Connections

Hosted Connection port hour pricing is consistent across all AWS Direct Connect locations globally with the exception of Japan. The table below lists the port hour price by Hosted Connection capacity selected.

Capacity Port-Hour rate (All AWS Direct Connect locations except in Japan) Port-hour rate in Japan
50M $0.03/hour $0.029/hour
100M $0.06/hour $0.057/hour
200M $0.08/hour $0.076/hour
300M $0.12/hour $0.114/hour
400M $0.16/hour $0.152/hour
500M $0.20/hour $0.190/hour
1G* $0.33/hour $0.314/hour
2G* $0.66/hour $0.627/hour
5G* $1.65/hour $1.568/hour
10G* $2.48/hour $2.361/hour


Tipe #2 – Colo to the Rescue:

Many organizations do not realize they can take advantage of standard colocation providers to connect to cloud environments.  These options include but not limited to OCI, Azure, IBM, AWS, and several common SaaS providers such as Cloudflare, SalesForce and other.  However, when designing which colocation provider offers the best connection, you must determine if you are required to collocate any equipment to qualify.  Some providers say yes, others say no and simply want to become that NSP of the future.  


Tip #3 – Dedicated vs. Hosted Connections

Options exist to connect and route traffic into cloud providers with a dedicated connection, unshared to the mass of other potential customers and a hosted, shared access option.  Neither is terrible, again it all depends on the requirement your organization has.  Do you need to connect to a SaaS provider for test/dev reasons with minimal bandwidth?  Perhaps a hosted connection with a small Ethernet Virtual Circuit (VC) could suffice?  Frost & Sullivan has good article about the genesis of this on Cission PR Newswire to understand where the industry is heading.


Tip #4 – Cost

You’ve built a multi-cloud environment, now how much does it cost to connect both IaaS environments?  In general terms, this is a fairly low-cost solution on a monthly basis.  We always begin with a 1-week trial option which we can design without costs.  Then, if it’s a software-defined option, usually costs can be by the hour depending on the provider(s) chosen.  We have learned after hundreds of deployments, the benchmark cost and free trial options which exist to test application load between clouds.  This is very important for latency, jitter, packet loss and ease of troubleshooting to select the right option.  We stay unbiased in the vendor evaluation process but biased in the right technology to support the design.


Tip #5 – Security considerations

Since Covid hit, organizations had to change their networking strategy overnight to support remote workers.  As a result, cloud connectivity took center stage with a larger emphasis compared to a traditional Wide Area Network(WAN).  However, work from home broadband connections to multi-cloud meant security (not that old Linksys router) needed to fill the void.  Enter SASE & Zero Trust Networking (ZTNA) to support these environments.  Replacing traditional VPNs with an agentless option to securely work in your multi-cloud environments is something you will need to consider. posted a quick article on multi-cloud SDWAN which outlines using various options across your WAN to do so.  

Press the “easy button” to secure multi-cloud connectivity


Learn more about how to connect to Azure Express and Oracle Cloud Fast Connect!

Azure Express Route

Oracle Cloud Fast Connect

Which cloud connectivity option do we recommend in 2021?  Easy, the only one which is suited for your business use case!  After hundreds of discussions, designs, and implementations we’ve found not every topology is the same.  We’ve approached this challenge by revieing where your workload instances are deployed, how users must access them. Then we build a review/design starting with layer 1-3 options for your business applications.  Just because a peer uses a brand of tech does not necessarily equate it’s the best option due to latency, security, costs, and convergence for your applications.  Ask us how we can show you architecture options free of a vendor bias.  

If you need to speak with someone on the engineering of your multi-cloud strategy, please contact us.  We can design something in an hour and provide a playbook for your architecture today and where you look to deploy in the future.  We have designed hundreds of environments all with Software Defined, physical X-connects, dedicated lines, colocation edge options and more support multi-cloud for the largest organizations in the world.