When interconnecting cloud solutions and providers, you will often want to integrate and interconnect them via your private, wide-area network. This has the same challenges as internet-based interconnection; for example, there is no direct peering between the cloud providers. You cannot configure an Azure ExpressRoute circuit to connect to AWS Direct Connect, and vice versa. In addition to that, your network providers need to have peering with the cloud provider in the right location, and you need to be able to route between the providers. You need to create the interconnection by yourself, or along with your network provider. 

The two most common use cases in this field are as follows:

• In the first one, you have a Layer 3 network, like with MPLS, and you can directly interconnect to the cloud providers; the routing is done by you or your provider. You only need to choose the right interconnection locations, to keep the locations close together and optimize data transfer. 
• The second option is to own a Layer 2 network, where routing is done by you, and the network is basically hardwired from location to location. Here, you need an additional colocation, like a colocation at a cloud exchange hub, to exchange data. 

There are providers on the market, such as Megaport, that offer virtual cloud routers, enabling customers to build a direct interconnection by themselves, without any need for colocations or hardware. An implementation of Megaport is shown at https://www.megaport.com/megaport-cloud-router/.

For both scenarios, you will need an interconnection point. The architectural diagram might look as follows: