I have been working on rearchitecting our backup cloud connectivity and am considering using Megaport’s cloud router (MCR) product. I’ll post again in the future with more details of the design and its implementation, but I wanted to write a short note of appreciation about Megaport’s provisioning interface.

They provide a complete self-service portal and REST API. In addition, they provide a separate “staging” portal and API, where “all actions mirror the production system, but services will not be deployed and you will not be billed for any activity.”

Megaport Staging Portal

I think this is wonderful, and wish more providers offered this service. Before I purchased anything, I wanted to validate that my expected design would work the way I planned. I was able to register an account and test provisioning the services without having to engage a salesperson.

Note that there is a 24 hour delay before new accounts have access to staging. Also, every day everything in staging is clobbered and all services configured in the current production environment are synced to the staging environment.

For this project, I specifically wanted to use their terraform provider to build an MCR and Azure and AWS virtual cross connects (VXCs). This was as simple as supplying portal credentials and specifying the environment to use the staging API.

terraform {
  required_providers {
    megaport = {
      source  = "megaport/megaport"
      version = ">=0.2.5"

provider "megaport" {
  username = "myportalusername"
  password = "myportalpassword"
  accept_purchase_terms = true
  delete_ports          = true
  environment           = "staging"

Before creating the first resource, I found that I needed to enable for the market(s) where I was provisioning services, “USA Market” in this case.

Choose Billing Markets under the Company menu
Enable Markets before using Terraform

Choose Billing Markets under the Company menu

Instantiating the cloud router (MCR) involves specifying its location. The provider has a location data source to lookup by name.

data "megaport_location" "datacenter" {
  name    = "Equinix DC4"
  has_mcr = true

resource "megaport_mcr" "mcr" {
  mcr_name    = "ashburn-mcr"
  location_id = data.megaport_location.datacenter.id

  router {
    port_speed    = 1000
    requested_asn = 133937

The cloud VXCs can be configured using megaport_aws_connection and megaport_azure_connection resources. There is a partner port data source that allows lookup of each by company and name.

data "megaport_partner_port" "aws_port" {
  connect_type = "AWS"
  company_name = "AWS"
  product_name = "US East (N. Virginia) (us-east-1)"
  location_id  = data.megaport_location.datacenter.id

resource "megaport_aws_connection" "aws_vxc" {
  vxc_name   = "AWS"
  rate_limit = 500

  a_end { port_id = megaport_mcr.mcr.id }

  csp_settings {
    requested_product_id = data.megaport_partner_port.aws_port.id
    amazon_asn           = 65301
    amazon_account       = "123456789012"
MCR with virtual cross-connects to AWS and Azure

This process provisions the MCR with default IPs and a IPv4 BGP session. I also verified (not shown here) how to use the a_end_mcr_configuration argument to specify from terraform custom IP addresses, multiple BGP sessions with individualized [advanced settings][bgp-advanced] for BFD, MED, and MD5 authentication in a dual-stack configuration.

Unfortunately, I can’t configure any physical connections using the staging API. Actual connections into our on-prem datacenters and the cloud providers will need to wait until I’m a paying customer. For now, this gave me the confidence to judge that I can integrate the provisioning of Megaport resources into our existing Terraform workflows.