Skip to main content
In this lesson we compare the ExpressRoute connectivity models and explain when to choose each. ExpressRoute delivers private, high-throughput connectivity between on-premises networks and Microsoft Azure. Different connectivity patterns trade off performance, security, management responsibility, and cost — choose the one that best matches your application SLAs and operational capabilities.

​
Service provider model

The service provider model uses a connectivity provider as the intermediary to bridge your network to Microsoft’s backbone. This model is flexible and commonly used by organizations that want managed connectivity options. There are three common patterns under this model:

​
Cloud exchange co-location

Place your equipment in the same co-location facility (cloud exchange) as the ExpressRoute provider. You typically establish a short physical cross-connect from your rack to the provider’s router; the provider carries traffic into Azure.
  • Typical scenario: Your servers live in a London data center that hosts an ExpressRoute provider (for example, Equinix or Megaport). A short cross-connect links you to the provider and then to Azure.
  • When to use: When you already operate in a co-location facility or need very low-latency, high-throughput connectivity for latency-sensitive workloads such as trading platforms.

​
Point-to-point Ethernet connection

A dedicated private Ethernet link connects your premises to the service provider’s network, which hands traffic off to Azure via ExpressRoute.
  • Typical scenario: A hospital or regulated enterprise orders a dedicated Ethernet private line from a telecom provider from their data center into the provider’s network, and onward to Azure.
  • When to use: When you require predictable performance, strong isolation, and strict security controls — common for regulated industries.

​
Any-to-any (IP VPN / MPLS) connection

Often implemented using the provider’s MPLS WAN or IP VPN service, any-to-any connectivity extends the provider’s WAN fabric to include Azure.
  • Typical scenario: A multinational retailer uses its MPLS WAN provider to give hundreds of branch stores and offices secure access to Azure without deploying many individual physical links.
  • When to use: When you want to extend Azure connectivity across a distributed site footprint with centralized WAN management by the provider.
Note: “Any-to-any” is typically delivered over provider-managed MPLS or IP VPN services. It’s a good fit when you want Azure connectivity integrated into an existing WAN fabric with centralized policy and routing.

​
Direct model (ExpressRoute Direct)

In the direct model you connect directly to Microsoft’s global network at an ExpressRoute Direct location, removing the service-provider intermediary. This option is aimed at organizations that need very high bandwidth and complete control.
  • Typical scenario: A global bank or large cloud operator installs physical connections at an ExpressRoute Direct site to achieve very high throughput (commonly 10 Gbps and up to 100 Gbps) and full control of traffic paths.
  • When to use: When you require the highest throughput, deterministic performance, and tight control over routing and traffic engineering. Note this option increases cost and operational responsibility.
Warning: ExpressRoute Direct provides extreme capacity and control but incurs higher cost and operational overhead. Verify your operations team has the skills and capacity planning required before adopting this model.

​
Quick comparison

Connectivity modelTypical bandwidthBest forKey advantagesKey trade-offs
Cloud exchange co-location1–10+ GbpsLow-latency workloads in co-loVery low latency; short physical path to providerRequires co-location presence
Point-to-point Ethernet50 Mbps–10 Gbps+Regulated enterprises, predictable workloadsDedicated, private link with predictable performanceHigher per-link cost
Any-to-any (IP VPN/MPLS)1 Mbps–10 Gbps (WAN-dependent)Distributed branch/retail footprintIntegrates with existing WAN; centralized managementDependent on provider WAN; variable latency
ExpressRoute Direct10–100 GbpsLarge throughput consumers, global operatorsHighest throughput and controlHighest cost; more operational responsibility

​
Design considerations

When choosing a connectivity model, consider the following factors:
  • Bandwidth needs: match throughput headroom to peak load and expected growth.
  • Latency and jitter: critical for real-time or financial applications; favor co-location or direct physical paths.
  • Redundancy and high availability: plan diverse physical connections, redundant routers, and failover designs.
  • Security and compliance: evaluate encryption, MPLS isolation, and where the provider’s security responsibilities end and yours begin.
  • Routing and traffic engineering: consider BGP configuration, prefix limits, and path control requirements.
  • Management and operational overhead: direct connections increase operational responsibility compared to provider-managed models.
  • Cost and procurement: factor in port fees, cross-connect charges, provider recurring fees, and long-term contracts.
  • Geographic proximity to ExpressRoute sites: selecting nearby exchange or Direct locations reduces latency and may lower cross-connect costs.
  • SLA and support: review provider SLAs and Microsoft’s ExpressRoute SLAs for availability and support expectations.
Now that you understand the models and the key design considerations, review your application SLAs, traffic patterns, and operational capabilities to choose the appropriate ExpressRoute approach.
Diagram titled "ExpressRoute Connectivity Models" showing Microsoft Azure clouds connected to on-premises sites via two approaches: a Service Provider Model with three connection types (cloud exchange co-location, point-to-point Ethernet, any-to-any IPVPN) and a Direct Model labeled ExpressRoute Direct to an ExpressRoute site.

Watch Video