Azure Service Endpoints versus Azure Private Links

Recently a lot of folks have been asking about Azure Service Endpoints and Azure Private Links — what’s the difference? when to use which? and why?

For starters, let’s review what is a Service Endpoint, and what is a Private Link? Followed by which solution is better to use, and why….

Azure Service Endpoints

With any Azure Virtual Network (VNet) you can leverage a ‘service endpoint’ that provides a secure connection and a direct connection to Microsoft Azure’s service over Microsoft’s backbone network infrastructure. The service endpoints allow you to run services/resources over the VNet and enables private IP Address within the VNet to communicate with the Azure service without the requirement of having a public IP on the VNet. Service Endpoints work by enabling your VNet or subnet(s) to support the Service Endpoint, and once enabled, you can configure which PaaS resource(s) can accept traffic from those subnet(s)/VNets. There is no requirement to do any IP filtering and/or NAT translation, all you need to tell is the PaaS resource(s) which VNet/Subnet to allow traffic from. When Service Endpoints are enabled, the PaaS resource sees traffic coming from your VNet private IP, not the public IP.

Azure Private Link

Azure Private Link allows you to access Azure (PaaS) services, like Key Vault, Storage, Log Analytics, etc., over a private endpoint within your Azure VNet. The communication between the Private Link (endpoint) and your VNet continue to travel over the Microsoft’s backbone network, however your service is no longer exposed over the Internet. One drawback with Private Link is that to support resolution of the PaaS resources using the same name, you do need to implement DNS to resolve the private link zone for that resource. There is integration with Azure Private DNS to set this up for you, but this can be problematic if you have your DNS service already running, or do not want to use Azure Private DNS with your VNet. Once enabled, you have now granted access to a specific PaaS resource within your VNet. Meaning, you can control the egress to the PaaS resource. Unlike Service Endpoints, Private Link allows access from your on-premises infrastructure to Azure resources over an ExpressRoute circuit, or Site to Site VPN tunnel, or via its peered VNets.

What’s the difference, and when to use?
  • The biggest difference between Private Links and Service Endpoints, is Public IPs. With Private Link, there is never any Public IP created and traffic can never go through the Internet, whereas with Service Endpoints, you have the option to limit access.
  • Second key difference with Private Link is, once enabled, you have now granted access to a specific PaaS resource within your VNet. Meaning, you can control the egress to the PaaS resource.
  • Service Endpoints are much simpler to implement and significantly reduce the complexity of your VNet/Architecture design.
  • Private Link will always ensure traffic stays within your VNet.
  • Another key difference between Private Links and Service Endpoints, is cost. There is a $0 cost to implement Service Endpoints, as the cost is already integrated within the VNet cost itself. Whereas Private Links costs can quickly grow depending on the total ingress and egress traffic and the runtime of the link. For example, within Azure Canada Central, to have a Private Link that is available for 730 hours in a given month, and that allows 100TB of ingress and egress (for both) can run over $2,000 monthly.
    • This is something to factor when designing or implementing either solution, as Private Links will quickly add to your monthly spend.
  • Another consideration is, availability, meaning Service Endpoints and Private Links are not generally available for all services, for example. There is no Service Endpoint as of writing this post, for Azure Log Analytics. However, there is a solution for Private Links for Log Analytics. Both services are available but not for all resources/services. For the complete list you can visit the links below, Service Endpoints: HERE ; Private Link: HERE.

Ultimately, if you are considering either solution, Private Link versus Service Endpoint, then you are probably concerned with security and with that said, Private Link is superior to Service Endpoints. The services available to Private Link will continue to grow like Service Endpoints, but based on my observation, it appears Private Link has a much deeper portfolio with Azure services integration.

Azure Virtual Network (VNet) Peering

In this blog post, I will go over,

  • What is Azure VNet (Virtual Network) Peering,
  • When to use VNet Peering,
  • How to implement VNet Peering.

What is Azure Virtual Network (VNet) Peering?

Azure VNet (Virtual Network) Peering enables resources within two separate virtual networks to communicate with one another. Leveraging Microsoft’s backbone infrastructure, the two peered virtual networks will communicate over its own isolated network.

Below we have two Virtual Networks (VNet01 and 02), that have different IP Address spaces. By implementing VNet Peering, the two networks will be able to communicate with one another, as if all resources are in one network. Some notes, VNet Peering is not transitive, ie. If VNet01 and VNet02 are Peered, and VNet02 and VNet03 are Peered. This means, VNet01 and VNet03 cannot communicate with one another. Another note, inbound and outbound traffic in the VNet Peer are $0.01 per GB. Prices are a bit higher for Global VNet Peering. Get the official numbers here, https://azure.microsoft.com/en-us/pricing/details/virtual-network/.

When to use Azure Virtual Network Peering?

As mentioned above, you want to enable Azure VNet Peering when you have two virtual networks that have resources (VMs) in both networks that need to communicate with one another. For example, let’s say you have exhausted 4,000 VM limit within a VNet…

Some of the benefits of VNet Peering is:

Before you go ahead and implement, there are a few requirements:

Finally, how to implement it!

In this example, both of my virtual networks (VNets) are in the same region, Canada Central.

Select VNet01, and select Peering:


Give the Peering a name, “VNet01Peering” and select the other VNet, VNet02.


Give it a few seconds, and it should now be connected to VNet02:

Next, we now need to apply the same concepts to VNet02. So let’s do that now.



Now if we go to the VMs within each of the Virtual Networks, and try to ping another VM in the other VNet, it should now work! Based on the images below, you can see the Ping failed, that was from a previous ping response prior to VNet Peering being implemented.

VM01 in VNet01 trying to Ping VM02 in VNet02; -> ->

And conversely, the other way around…

VM02 in VNet02 trying to Ping VM01 in VNet01; -> -> : ->