Twin-region deployments utilizing Safe Digital WAN Hub with Routing-Intent with out World Attain

This text describes the very best practices for connectivity, visitors flows, and excessive availability of dual-region Azure VMware Answer when utilizing Azure Safe Digital WAN with Routing Intent. You study the design particulars of utilizing Safe Digital WAN with Routing-Intent, with out World Attain. This text breaks down Digital WAN with Routing Intent topology from the attitude of Azure VMware Answer non-public clouds, on-premises websites, and Azure native. The implementation and configuration of Safe Digital WAN with Routing Intent are past the scope and are not mentioned on this doc.

In areas with out World Attain assist or with a safety requirement to examine visitors between Azure VMware Answer and on-premises on the hub firewall, a assist ticket have to be opened to allow ExpressRoute to ExpressRoute transitivity for each regional hubs. ExpressRoute to ExpressRoute transitivity is not supported by default with Digital WAN. – see Transit connectivity between ExpressRoute circuits with routing intent 

Safe Digital WAN with Routing Intent is just supported with Digital WAN Normal SKU. Safe Digital WAN with Routing Intent gives the aptitude to ship all Web visitors and Non-public community visitors to a safety resolution like Azure Firewall, a third-party Community Digital Equipment (NVA), or SaaS resolution. Within the state of affairs, we’ve a community topology that spans two areas. There’s one Digital WAN with two Hubs, Hub1 and Hub2. Hub1 is in Area 1, and Hub2 is in Area 2. Every Hub has its personal occasion of Azure Firewall deployed(Hub 1 Firewall, Hub 2 Firewall), basically making them every Safe Digital WAN Hubs. Having Safe Digital WAN hubs is a technical prerequisite to Routing Intent. Safe Digital WAN Hub1 and Hub2 have Routing Intent enabled.

Every area additionally has an Azure VMware Answer Non-public Cloud and an Azure Digital Community. There’s additionally an on-premises web site connecting to each areas, which we overview in additional element later on this doc.

Understanding Topology Connectivity

The next sections cowl visitors flows and connectivity for Azure VMware Answer, on-premises, Azure Digital Networks, and the Web.

This part focuses on solely the Azure VMware Answer non-public clouds in each areas. Every Azure VMware Answer non-public cloud has an ExpressRoute connection to the hub (connections labeled as “D”).

With ExpressRoute to ExpressRoute transitivity enabled on the Safe Hub and Routing-Intent enabled, the Safe Hub sends the default RFC 1918 addresses (10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16) to each Azure VMware Answer non-public clouds over connection “D”. Along with the default RFC 1918 addresses, each Azure VMware Answer non-public clouds study extra particular routes from Azure Digital Networks and Networks (S2S VPN, P2S VPN, SDWAN) which can be linked to each Hub 1 and Hub 2. Each Azure VMware Answer non-public clouds do not study particular routes from on-premises networks. For routing visitors again to on-premises networks, it makes use of the default RFC 1918 addresses that it realized through connection “D” from its native regional hub. This visitors transits by the native regional Hub firewall, as proven within the diagram. The Hub firewall has the precise routes for on-premises networks and routes visitors towards the vacation spot over connection “E”. Visitors from each Azure VMware Answer non-public clouds, heading in the direction of Digital Networks, will transit the Hub firewall. For extra data, see the visitors stream part.

The diagram illustrates visitors flows from the attitude of the Azure VMware Answer Non-public Cloud Area 1 and Azure VMware Answer Non-public Cloud Area 2.

Visitors Circulation Chart

This part focuses solely on the on-premises web site. As proven within the diagram, the on-premises web site has an ExpressRoute connection to each Hub 1 and Hub 2 (connection labeled as “E”).

With ExpressRoute to ExpressRoute transitivity enabled on each Safe Hubs and Routing-Intent enabled, every Safe Hub sends the default RFC 1918 addresses (10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16) to on-premises over connection “E”. Along with the default RFC 1918 addresses, on-premises learns extra particular routes from Azure Digital Networks and Department Networks (S2S VPN, P2S VPN, SDWAN) which can be linked to each Hub 1 and Hub 2.

By default, on-premises doesn’t study the precise routes for each Azure VMware Answer Non-public Clouds. As an alternative, it routes to each Azure VMware Answer Non-public Clouds utilizing the default RFC 1918 addresses it learns over connection “E”. On-premises will study the default RFC 1918 addresses from each Hub 1 and Hub 2 through connection “E”.

To promote a extra particular route right down to on-premises, it must be achieved from the “Non-public Visitors Prefixes” field inside Routing Intent. – see Configure routing intent and insurance policies by Digital WAN portal. It’s good to add a summarized route that encompasses each your Azure VMware Answer /22 block and your Azure VMware Answer subnets. If you happen to add the identical precise prefix or a extra particular prefix as an alternative of a abstract route, you introduce routing points throughout the Azure atmosphere. Due to this fact, it’s necessary to keep in mind that any prefixes added to the “Non-public Visitors Prefixes” field have to be summarized routes.

As illustrated within the diagram, Azure VMware Answer Non-public Cloud 1 consists of workload subnets from 10.10.0.0/24 to 10.10.7.0/24. On Hub 1, the abstract route 10.10.0.0/21 is added to “Non-public Visitors Prefixes” as a result of it encompasses all eight subnets. Moreover, on Hub 1, the abstract route 10.150.0.0/22 is added to “Non-public Visitors Prefixes” to cowl the Azure VMware Answer administration block. Abstract routes 10.10.0.0/21 and 10.150.0.0/22 are then marketed right down to on-premises through connection “E”, offering on-premises with a extra particular route than 10.0.0.0/8.

Azure VMware Answer Non-public Cloud 2 consists of workload subnets from 10.20.0.0/24 to 10.20.7.0/24. On Hub 2, the abstract route 10.20.0.0/21 is added to “Non-public Visitors Prefixes” as a result of it encompasses all eight subnets. Moreover, on Hub 2, the abstract route 10.250.0.0/22 is added to “Non-public Visitors Prefixes.” This covers the Azure VMware Answer administration block. Abstract routes 10.20.0.0/21 and 10.250.0.0/22 are then marketed right down to on-premises through connection “E.” This gives on-premises with a extra particular route than 10.0.0.0/8.

There’s no concern in including the whole Azure VMware Answer Administration /22 block underneath “Non-public Visitors Prefixes” as a result of Azure VMware Answer doesn’t promote the precise /22 block again to Azure; it all the time advertises extra particular routes.

As talked about earlier, while you allow ExpressRoute to ExpressRoute transitivity on the Hub, it sends the default RFC 1918 addresses (10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16) to your on-premises community. Due to this fact, you should not promote the precise RFC 1918 prefixes (10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16) again to Azure. Promoting the identical precise routes creates routing issues inside Azure. As an alternative, it is best to promote extra particular routes again to Azure on your on-premises networks.

The diagram illustrates visitors flows from the attitude of on-premises.

Visitors Circulation Chart

This part focuses solely on connectivity from the Azure Digital Networks perspective. As depicted within the diagram, every Digital Community has a Digital Community peering on to its regional hub.

The diagram illustrates how all Azure native sources in each Digital Networks study routes underneath their “Efficient Routes”. With Routing Intent enabled, Hub 1 and Hub 2 all the time ship the default RFC 1918 addresses (10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16) to their peered Digital Networks. Azure native sources within the Digital Networks don’t study particular routes from exterior their Digital Community. With Routing Intent enabled, all sources within the Digital Community at the moment study the default RFC 1918 tackle and use their regional hub firewall as the subsequent hop. Azure VMware Answer Non-public Clouds talk with one another through connection “D” to their native regional hub firewall. From there, they traverse the Digital WAN inter-hub and endure inspection on the cross-regional hub firewall. Moreover, Azure VMware Answer non-public clouds talk with on-premises through connection “D” over their native regional hub firewall. All visitors ingressing and egressing the Digital Networks will all the time transit their regional hub firewalls. For extra data, see the visitors stream part.

The diagram illustrates visitors flows from the Azure Digital Networks perspective.

Visitors Circulation Chart

This part focuses solely on how web connectivity is supplied for Azure native sources in Digital Networks and Azure VMware Answer Non-public Clouds with twin area. There are a number of choices to offer web connectivity to Azure VMware Answer. – see  Web Entry Ideas for Azure VMware Answer 

Possibility 1: Web Service hosted in Azure
Possibility 2: VMware Answer Managed SNAT
Possibility 3: Azure Public IPv4 tackle to NSX-T Knowledge Middle Edge

Though you should use all three choices with Twin Area Safe Digital WAN with Routing Intent, “Possibility 1: Web Service hosted in Azure” is the best choice when utilizing Safe Digital WAN with Routing Intent and is the choice that’s used to offer web connectivity within the state of affairs. The rationale why “Possibility 1” is taken into account the best choice with Safe Digital WAN is because of its ease of safety inspection, deployment, and manageability.

As talked about earlier, while you allow Routing Intent on each Safe Hubs, it advertises RFC 1918 to all straight peered Digital Networks. Nonetheless, you may as well promote a default route 0.0.0.0/0 for web connectivity to downstream sources. With Routing Intent, you may select to generate a default route from each hub firewalls. This default route is marketed to its straight peered Digital Networks and to its straight linked Azure VMware Answer. This part is damaged into two sections, one which explains web connectivity from each regional Azure VMware Answer perspective and one other from the Digital Networks perspective.

When Routing Intent is enabled for web visitors, the default habits of the Safe Digital WAN Hub is to not promote the default route throughout ExpressRoute circuits. To make sure the default route is propagated to its straight linked Azure VMware Answer from the Azure Digital WAN, you need to allow default route propagation in your Azure VMware Answer ExpressRoute circuits – see To promote default route 0.0.0.0/0 to endpoints. As soon as adjustments are full, the default route 0.0.0.0/0 is then marketed through connection “D” from the hub. It’s necessary to notice that this setting should not be enabled for on-premises ExpressRoute circuits. As a greatest observe, it’s advisable to implement a BGP Filter in your on-premises gear. A BGP Filter in place prevents the inadvertent studying of the default route, provides an additional layer of precaution, and ensures that on-premises web connectivity is not impacted.

If you allow Routing Intent for web entry, it routinely generates a default route from each regional hubs and advertises it to their hub-peered Digital Community connections. You may discover underneath Efficient Routes for the Digital Machines’ NICs within the Digital Community that the 0.0.0.0/0 subsequent hop is the regional hub firewall. The default route isn’t marketed throughout regional hubs over the ‘inter-hub’ hyperlink. Due to this fact, Digital Networks use their native regional hub for web entry and haven’t any backup web connectivity to the cross-regional hub.

For extra data, see the visitors stream part.

The diagram illustrates visitors flows from the Digital Networks and Azure VMware Answer Non-public Clouds perspective.

Visitors Circulation Chart

With Azure VMware Answer utilizing the Twin-Area with out World Attain design, you don’t have outbound web connectivity redundancy as a result of every Azure VMware Answer non-public cloud learns the default route from each its native regional hub and isn’t straight linked to its cross-regional hub. If a regional outage that impacts the native regional hub, you’ve got two choices as a way to obtain web redundancy which can be handbook configurations.

Possibility 1: For Outbound Web Entry Solely
Throughout a neighborhood regional outage, should you want outbound web entry on your Azure VMware Answer workload, you may go for VMware Answer Managed SNAT. It’s an easy resolution that rapidly gives the entry you want. – see Activate Managed SNAT for Azure VMware Answer workloads 

Possibility 2: For Inbound and Outbound Web Entry
Throughout a neighborhood regional outage, should you want each inbound and outbound web entry on your Azure VMware Answer cloud, begin by eradicating the “D” connection on your native regional hub. Take away the Authorization Key created for the “D” connection from the Azure VMware Answer blade within the Azure portal. Then, create a brand new connection to the cross-regional hub. For dealing with inbound visitors, think about using Azure Entrance Door or Visitors Supervisor to take care of regional excessive availability.

HCX Mobility Optimized Networking (MON) is an optionally available characteristic to allow when utilizing HCX Community Extensions (NE). Mobility Optimized Networking (MON) gives optimum visitors routing underneath sure eventualities to stop community tromboning between the on-premises-based and cloud-based sources on prolonged networks.

Enabling Mobility Optimized Networking (MON) for a particular prolonged community and a digital machine adjustments the visitors stream. For Mobility Optimized Networking (MON), egress visitors from that digital machine does not trombone again to on-premises. As an alternative, it bypasses the Community Extensions (NE) IPSEC tunnel. Visitors for that digital machine will now egress out of the Azure VMware Answer NSX-T Tier-1 Gateway> NSX-T Tier-0 Gateway>Azure Digital WAN.

Enabling Mobility Optimized Networking (MON) for a particular prolonged community and a digital machine ends in a change. From Azure VMware Answer NSX-T, it injects a /32 host route again to Azure Digital WAN. Azure Digital WAN advertises this /32 route again to on-premises, Digital Networks, and Department Networks (S2S VPN, P2S VPN, SDWAN). The aim of this /32 host route is to make sure that visitors from on-premises, Digital Networks, and Department Networks (S2S VPN, P2S VPN, SDWAN) does not use the Community Extensions (NE) IPSEC tunnel when destined for the Mobility Optimized Networking (MON) enabled Digital Machine. Visitors from supply networks is directed straight to the Mobility Optimized Networking (MON) enabled Digital Machine as a result of /32 route that’s realized.

With ExpressRoute to ExpressRoute transitivity enabled on the Safe Hub and Routing-Intent enabled, the Safe Hub sends the default RFC 1918 addresses (10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16) to each the on-premises and Azure VMware Answer. Along with the default RFC 1918 addresses, each on-premises and Azure VMware Answer study extra particular routes from Azure Digital Networks and Department Networks (S2S VPN, P2S VPN, SDWAN) which can be linked to the hub. Nonetheless, on-premises networks do not study any particular routes from the Azure VMware Answer, nor does the reverse happen. As an alternative, each environments depend on the default RFC 1918 addresses to facilitate routing again to at least one one other through their native regional Hub firewall. Which means that extra particular routes, resembling HCX Mobility Optimized Networking (MON) Host Routes, aren’t marketed from the Azure VMware Answer ExpressRoute to the on-premises-based ExpressRoute circuit and vice-versa. The lack to study particular routes introduces uneven visitors flows. Visitors egresses Azure VMware Answer through the NSX-T Tier-0 gateway, however returning visitors from on-premises returns over the Community Extensions (NE) IPSEC tunnel.

To appropriate any visitors asymmetry, it is advisable to alter the HCX Mobility Optimized Networking (MON) Coverage Routes. Mobility Optimized Networking (MON) coverage routes decide which visitors goes again to the on-premises Gateway through an L2 extension. Additionally they determine which visitors is routed by the Azure VMware Answer NSX Tier-0 Gateway.

If a vacation spot IP matches and is ready to “enable” within the Mobility Optimized Networking (MON) coverage configuration, then two actions happen. First, the packet is recognized. Second, its despatched to the on-premises gateway by the HCX Community Extension equipment.

If a vacation spot IP does not match or is ready to “deny” within the Mobility Optimized Networking (MON) coverage, the system sends the packet to the Azure VMware Answer Tier-0 for routing.

HCX Coverage Routes