Dedicated Interconnect provides direct physicalconnections between your on-premises network and Google's network.Dedicated Interconnect lets you transfer largeamounts of data between networks, which can be more cost-effective thanpurchasing additional bandwidth over the public internet.

Before you use Dedicated Interconnect

Make sure that you meet the following requirements:

• Be familiar with basic networkinterconnections so that you can order and configure circuits.
• Be familiar withCloud Interconnectterminology.
• Your network must physically meet Google's network in acolocationfacility. You mustprovide your own routing equipment. Your on-premises router is typicallylocated in the colocation facility. However, you can also extend yourconnection to a router outside of the colocation facility.

• In the colocation facility, your network devices must support the followingtechnical requirements:

  • 10-Gbps circuits, single mode fiber, 10GBASE-LR (1310 nm), or 100-Gbpscircuits, single mode fiber, 100GBASE-LR4

    Note: The maximum supported fiber length for 10GBASE-LR or 100GBASE-LR4optics is 10 km. If your on-premises router is more than 10 km away from the Googledemarcation, you must use a lit wavelength,DWDM,or other managed circuit to complete your connection.

  • IPv4 link local addressing

  • LACP, even if you're using a single circuit

  • EBGP-4 with multi-hop

  • 802.1Q VLANs

How does Dedicated Interconnect work?

For Dedicated Interconnect, you provision aDedicated Interconnect connection between the Google network andyour own network.

The following example diagram shows a singleDedicated Interconnect connection between aVirtual Private Cloud (VPC) network and your on-premises network.

For the basic setup shown in the diagram, aDedicated Interconnect connection is provisioned between theGoogle network and the on-premises router in a common colocation facility. Yoursetup might be different if your on-premises router is not in the samecolocation facility as your Dedicated Interconnect demarcation.

When you create a VLAN attachment, you associate it with a Cloud Router. ThisCloud Router creates a BGP session for the VLAN attachment and itscorresponding on-premises peer router. The Cloud Router receives theroutes that your on-premises router advertises. These routes are added ascustom dynamic routes in your VPC network. TheCloud Router also advertises routes for Google Cloud resourcesto the on-premises peer router.

VLAN attachment MTU options

VLAN attachments can have a maximum transmission unit (MTU) of 1440, 1460, 1500or 8896 bytes.

The following limitations apply:

• VLAN attachments with an MTU of 8896 (also known asjumbo frames) aresupported only on unencrypted IPv4 and IPv6 VLAN attachments.

• Requests to Google API Client Libraries automatically use 1440 MTU packets, evenif your VLAN attachment is set to a higher MTU value, including jumboframes.

We recommend that you use the same MTU for all VLAN attachments that areconnected to the same VPC network, and that you set the MTU ofthe VPC networks to the same value. For more information aboutCloud Interconnect MTUs, seeCloud InterconnectMTU.

Provisioning

To create and configure a Dedicated Interconnect connection, youstart by deciding where you want a Dedicated Interconnectconnection and whether you wantMACsec for Cloud Interconnect. Then, youorder a Dedicated Interconnect connection so that Google canallocate the necessary resources and send you a Letter of Authorization andConnecting Facility Assignment (LOA-CFA). After you receive the LOA-CFA, youneed to submit it to your vendor so that they can provision the connectionsbetween Google's network and your network.

You then need to configure and test the connections with Google before you canuse them. After they're ready, you can create VLAN attachments to allocate aVLAN on the connection.

For detailed steps to provision a Dedicated Interconnectconnection, see theProvisioning overview.

Note: If you encounter any issues during the provisioning stage, we advise youto reply directly to the email received with yourLOA-CFA.Replying to that email lets you receive support faster than creating asupport case.

Fixed pricing

Dedicated Interconnect offers fixed port pricing for outbounddata transfers for VLAN attachments. You can choose fixed pricing for eachDedicated Interconnect connection and choose which types ofconnections you want the fixed pricing to apply to. This lets you get a fixedmonthly invoice for outbound data transfers.

Fixed port pricing considers the following types of connections:

• Local connection: The VLAN attachment is in the same a metro locationwhere the destination Google Cloudregion is located.

  For example, if you obtain a VLAN attachment in the Los Angeles, Californiametro location and the destination Google Cloud region isus-west2,then the VLAN attachment's location and destination Google Cloudregion are the same. This is considered a local connection.

• Remote connection: The VLAN attachment is in a different metrolocation where the destination Google Cloud region is located.

  For example, if you create a VLAN attachment in the Los Angeles, Californiametro location and the destination Google Cloud region isus-east4,then the VLAN attachment's location and destination Google Cloudregion are different. This is considered a remote connection.

  Similarly, if you obtain a VLAN attachment in the Portland, Oregon metrolocation, then there isn't a local Google Cloud region availablewithin that metro location. Because you can't connect to a localGoogle Cloud region, this is considered a remote connection.

The following Google Cloud regions only offer support for fixed pricingfor remote connections on Dedicated Interconnect:

• us-west1
• us-east1
• europe-west4
• europe-north1
• asia-east1

To request a Dedicated Interconnect connection with fixed portpricing, contact your account team.

Cloud Interconnect groups

You can use connection groups to communicate your intended level of reliability,and to receive feedback on how your Cloud Interconnect resources meetthat intended level of reliability. There are two types ofCloud Interconnect resource groups:

• Interconnect groups. Use these to group your Cloud Interconnectconnections.
• Attachment groups. Use these to group your VLAN attachments.

When you group your Cloud Interconnect connections and VLANattachments, you receive information about the group's structure, global routingrequirements, operational status, SLAeligibilityand blockers. Use these groups to establish thatyour deployment meets the requirements for your chosen level of reliability. Youcan use Cloud Interconnect connection groups without configuring VLANattachment groups, and you can use VLAN attachment groups without configuringCloud Interconnect connection groups.

Connection groups are only an informational and organizational tool.Connection groupsnever change the behavior of yourCloud Interconnect connections or your VLAN attachments. This makesconnection groups safe to use for your production resources.

You can add a Cloud Interconnect connection to more than oneCloud Interconnect connection group. The connection group must be inthe same project as the resources in the group.

You can't add a VLAN attachment to more than one VLAN attachment group. Inaddition, all of the VLAN attachments in an attachment group must be configuredfor the same VPC network.

To configure Cloud Interconnect groups, seeCreate a Dedicated Interconnect connection group.

Redundancy and SLA

When you configure a Cloud Interconnect connection, you can do so ata specific level of reliability. You can choose between the followingreliability options:

• Critical production. Choose this option for critical production workloadsthat require maximum resiliency. This option provides a 99.99% uptime SLA.
• Non-critical production. Choose this option for non-critical productionand development workloads. This option provides a 99.9% uptime SLA.
• No SLA. You can configure your connection group without specifying anintended level of reliability. We recommend that you avoid using this optionfor resources in which extended downtime is undesirable. This option doesn'tprovide an uptime SLA.

For more information about Cloud Interconnect SLAs, seeDedicated Interconnect and Partner Interconnect SLA.

For the highest level availability, we recommend the configuration for 99.99%availability as the base configuration, as shown in the following diagram.Clients in the on-premises network can reach the IP addresses of virtual machine(VM) instances in theus-central1 region through at least one of the redundantpaths. If one path is unavailable, the other paths can continue to servetraffic.

We recommend that you use the 99.99% availability configuration forproduction-level applications with a low tolerance for downtime. If yourapplications aren't mission-critical and can tolerate some downtime, you canuse the 99.9% availability configuration.

The SLA requires properly configured topologies that are defined by the 99.99%and 99.9% configurations. These configurations support availability and providean SLA.

Each level of reliability has a different set of metropolitan area (metro) andedge availability domain (metro availability zone) requirements:

• Critical production.At least four Dedicated Interconnectconnections, two connections in one metropolitan area (metro) and twoconnections in another metro. Connections that are in the same metro must beplaced indifferentedge availability domains(metro availability zones) to achieve 99.99% availability. PlacingDedicated Interconnect connections in two separate edgeavailability domains within the same metro is important because maintenancewindows are coordinated across edge availability domains within a metro,while maintenance windows arenot coordinated across metros.
• Non-critical production.At least two Cloud Interconnectconnections. The connections must be located in the same metropolitan area(metro) but in different edge availability domains (metro availabilityzones). It doesn't matter if the connections are in the same facility or indifferent facilities, as long as they are located in twodifferent edgeavailability domains.

Balance egress traffic with redundant connections

When you have a redundant topology similar to the 99.99% configuration, thereare multiple paths for traffic to traverse from the VPC networkto your on-premises network.

Google Cloud usesECMP to balance the egress traffic acrossconnections. To use ECMP, the Cloud Routers used by the VLANattachments must receive the same announcement with equal cost (the same CIDRrange and the same MED values).

Google Cloud balances the traffic between the VLAN attachments basedupon the configured capacity of each VLAN attachment.

Create redundant connections with sufficient capacity

The Best practices document describes best practices forcreating redundant Cloud Interconnect connections that have sufficientcapacity in a failover scenario.When you follow these practices, events such as planned maintenance or hardwarefailures are less likely to cause a loss of connectivity.

Dedicated Interconnect availability

A Dedicated Interconnect connection is considered available ifyou can send and receive ICMP packets (ping) between a VM in a specificGoogle Cloud region and a correctly configured machine in youron-premises network. You should be able to send and receive packets through atleast one of your redundant connections.

Note: When testing availability, Google assumes that all connections and relateddevices are correctly configured and connected.

IPv6 support

Dedicated Interconnect supports IPv6 traffic.

To support IPv6 traffic in a Dedicated Interconnect, do thefollowing:

• Configure your VPC networks to use either IPv4 and IPv6 (dualstack) or IPv6-onlysubnets.

• Assign subnets withininternal IPv6ranges.

• Configure IPv6 addresses for VMs and instancetemplates within thesubnet.

For more information about configuring IPv6 within a subnet, see thefollowing:

• To create a custom mode VPC network with internal IPv6addresses, seeCreate a custom mode VPC network with adual-stacksubnet.

• To create a dual-stack subnet, seeAdd a dual-stacksubnet.

• To enable IPv6 in an existing IPv4-only subnet, seeChange an IPv4-onlysubnet to dual stack.

• To add an IPv6-only subnet, seeAdd an IPv6-onlysubnet

• To create or enable VMs with IPv6, seeConfigure IPv6 for instances andinstance templates.

For information about using internal IPv6 ranges in your VPCnetwork and subnets, seeInternal IPv6specifications.

After configuring IPv6 in your VPC network, subnets, and VMs,configure your VLAN attachments.

Stack types and BGP sessions

With Dedicated Interconnect, you can choose between two differentstack types for your VLAN attachment.

• Single stack (IPv4 only)
• Dual stack (IPv4 and IPv6)

The stack type that you select for your VLAN attachment determines what versionof IP traffic is supported by your Dedicated Interconnectconnection.

When you create the BGP sessions for a dual-stack VLAN attachment,you have the following options for IPv6 route exchange:

• You can create an IPv6 BGP session
• You can create an IPv4 BGP session thatexchanges IPv6 routes by usingmultiprotocol BGP(MP-BGP).

The following table summarizes the types of BGP sessions allowed for eachDedicated Interconnect VLAN attachment.

Stack type	Supported BGP sessions
IPv4 only	IPv4 BGP
IPv4 and IPv6	IPv4 BGP, with or without MP-BGP IPv6 BGP, with or without MP-BGP Both IPv4 BGP and IPv6 BGP, no MP-BGP

For more information about BGP sessions, seeEstablish BGP sessionsin the Cloud Router documentation.

Single-stack IPv4-only VLAN attachments

By default, a Dedicated Interconnect VLAN attachment is assignedthe IPv4-only stack type.

An IPv4-only VLAN attachment can support only IPv4 traffic.

Use the following procedure to create an IPv4-onlyDedicated Interconnect VLAN attachment and an IPv4 BGP session.

• Create VLAN attachments

Dual-stack IPv4 and IPv6 VLAN attachments

A Dedicated Interconnect VLAN attachment that is configured withthe dual-stack (IPv4 and IPv6) stack type can support both IPv4 and IPv6traffic.

For a dual-stack VLAN attachment, you can configure your Cloud Routerwith an IPv4 BGP session, an IPv6 BGP session, or both. If you configure onlyone BGP session, you can enable MP-BGP to allow that session to exchange bothIPv4 and IPv6 routes. If you create an IPv4 BGP session and an IPv6 BGP session,you can't enable MP-BGP on either session.

Use the following procedure to create a dual-stackDedicated Interconnect VLAN attachment and all supported BGPsessions.

• Create VLAN attachments

You can also change the stack type of your VLAN attachment afteryou create the attachment. For more information,seeModify VLAN attachment.

Custom IP address ranges

When you create a VLAN attachment for Dedicated Interconnect,you can configure custom IP address ranges for the Cloud Router andcustomer router ends of the attachment. For information about how it works,including limitations and best practices, see theCustom IP address ranges section inthe Cloud Interconnect overview.

When you configure custom IP address ranges for VLAN attachments that you usewith Dedicated Interconnect, you must provide the custom IPaddress ranges during VLAN creation. However, you can add IPv6 custom IP addressranges when you upgrade your stack type fromIPV4_ONLY toIPV4_IPV6. Toconfigure custom IP address ranges with Dedicated Interconnect,seeConfigure custom IP address ranges.

Restrict Dedicated Interconnect usage

By default, any VPC network can use Cloud Interconnect.To control which VPC networks can use Cloud Interconnect, you can set anorganization policy. For more information, seeRestrict Cloud Interconnect usage.

What's next?

• To find answers to common questions about Cloud Interconnectarchitecture and features, see theCloud Interconnect FAQ.
• To find out more about Cloud Interconnect, see theCloud Interconnect overview.
• To learn about best practices when planning for and configuringCloud Interconnect, seeBest practices.
• To find Google Cloud resource names, see theCloud Interconnect APIs.