We are rolling out networking infrastructure upgrades to all of our existing data centers. These upgrades increase the stability and resiliency of our already reliable network. It also enables us to bring features, such as VLAN and IP Sharing, to every data center.

For most customers, these upgrades are performed seamlessly behind the scenes. For customers that use certain features, such as IP Sharing and /116 IPv6 pools, there may be some changes that impact your current configuration. This document outlines what is changing, what data centers are impacted, and what, if anything, you may need to do in order to prepare for these upcoming changes.

What's new?

• IP Sharing (IP failover) availability: The IP Sharing feature, as it exists before these upgrades, enables IP failover for public IPv4 addresses in select data centers. After the upgrades have been completed, this feature will be expanded to all data centers and will also support IPv6 routed ranges (/64 and /56). See our Configure IP failover on a Compute Instance guide to learn more about configuring IP failover.

• VLAN availability: VLANs, which enable private layer 2 networking, will be launched across all data centers soon after the network upgrades have occurred.

What's changing?

The following is a list of breaking changes and any action that may be required if you are impacted by that change:

• Deprecation of IPv6 /116 pools: /116 pools will no longer be provided to new Compute Instances. Existing /116 pools will be removed from Compute Instances when data center is undergoing upgrades.

  Action: If you are using /116 for IPv6 failover, consider using an IPv6 /64 instead.

• IP failover through BGP: IP failover (IP Sharing) for public IPv4 addresses and IPv6 routed ranges will be facilitated through BGP instead of ARP (configured through keepalived).

  Action: If you have previously configured IP failover for a public IPv4 address, review the Configure IP failover on a Compute Instance guide to learn more about configuring IP failover using BGP. You can configure BGP ahead of time, but will not be able to test or use the configuration until after the network upgrades have been completed.

Which data centers have been upgraded?

Review the table below to learn which data centers have been upgraded with the latest network enhancements. This table covers the regions that existed at the start of this initiative. If a region is not listed, it was launched after this initiative began and, as such, already has the relevant network upgrades.

A status of complete indicates that all new Compute Instances (and most existing Compute Instances) are located on fully upgraded hardware. Compute Instances using legacy features, such as ARP-based failover and /116 ranges, may still be located on hardware that hasn't yet been upgraded. These customers have been notified and a migration timeline has been shared.

📘

If a data center is marked as undergoing network upgrades, customers may encounter issues enabling IP Sharing and configuring failover. For Compute Instances that already have IP Sharing enabled, this feature should still function as intended. Once the network upgrades are completed, IP Sharing will be supported through the new method (BGP). See Configure IP failover on a Compute Instance.

What action is required?

• Migration of Compute Instances: Once a data center has started the network infrastructure upgrades, live migrations are scheduled for all Compute Instances that do not reside on upgraded hardware. This live migration occurs while your Compute Instance is powered on and operating normally. Immediately following a successful live migration, there may be a small amount of packet loss as traffic is routed to the newer hardware.

  🚧

  Important

  If a live migration is not possible, a cold migration is scheduled instead. During a cold migration, the Compute Instance is powered off while its data is migrated to newer hardware. This results in downtime unless your application is configured to fail over to another Compute Instance or server. If a Compute Instance requires a cold migration, the account owner will receive a ticket from the Support team outlining this process and the scheduled migration date.

• Update IP failover configuration: If you have configured IP failover for a public IPv4 address, review the Configure IP failover on a Compute Instance guide to learn more about configuring IP failover using BGP. If you were using a now deprecated IPv6 /116 pool for IP failover, consider using an IPv6 /64 range instead. You can configure BGP ahead of time, but will not be able to test or use the configuration until after your Compute Instances are migrated to upgraded hardware.