In Cloud Container Distribution, IBD workers' IP address allocation on external interfaces is done based on the corresponding network specification. The IP address is selected from the configured address allocation pool based on the IPAM policy implemented by the cloud provider. It is not possible to pre-select fixed IP for a specific worker once the allocated IP address is stable, that is, does not change during node upgrades or rebuilds.The recommended way of configuring upstream router is to configure the range of peering neighbors, that accommodates future uses (for example, possible scale out) and configure that range as allocation pool on the corresponding external networks to be used by Cloud Container Distribution workers. Dormant or unavailable workers do not run ECFE speaker component, thus are not advertised as routing peers, and are not used for ingress packet distribution.Pre-configuring the router before the Cloud Container Distribution cluster deployment allows to start receiving ingress traffic immediately. Pre-configuring all possible peers allows to avoid router configuration after the cluster scale-out.There is an important limitation within the BGP mode of ECFE. Using BGP as a load-balancing mechanism has the advantage that allows to use standard router hardware, rather than bespoke load balancers. However, this comes with downsides as well. The biggest is that BGP-based load balancing does not react gracefully to changes in the backend set for an address. This means that when a cluster node goes down, one can expect all active connections to the cluster external services to be broken. This situation naturally occurs during cluster rebuild and upgrades or worker pool scale-in and scale-out, since nodes are added or removed (during scale-in and out) or restarted (during rebuild and upgrade).