This introduces the concept of inference endpoints and inference endpoint replicas. An inference endpoint describes a model and is backed by a private network, a load balancer and multiple replicas. A replica is basically a VM that runs the model. The replica VM is part of the inference endpoints private subnet and attached to its load balancer.
15 lines
303 B
Ruby
15 lines
303 B
Ruby
# frozen_string_literal: true
|
|
|
|
require_relative "../../model"
|
|
|
|
class InferenceEndpointReplica < Sequel::Model
|
|
one_to_one :strand, key: :id
|
|
one_to_one :vm, key: :id, primary_key: :vm_id
|
|
many_to_one :inference_endpoint
|
|
|
|
include ResourceMethods
|
|
include SemaphoreMethods
|
|
|
|
semaphore :destroy
|
|
end
|