This repository contains a Go-based implementation of an API server that
implements the discovery service APIs defined in
data-plane-api.
Scope
Due to the variety of platforms out there, there is no single
control plane implementation that can satisfy everyone's needs. Hence this
code base does not attempt to be a full scale control plane for a fleet of
Envoy proxies. Instead, it provides infrastructure that is shared by
multiple different control plane implementations. The components provided
by this library are:
API Server: A generic gRPC based API server that implements xDS APIs as defined
in the
data-plane-api. The API
server is responsible for pushing configuration updates to
Envoys. Consumers should be able to import this go library and use the
API server as is, in production deployments.
Configuration Cache: The library will cache Envoy configurations in
memory in an attempt to provide fast response to consumer Envoys. It is the
responsibility of the consumer of this library to populate the cache as
well as invalidate it when necessary. The cache will be keyed based on a
pre-defined hash function whose keys are based on the
Node information.
At this moment, this repository will not tackle translating platform
specific representation of resources (e.g., services, instances of
services, etc.) into Envoy-style configuration. Based on usage and
feedback, we might decided to revisit this aspect at a later point in time.
Requirements
Go 1.16+
Quick start
It's recommended to run the tests with make docker_tests as it executes the tests
in the same environment as CI. This makes sure to produce a consistent set of generated files.
Note: It is recommended to use a previous SHA if there is still a need for V2.
Resource caching
Because Envoy clients are assumed to be ephemeral, and thus, can come and go
away arbitrarily, the server relies on a configuration cache to minimize the
client load on the server. There are several caches available in this
repository:
Simple cache is a snapshot-based cache that maintains a consistent view of
the configuration for each group of proxies. It supports running as an ADS
server or as regular dis-aggregated xDS servers. In ADS mode, the cache can
hold responses until the complete set of referenced resources is requested
(e.g. the entire set of RDS as referenced by LDS). Holding the response
enables an atomic update of xDS collections.
Linear is an eventually consistent cache for a single type URL collection.
The cache maintains a single linear version history and a version vector for
the resources in the cache. For each request, it compares the request version
against latest versions for the requested resources, and responds with any
updated resources. This cache assumes the resources are entirely opaque.
Mux cache is a simple cache combinator. It allows mixing multiple caches
for different type URLs, e.g use a simple cache for LDS/RDS/CDS and a linear
cache for EDS.
Usage
The example server demonstrates how to integrate the go-control-plane with your code.
请发表评论