open-consul/website/content/docs/connect/l7-traffic/index.mdx

122 lines
4.7 KiB
Plaintext
Raw Normal View History

2019-07-09 02:11:19 +00:00
---
2020-04-07 18:55:19 +00:00
layout: docs
page_title: Connect - L7 Traffic Management
description: >-
Layer 7 traffic management allows operators to divide L7 traffic between
different subsets of service instances when using Connect.
2019-07-09 02:11:19 +00:00
---
2020-04-06 20:27:35 +00:00
-> **1.6.0+:** This feature is available in Consul versions 1.6.0 and newer.
2019-07-09 02:11:19 +00:00
# L7 Traffic Management
2019-07-09 02:11:19 +00:00
Layer 7 traffic management allows operators to divide L7 traffic between
different
[subsets](/docs/connect/config-entries/service-resolver#service-subsets) of
2019-07-09 02:11:19 +00:00
service instances when using Connect.
There are many ways you may wish to carve up a single datacenter's pool of
services beyond simply returning all healthy instances for load balancing.
Canary testing, A/B tests, blue/green deploys, and soft multi-tenancy
(prod/qa/staging sharing compute resources) all require some mechanism of
carving out portions of the Consul catalog smaller than the level of a single
service and configuring when that subset should receive traffic.
2020-04-06 20:27:35 +00:00
-> **Note:** This feature is not compatible with the
2020-04-09 23:46:54 +00:00
[built-in proxy](/docs/connect/proxies/built-in),
[native proxies](/docs/connect/native),
and some [Envoy proxy escape hatches](/docs/connect/proxies/envoy#escape-hatch-overrides).
2019-07-09 02:11:19 +00:00
## Stages
Connect proxy upstreams are discovered using a series of stages: routing,
splitting, and resolution. These stages represent different ways of managing L7
traffic.
2020-04-07 23:56:08 +00:00
![diagram showing l7 traffic discovery stages: routing to splitting to resolution](/img/l7-traffic-stages.svg)
2019-07-09 02:11:19 +00:00
Each stage of this discovery process can be dynamically reconfigured via various
[configuration entries](/docs/agent/config-entries). When a configuration
2019-07-09 02:11:19 +00:00
entry is missing, that stage will fall back on reasonable default behavior.
### Routing
A [`service-router`](/docs/connect/config-entries/service-router) config
2019-07-09 02:11:19 +00:00
entry kind is the first configurable stage.
A router config entry allows for a user to intercept traffic using L7 criteria
such as path prefixes or http headers, and change behavior such as by sending
traffic to a different service or service subset.
These config entries may only reference `service-splitter` or
`service-resolver` entries.
[Examples](/docs/connect/config-entries/service-router#sample-config-entries)
2019-07-09 02:11:19 +00:00
can be found in the `service-router` documentation.
### Splitting
A [`service-splitter`](/docs/connect/config-entries/service-splitter) config
2019-07-09 02:11:19 +00:00
entry kind is the next stage after routing.
A splitter config entry allows for a user to choose to split incoming requests
across different subsets of a single service (like during staged canary
rollouts), or perhaps across different services (like during a v2 rewrite or
other type of codebase migration).
These config entries may only reference `service-splitter` or
`service-resolver` entries.
If one splitter references another splitter the overall effects are flattened
into one effective splitter config entry which reflects the multiplicative
union. For instance:
2020-04-06 20:27:35 +00:00
splitter[A]: A_v1=50%, A_v2=50%
splitter[B]: A=50%, B=50%
---------------------
splitter[effective_B]: A_v1=25%, A_v2=25%, B=50%
2019-07-09 02:11:19 +00:00
[Examples](/docs/connect/config-entries/service-splitter#sample-config-entries)
2019-07-09 02:11:19 +00:00
can be found in the `service-splitter` documentation.
### Resolution
A [`service-resolver`](/docs/connect/config-entries/service-resolver) config
2019-07-09 02:11:19 +00:00
entry kind is the last stage.
A resolver config entry allows for a user to define which instances of a
service should satisfy discovery requests for the provided name.
Examples of things you can do with resolver config entries:
- Control where to send traffic if all instances of `api` in the current
2020-04-06 20:27:35 +00:00
datacenter are unhealthy.
2019-07-09 02:11:19 +00:00
- Configure service subsets based on `Service.Meta.version` values.
- Send all traffic for `web` that does not specify a service subset to the
2020-04-06 20:27:35 +00:00
`version1` subset.
2019-07-09 02:11:19 +00:00
- Send all traffic for `api` to `new-api`.
- Send all traffic for `api` in all datacenters to instances of `api` in `dc2`.
- Create a "virtual service" `api-dc2` that sends traffic to instances of `api`
2020-04-06 20:27:35 +00:00
in `dc2`. This can be referenced in upstreams or in other config entries.
2019-07-09 02:11:19 +00:00
If no resolver config is defined for a service it is assumed 100% of traffic
flows to the healthy instances of a service with the same name in the current
datacenter/namespace and discovery terminates.
This should feel similar in spirit to various uses of Prepared Queries, but is
not intended to be a drop-in replacement currently.
These config entries may only reference other `service-resolver` entries.
[Examples](/docs/connect/config-entries/service-resolver#sample-config-entries)
2019-07-09 02:11:19 +00:00
can be found in the `service-resolver` documentation.
-> **Note:** `service-resolver` config entries kinds can function at L4 (unlike
2019-07-09 02:11:19 +00:00
`service-router` and `service-splitter` kinds). These can be created for
services of any protocol such as `tcp`.