Easegress
is a Cloud Native traffic orchestration system designed for:
- High Availability: Built-in Raft consensus & leader election provides 99.99% availability.
- Traffic Orchestration: Simple orchestration of various filters for each traffic pipeline.
- High Performance: Lightweight and essential features speed up the performance.
- Observability: There are many meaningful statistics periodically in a readable way.
- Extensibility: It's easy to develop your own filter or controller with high-level programming language.
- Integration: The simple interfaces make it easy to integrate with other systems, such as Kubernetes Ingress, EaseMesh sidecar, Workflow, etc.
The architecture of Easegress:
- Service Management
- Multiple protocols:
- HTTP/1.1
- HTTP/2
- HTTP/3(QUIC)
- MQTT
- Rich Routing Rules: exact path, path prefix, regular expression of the path, method, headers.
- Resilience&Fault Tolerance
- Circuit breaker: temporarily blocks possible failures.
- Rate limiter: limits the rate of incoming requests.
- Retryer: repeats failed executions.
- Time limiter: limits the duration of execution.
- Deployment Management
- Blue-green Strategy: switches traffic at one time.
- Canary Strategy: schedules traffic slightly.
- API Management
- API Aggregation: aggregates results of multiple APIs.
- API Orchestration: orchestrates the flow of APIs.
- Security
- Pipeline-Filter Mechanism
- Chain of Responsibility Pattern: orchestrates filters chain.
- Filter Management: makes it easy to develop new filters.
- Service Mesh
- Mesh Master: is the control plane to manage the lifecycle of mesh services.
- Mesh Sidecar: is the data plane as the endpoint to do traffic interception and routing.
- Mesh Ingress Controller: is the mesh-specific ingress controller to route external traffic to mesh services.
Notes: This feature is leveraged by EaseMesh
- Third-Part Integration
- FaaS integrates with the serverless platform Knative.
- Service Discovery integrates with Eureka, Consul, Etcd, and Zookeeper.
- Ingress Controller integrates with Kubernetes as an ingress controller.
- Multiple protocols:
- Extensibility
- WebAssembly executes user developed WebAssembly code.
- High Performance and Availability
- Adaption: adapts request, response in the handling chain.
- Validation: headers validation, OAuth2, JWT, and HMAC verification.
- Load Balance: round-robin, random, weighted random, IP hash, header hash.
- Cache: for the backend servers.
- Compression: compresses body for the response.
- Hot-Update: updates both config and binary of Easegress in place without losing connections.
- Operation
- Easy to Integrate: command line(
egctl
), MegaEase Portal, HTTP clients such as curl, postman, etc. - Distributed Tracing
- Built-in Open Zipkin
- Open Tracing for vendor-neutral APIs
- Observability
- Node: role(leader, writer, reader), health or not, last heartbeat time, and so on
- Traffic: in multi-dimension: server and backend.
- Throughput: total and error statistics of request count, TPS/m1, m5, m15, and error percent, etc.
- Latency: p25, p50, p75, p95, p98, p99, p999.
- Data Size: request and response size.
- Status Codes: HTTP status codes.
- TopN: sorted by aggregated APIs(only in server dimension).
- Easy to Integrate: command line(
The following examples show how to use Easegress for different scenarios.
- API Aggregator - Aggregating many APIs into a single API.
- Distributed Tracing - How to do APM tracing - Zipkin.
- FaaS - Supporting Knative FaaS integration
- Flash Sale - How to do high concurrent promotion sales with Easegress
- Kubernetes Ingress Controller - How to integrate with Kubernetes as ingress controller
- LoadBalancer - A number of the strategies of load balancing
- MQTTProxy - An Example to MQTT proxy with Kafka backend.
- Performance - Performance optimization - compression, caching etc.
- Pipeline - How to orchestrate HTTP filters for requests/responses handling
- Resilience and Fault Tolerance - Circuit Breaker, Rate Limiter, Retryer, Time limiter, etc. (Porting from Java resilience4j)
- Security - How to do authentication by Header, JWT, HMAC, OAuth2, etc.
- Service Proxy - Supporting the Microservice registries - Zookeeper, Eureka, Consul, Nacos, etc.
- WebAssembly - Using AssemblyScript to extend the Easegress
- Workflow - An Example to make a workflow for a number of APIs.
For full list, see Cookbook.
The basic common usage of Easegress is to quickly set up proxy for the backend servers. We split it into multiple simple steps to illustrate the essential concepts and operations.
We can download the latest or history binaries from the release page. Following shell script will download and extract the latest binaries to ./easegress
folder:
/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/megaease/easegress/main/scripts/install.sh)"
or if we can install Easegress from source code:
$ git clone https://github.com/megaease/easegress && cd easegress
$ make
Note:
- We need go 1.16+ complier for build.
- If you need the WebAssembly feature, please run
make wasm
.
Then we can add the binary directory to the PATH
and execute the server:
$ export PATH=${PATH}:$(pwd)/bin/
$ easegress-server
2021-05-17T16:45:38.185+08:00 INFO cluster/config.go:84 etcd config: init-cluster:eg-default-name=https://localhost:2380 cluster-state:new force-new-cluster:false
2021-05-17T16:45:38.185+08:00 INFO cluster/cluster.go:379 client is ready
2021-05-17T16:45:39.189+08:00 INFO cluster/cluster.go:590 server is ready
2021-05-17T16:45:39.21+08:00 INFO cluster/cluster.go:451 lease is ready
2021-05-17T16:45:39.231+08:00 INFO cluster/cluster.go:187 cluster is ready
2021-05-17T16:45:39.253+08:00 INFO supervisor/supervisor.go:180 create system controller StatusSyncController
2021-05-17T16:45:39.253+08:00 INFO cluster/cluster.go:496 session is ready
2021-05-17T16:45:39.253+08:00 INFO api/api.go:96 api server running in localhost:2381
2021-05-17T16:45:44.235+08:00 INFO cluster/member.go:210 self ID changed from 0 to 689e371e88f78b6a
2021-05-17T16:45:44.236+08:00 INFO cluster/member.go:137 store clusterMembers: eg-default-name(689e371e88f78b6a)=https://localhost:2380
2021-05-17T16:45:44.236+08:00 INFO cluster/member.go:138 store knownMembers : eg-default-name(689e371e88f78b6a)=https://localhost:2380
The default target of Makefile is to compile two binary into the directory bin/
. bin/easegress-server
is the server-side binary, bin/egctl
is the client-side binary. We could add it to the $PATH
for simplifying the following commands.
We could run easegress-server
without specifying any arguments, which launch itself by opening default ports 2379, 2380, 2381. Of course, we can change them in the config file or command arguments that are explained well in easegress-server --help
.
$ egctl member list
- options:
name: eg-default-name
labels: {}
cluster-name: eg-cluster-default-name
cluster-role: writer
cluster-request-timeout: 10s
cluster-listen-client-urls:
- https://127.0.0.1:2379
cluster-listen-peer-urls:
- https://127.0.0.1:2380
cluster-advertise-client-urls:
- https://127.0.0.1:2379
cluster-initial-advertise-peer-urls:
- https://127.0.0.1:2380
cluster-join-urls: []
api-addr: localhost:2381
debug: false
home-dir: ./
data-dir: data
wal-dir: ""
log-dir: log
member-dir: member
cpu-profile-file: ""
memory-profile-file: ""
lastHeartbeatTime: "2021-05-05T15:43:27+08:00"
etcd:
id: a30c34bf7ec77546
startTime: "2021-05-05T15:42:37+08:00"
state: Leader
After launched successfully, we could check the status of the one-node cluster. It shows the static options and dynamic status of heartbeat and etcd.
Now let's create an HTTPServer listening on port 10080 to handle the HTTP traffic.
$ echo '
kind: HTTPServer
name: server-demo
port: 10080
keepAlive: true
https: false
rules:
- paths:
- pathPrefix: /pipeline
backend: pipeline-demo' | egctl object create
The rules of routers above mean that it will lead the traffic with the prefix /pipeline
to the pipeline pipeline-demo
, which will be created below. If we curl
it before created, it will return 503.
$ echo '
name: pipeline-demo
kind: HTTPPipeline
flow:
- filter: proxy
filters:
- name: proxy
kind: Proxy
mainPool:
servers:
- url: https://127.0.0.1:9095
- url: https://127.0.0.1:9096
- url: https://127.0.0.1:9097
loadBalance:
policy: roundRobin' | egctl object create
The pipeline means it will do proxy for 3 backend endpoints in load balance policy roundRobin
.
Now you can use some HTTP clients such as curl
to test the feature:
$ curl -v https://127.0.0.1:10080/pipeline
If you are not set up some applications to handle the 9095, 9096, and 9097 in the localhost, it will return 503 too. We prepare a simple service to let us test handily, the example shows:
$ go run example/backend-service/mirror/mirror.go & # Running in background
$ curl https://127.0.0.1:10080/pipeline -d 'Hello, Easegress'
Your Request
===============
Method: POST
URL : /pipeline
Header: map[Accept:[*/*] Accept-Encoding:[gzip] Content-Type:[application/x-www-form-urlencoded] User-Agent:[curl/7.64.1]]
Body : Hello, Easegress
Now we want to add more features to the pipeline, then we could add kinds of filters to the pipeline. For example, we want validation and request adaptation for the pipeline-demo
.
$ cat pipeline-demo.yaml
name: pipeline-demo
kind: HTTPPipeline
flow:
- filter: validator
jumpIf: { invalid: END }
- filter: requestAdaptor
- filter: proxy
filters:
- name: validator
kind: Validator
headers:
Content-Type:
values:
- application/json
- name: requestAdaptor
kind: RequestAdaptor
header:
set:
X-Adapt-Key: goodplan
- name: proxy
kind: Proxy
mainPool:
servers:
- url: https://127.0.0.1:9095
- url: https://127.0.0.1:9096
- url: https://127.0.0.1:9097
loadBalance:
policy: roundRobin
$ egctl object update -f pipeline-demo.yaml
After updating the pipeline, the original curl -v https://127.0.0.1:10080/pipeline
will get 400 because of the validating. So we changed it to satisfy the limitation:
$ curl https://127.0.0.1:10080/pipeline -H 'Content-Type: application/json' -d '{"message": "Hello, Easegress"}'
Your Request
===============
Method: POST
URL : /pipeline
Header: map[Accept:[*/*] Accept-Encoding:[gzip] Content-Type:[application/json] User-Agent:[curl/7.64.1] X-Adapt-Key:[goodplan]]
Body : {"message": "Hello, Easegress"}
We can also see Easegress send one more header X-Adapt-Key: goodplan
to the mirror service.
See reference and developer guide for more information.
See Easegress Roadmap for details.
- Join Slack Workspace for requirement, issue and development.
- MegaEase on Twitter
Easegress is under the Apache 2.0 license. See the LICENSE file for details.