Scalable, Transactional and Reliable Game Server Framework based on Pomelo and MemDB
- Fast in-memory data access.
- Distributed architecture, system capacity is horizontally scalable. Performance can be linearly increased by simply add more servers.
- ACID(Stands for Atomicity, Consistency, Isolation, Durability) transaction support on distributed environment.
- Data atomicity and consistency guarantee, never leave dirty data in memory.
- Concurrency and locking control, which make it very easy to write concurrency code. High performance on concurrent system.
- Each server is backed by one or more replica, no single point of failure.
- Simple and clear Module-Controller architecture.
- Use Mongoose to define data models.
- Promise A+ compatible
- Support ES6 generators (yield)
- Very powerful built-in modules, like push module. You can build a full featured push/chat service with almost zero code.
- Home Page: http://quickpomelo.com
- Github: https://github.com/rain1017/quick-pomelo
- Wiki: https://github.com/rain1017/quick-pomelo/wiki
- Demo Game: https://github.com/rain1017/quick-pomelo-demo
- Email: [email protected]
- QQ: 9040044
- QQ discussion group: 292495320
Quick-pomelo is based on pomelo, have a draft idea of pomelo framework is required.
Quick-pomelo use memdb for underlaying data storage, so understanding memdb is required.
- Install Node.js
- Install Redis (required for memdb)
- Install MongoDB (required for memdb)
- Install MemDB globally
sudo npm install -g memdb-server
- Install pomelo globally
sudo npm install -g rain1017/pomelo
First copy the template to your working directory. The template contains most common skeletons for a quick-pomelo game server.
Define data models in app/models directory
// app/models/player.js
module.exports = function(app){
var mdbgoose = app.memdb.goose;
var PlayerSchema = new mdbgoose.Schema({
_id : {type : String},
areaId : {type : String},
teamId : {type : String},
connectorId : {type : String},
name : {type : String},
}, {collection : 'players'});
mdbgoose.model('Player', PlayerSchema);
};
Write controllers in app/controllers directory
// app/controllers/player.js
var Controller = function(app){
this.app = app;
};
var proto = Controller.prototype;
proto.createPlayerAsync = function(opts){
// Get model by this.app.models.[model]
var player = new this.app.models.Player(opts);
yield player.saveAsync();
};
proto.removePlayerAsync = function(playerId){
var player = yield this.app.models.Player.findAsync(playerId);
if(!player){
throw new Error('player not exist');
}
yield player.removeAsync();
};
module.exports = function(app){
return new Controller(app);
};
For each type of server, write a route in app/routes directory.
// app/routes/player.js
module.exports = {
// Routing in handler
handler : function(session, method, msg){
// Return a string routing key
// Same routing key always route to the same backend server.
return session.uid || msg.playerId;
},
// Routing in remote
remote : function(routeParam, method, args){
return routeParam;
}
};
Write server handlers in app/servers/[server]/handler
// app/servers/player/playerHandler.js
var Handler = function(app){
this.app = app;
};
Handler.prototype.createPlayer = function(msg, session, next){
// Get controller by this.app.controllers.[controler]
return this.app.controllers.player.createPlayerAsync(msg.opts)
.nodeify(next);
};
module.exports = function(app){
return new Handler(app);
};
Before start
- Make sure Redis and MongoDB has started.
- Create a symbol link to
[PROJECT_ROOT]/config/memdb.conf.js
in~/.memdb/
mkdir ~/.memdb
ln -s [PROJECT_ROOT]/config/memdb.conf.js ~/.memdb/memdb.conf.js
- Start memdb cluster
memdbcluster start
Start server
pomelo start --harmony (--harmony is required on node < 0.4.0)
A typical realtime game server framework is stateful for the sake of performance, all states is kept in server local memory. However, this approach has significant drawbacks:
- Any exceptions (may be caused by bugs or unexpected client input) may result in non-consistent state (half-modified dirty data), which is very difficult to recover
- Concurrency control is very difficult to implement
- We must remember which server the data is located, and use rpc to get data, which is error prone.
- In memory data will be lost on server failure, it's very difficult to support HA
Thanks to MemDB, quick pomelo un-invent the stateful approach and use a web server like 'MVC' based architecture. All servers become stateless and all states is stored in memdb. You can now get all benefits from a typical stateless web server, without losing performance and scalability of in memory stateful server.
Copyright 2015 rain1017.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. See the AUTHORS file for names of contributors.