引言
手头上的项目有一些采用django框架编写, 如果说并发量比较小的时候简单的runserver是可以应对的.
那么当并发达到一两千的时候,该怎么提高django的并发能力呢?
Overview
- 环境说明:
- python: 3.5
- django: 1.8.2
- gunicorn: 19.7.1
- 系统:
- 服务器: centos 4核
- 压测机器: centos 4核
- 压测环境
- siege/ysab
- 4核centos测试机
- 为什么用django
- 开发效率高
- 好上手
- 关于gunicorn
- Gunicorn 'Green Unicorn' is a Python WSGI HTTP Server for UNIX.It's a pre-fork worker model. The Gunicorn server is broadly compatible with various web frameworks, simply implemented, light on server resources, and fairly speedy.(这是官方给出的回答)
压测方式及命令
-
压测方式:
-
压测命令:
- siege: siege -c255 -t200S -v -b 'http://B_ip:8080/test POST appid=111'
- ysab: ysab -n 900 -m POST -u http://B_ip:8080/test -d '{"appid": "111", "other": "other"}'
- 备注: 欢迎使用ysab, ysab文档
本次实验业务场景
代码展示
settings部分
# 这里我们用mysql,其他配置都是默认
DATABASES = {
'default': {
'ENGINE': 'django.db.backends.mysql',
'NAME': 'ce',
'USER': 'root',
'PASSWORD': '',
'HOST': '192.168.96.95',
'PORT': '3306',
# 'CONN_MAX_AGE': 600,
}
}
models部分
class Test(models.Model):
url = models.CharField(max_length=228, blank=True, null=True)
img_url = models.CharField(max_length=228, blank=True, null=True)
title = models.CharField(max_length=228, blank=True, null=True)
content = models.CharField(max_length=228, blank=True, null=True)
class Meta:
db_table = 'test'
verbose_name = "test表"
def __unicode__(self):
return self.id
views部分
class Test(APIView):
def post(self, requsts):
Test.objects.create(
**{'url': str(1000000 * time.time())})
return Response({"status": 200})
开始压测
数据说明
目前数据库test表的数据量是, 其中id是自增主键
MySQL [ce]> select id from test order by id desc limit 2;
+--------+
| id |
+--------+
| 627775 |
| 627774 |
+--------+
runserver 方式压测结果
Lifting the server siege... done.
Transactions: 24041 hits
Availability: 99.93 %
Elapsed time: 99.60 secs
Data transferred: 0.32 MB
Response time: 1.03 secs
Transaction rate: 241.38 trans/sec # 并发量只有241
Throughput: 0.00 MB/sec
Concurrency: 248.94
Successful transactions: 24041
Failed transactions: 16
Longest transaction: 32.55
Shortest transaction: 0.05
gunicorn + gevent (4个worker)
Lifting the server siege... done.
Transactions: 23056 hits
Availability: 100.00 %
Elapsed time: 99.49 secs
Data transferred: 0.31 MB
Response time: 1.09 secs
Transaction rate: 231.74 trans/sec # 并发量只有231
Throughput: 0.00 MB/sec
Concurrency: 252.95
Successful transactions: 23056
Failed transactions: 0
Longest transaction: 8.21
Shortest transaction: 0.01
gunicorn + gthread (4个worker, --threads=50)
启动方式
[官方有相应说明]((http://docs.gunicorn.org/en/latest/settings.html)
gunicorn --env DJANGO_SETTINGS_MODULE=ce.settings ce.wsgi:application -w 4 -b 0.0.0.0:8080 -k gthread --threads 40 --max-requests 4096 --max-requests-jitter 512
压测结果
启动方式:
done.
siege aborted due to excessive socket failure; you
can change the failure threshold in $HOME/.siegerc
Transactions: 28231 hits
Availability: 95.67 %
Elapsed time: 30.71 secs
Data transferred: 0.41 MB
Response time: 0.27 secs
Transaction rate: 919.28 trans/sec # 提高了不少吧,能不能在提高?
Throughput: 0.01 MB/sec
Concurrency: 251.06
Successful transactions: 28231
Failed transactions: 1278 # 但是失败的有些多
Longest transaction: 8.06
Shortest transaction: 0.01
gunicorn + gthread + CONN_MAX_AGE(4个worker, --threads=50)
CONN_MAX_AGE: 复用数据库链接
Lifting the server siege... done.
Transactions: 110289 hits
Availability: 99.62 %
Elapsed time: 99.65 secs
Data transferred: 1.47 MB
Response time: 0.23 secs
Transaction rate: 1106.76 trans/sec # 这次又提升了不少啊
Throughput: 0.01 MB/sec
Concurrency: 253.84
Successful transactions: 110289
Failed transactions: 422
Longest transaction: 3.85
Shortest transaction: 0.01
能不能gunicorn+gevent+CONN_MAX_AGE(4个worker)
这里我不建议使用,这样的话你的数据库连接数会飚的很高,服务会挂的很惨, 毕竟数据库是不会允许
无休止的建立连接的。前边的提高手段无非用的多线程,如果一定要用协程(gevent)的方式呢,能不
能解决数据库连接数过高的问题,而且还能有不错的性能呢?可以看一下这篇文章:
gunicorn+gevent+django数据库连接池
如何再次增加并发量
采用nginx做负载
[图片上传失败...(image-55dc84-1555064167047)]
去掉自增主键
原因很简单,因为自增主键的存在写库存在抢锁, 可以利用全局id生成器提前生成id直接写入数据库
换成异步任务去写库
如果数据只是存在mysql中做备份,建议使用异步的方式写入库,先把数据写到缓存下发给用户,之后在
利用后台异步任务一点点的写入,例如聊天系统可以这样干
换成更高效的框架或者语言
可以试试tornado, 如果tornado依然无法满足,可以尝试使用golango,毕竟golang是以高并发著称,
而且是编译语言,而且基于它的web框架也很容易上手,性能很可观,例如Gin
Gin