Laravel 如何生成一千万条数据表记录
2017-03-07 本文已影响992人
zshanjun
Laravel 批量填充数据流程
一般我们使用laravel填充数据,一般按照下面的流程:
1、使用model factory定义要填充的数据结果
2、在DatabaseSeeder.php里面进行调用填充
示例:
//model factory
$factory->define(NewProduct::class, function (Faker\Generator $faker) {
$fabrication_ids = Fabrication::lists('id')->toArray();
$fabric_width_ids = FabricWidth::lists('id')->toArray();
$fabric_weight_ids = FabricWeight::lists('id')->toArray();
$gauge_ids = Gauge::lists('id')->toArray();
$diameter_ids = Diameter::lists('id')->toArray();
$production_property_ids = ProductionProperty::lists('id')->toArray();
$content_ids = Content::lists('id')->toArray();
$creator_ids = Account::lists('FID')->toArray();
return [
'code' => $faker->word,
'name' => $faker->word,
'name_en' => $faker->word,
'fabrication_id' => $faker->randomElement($fabrication_ids),
'fabric_width_id' => $faker->randomElement($fabric_width_ids),
'fabric_weight_id' => $faker->randomElement($fabric_weight_ids),
'gauge_id' => $faker->randomElement($gauge_ids),
'diameter_id' => $faker->randomElement($diameter_ids),
'production_property_id' => $faker->randomElement($production_property_ids),
'content_id' => $faker->randomElement($content_ids),
'creator_id' => $faker->randomElement($creator_ids),
'remark' => $faker->text(200),
];
});
//DatabaseSeeder.php
factory(NewProduct::class, 1000)->create();
以上代码,存在一个很严重的问题,create()方法会每创建一条记录都执行一条sql语句。脚本运行大概需要40s。
Laravel 优化批量填充数据
1、新建一个批量填充的文件,这里以NewProductSeeder.php为例
2、在DatabaseSeeder.php里进行调用
示例:
//NewProductSeeder.php
//获取相关联属性的值,为防止数据过多,限制每次只取10000条
//为加快获取数据速度,在有索引的id列上进行排序
$fabrication_ids = Fabrication::orderBy('id', 'desc')->take(10000)->lists('id')->toArray();
$fabric_width_ids = FabricWidth::orderBy('id', 'desc')->take(10000)->lists('id')->toArray();
$fabric_weight_ids = FabricWeight::orderBy('id', 'desc')->take(10000)->lists('id')->toArray();
$gauge_ids = Gauge::orderBy('id', 'desc')->take(10000)->lists('id')->toArray();
$diameter_ids = Diameter::orderBy('id', 'desc')->take(10000)->lists('id')->toArray();
$production_property_ids = ProductionProperty::orderBy('id', 'desc')->take(10000)->lists('id')->toArray();
$content_ids = Content::orderBy('id', 'desc')->take(10000)->lists('id')->toArray();
$creator_ids = Account::lists('FID')->toArray();
//利用两层循环,脚本每次运行总共生成20000条记录
//利用一个临时数组来批量保存临时数据
//数组不宜设置过大,否则会导致异常
foreach (range(0, 10) as $index) {
$products = [];
$date = $faker->dateTime;
foreach (range(0, 2000) as $index) {
$products[] = [
'code' => $faker->word,
'name' => $faker->word,
'name_en' => $faker->word,
'fabrication_id' => $faker->randomElement($fabrication_ids),
'fabric_width_id' => $faker->randomElement($fabric_width_ids),
'fabric_weight_id' => $faker->randomElement($fabric_weight_ids),
'gauge_id' => $faker->randomElement($gauge_ids),
'diameter_id' => $faker->randomElement($diameter_ids),
'production_property_id' => $faker->randomElement($production_property_ids),
'content_id' => $faker->randomElement($content_ids),
'creator_id' => $faker->randomElement($creator_ids),
'remark' => $faker->text(200),
'created_at' => $date,
'updated_at' => $date
];
}
NewProduct::insert($products);
}
//DatabaseSeeder.php
$this->call(NewProductSeeder::class);
填充一千万条记录
利用上面的脚本,只需要结合linux里面的cron job即可很简单的实现填充一千万条记录的目标
1、添加任务
在app/Console/Kernel.php文件里面添加上任务:
protected function schedule(Schedule $schedule)
{
$schedule->exec('php artisan db:seed')
->everyMinute()
->withoutOverlapping();
}
2、添加crontab任务
在linux里面输入crontab -e
打开crontab文件,添加下面代码,保存退出即可运行代码:
* * * * * php /home/vagrant/Code/sfabric/artisan schedule:run >> /dev/null 2>&1
3、坐等成功
按照上面的填充速度(一分钟填充4w条记录),需要250分钟即可。
注意,实际使用时请根据上面脚本的执行速度,来动态调整每分钟生成的记录条数(调整最外层range即可)
(这个例子中,实际代码涉及到更多插入,4w条主表记录加8w条关联表记录加3.5k条其它属性记录,只需要50s)。
tips:为加快速度,可在批量添加时,提前删除索引,添加完成后再添加索引。
参考网站: