Couchbase性能优化 #

优化层次：
├── 应用层
│   ├── 查询优化
│   ├── 连接池管理
│   └── 批量操作
├── 数据层
│   ├── 索引设计
│   ├── 文档设计
│   └── 数据分布
├── 配置层
│   ├── 内存配置
│   ├── 线程配置
│   └── 超时配置
└── 架构层
    ├── 集群规划
    ├── 分片策略
    └── 缓存策略

EXPLAIN SELECT * FROM `my-bucket`.`_default`.`_default`
WHERE type = 'user' AND city = '北京';

SELECT * FROM `my-bucket`.`_default`.`_default`
WHERE type = 'user';

SELECT * FROM `my-bucket`.`_default`.`_default`
WHERE type = 'user' AND city = '北京'
LIMIT 100;

CREATE INDEX idx_users_city_name_email 
ON `my-bucket`.`_default`.`_default`(city, name, email)
WHERE type = 'user';

SELECT city, name, email
FROM `my-bucket`.`_default`.`_default`
WHERE type = 'user' AND city = '北京';

SELECT u.name, o.order_id
FROM `my-bucket`.`_default`.`_default` u
JOIN `my-bucket`.`_default`.`_default` o 
    ON META(u).id = o.user_id
WHERE u.type = 'user' AND o.type = 'order'
    AND u.city = '北京';

CREATE INDEX idx_orders_user_id 
ON `my-bucket`.`_default`.`_default`(user_id)
WHERE type = 'order';

SELECT * FROM `my-bucket`.`_default`.`_default`
WHERE type = 'user'
ORDER BY name
LIMIT 10 OFFSET 10000;

SELECT * FROM `my-bucket`.`_default`.`_default`
WHERE type = 'user' AND name > 'last_name'
ORDER BY name
LIMIT 10;

from couchbase.options import QueryOptions

result = cluster.query('''
    SELECT * FROM `my-bucket`.`_default`.`_default`
    WHERE type = 'user' AND city = $city
''', QueryOptions(named_parameters={'city': '北京'}))

CREATE INDEX idx_users_city 
ON `my-bucket`.`_default`.`_default`(city)
WHERE type = 'user';

CREATE INDEX idx_users_city_age 
ON `my-bucket`.`_default`.`_default`(city, age)
WHERE type = 'user';

CREATE INDEX idx_users_city_age_name 
ON `my-bucket`.`_default`.`_default`(city, age, name)
WHERE type = 'user';

CREATE INDEX idx_composite 
ON `my-bucket`.`_default`.`_default`(a, b, c);

SELECT * FROM t WHERE a = 1;
SELECT * FROM t WHERE a = 1 AND b = 2;
SELECT * FROM t WHERE a = 1 AND b = 2 AND c = 3;
SELECT * FROM t WHERE b = 2;

CREATE INDEX idx_active_users 
ON `my-bucket`.`_default`.`_default`(name)
WHERE type = 'user' AND status = 'active';

SELECT 
    name,
    num_docs,
    num_requests,
    avg_scan_time,
    cache_hits,
    cache_misses
FROM system:indexes
WHERE keyspace_id = 'my-bucket';

ALTER INDEX idx_users_city 
ON `my-bucket`.`_default`.`_default`
WITH {"action": "cleanup"};

内存分配原则：
总内存 = 系统预留 + Data服务 + Index服务 + Query服务 + 其他服务

示例（16GB服务器）：
系统预留：4GB
Data服务：8GB
Index服务：2GB
Query服务：1GB
其他服务：1GB

/opt/couchbase/bin/couchbase-cli bucket-edit \
    --cluster localhost:8091 \
    --username Administrator \
    --password your-password \
    --bucket my-bucket \
    --bucket-ramsize 2048

/opt/couchbase/bin/couchbase-cli bucket-edit \
    --cluster localhost:8091 \
    --username Administrator \
    --password your-password \
    --bucket my-bucket \
    --bucket-eviction-policy valueOnly

工作集 = 活跃数据大小

建议：
工作集大小 < 内存配额 × 70%

计算：
活跃用户数 × 平均文档大小 = 工作集大小

from couchbase.cluster import Cluster, ClusterOptions
from couchbase.auth import PasswordAuthenticator
from couchbase.options import ClusterOptions, ClusterTimeoutOptions
from datetime import timedelta

options = ClusterOptions(
    PasswordAuthenticator('Administrator', 'password'),
    timeout_options=ClusterTimeoutOptions(
        connect_timeout=timedelta(seconds=10),
        kv_timeout=timedelta(seconds=5),
        query_timeout=timedelta(seconds=30)
    )
)

cluster = Cluster('couchbase://node1,node2,node3', options)

class CouchbaseConnection:
    _instance = None
    
    def __new__(cls):
        if cls._instance is None:
            cls._instance = super().__new__(cls)
            cls._instance.cluster = Cluster(
                'couchbase://localhost',
                ClusterOptions(PasswordAuthenticator('Administrator', 'password'))
            )
        return cls._instance
    
    def get_collection(self, bucket, scope, collection):
        return self._instance.cluster.bucket(bucket).scope(scope).collection(collection)

建议配置：
- 最小连接数：CPU核心数 × 2
- 最大连接数：CPU核心数 × 4
- 连接超时：5-10秒
- 空闲超时：60秒

from couchbase.options import GetOptions

keys = ['user::001', 'user::002', 'user::003']
results = collection.get_multi(keys)

from couchbase.options import UpsertOptions

docs = {
    'user::001': {'name': '张三'},
    'user::002': {'name': '李四'},
    'user::003': {'name': '王五'}
}
results = collection.upsert_multi(docs)

import asyncio
from acouchbase.cluster import Cluster

async def batch_insert(docs):
    cluster = await Cluster.connect(
        'couchbase://localhost',
        ClusterOptions(PasswordAuthenticator('Administrator', 'password'))
    )
    
    bucket = cluster.bucket('my-bucket')
    collection = bucket.default_collection()
    
    tasks = [collection.upsert(k, v) for k, v in docs.items()]
    await asyncio.gather(*tasks)

asyncio.run(batch_insert(docs))

推荐配置（5节点）：
Node 1-3: Data, Index, Query
Node 4-5: Eventing, Search, Analytics

或者：
Node 1-3: Data
Node 4-5: Index, Query

/opt/couchbase/bin/couchbase-cli bucket-edit \
    --cluster localhost:8091 \
    --username Administrator \
    --password your-password \
    --bucket my-bucket \
    --compaction-db-percentage 30 \
    --compaction-view-percentage 30 \
    --enable-compaction-parallel true

/opt/couchbase/bin/couchbase-cli setting-cluster \
    --cluster localhost:8091 \
    --username Administrator \
    --password your-password \
    --cluster-max-parallel-indexers 4

curl -X GET http://localhost:8091/pools/default/buckets/my-bucket/stats \
    -u Administrator:password | jq '.op.samples'

SELECT * FROM system:completed_requests
WHERE elapsed_time > 1000000000
ORDER BY elapsed_time DESC;

/opt/couchbase/bin/cbstats localhost:11210 all \
    -u Administrator -p password \
    -b my-bucket

症状：
- 响应延迟增加
- 频繁GC
- 操作超时

解决：
1. 增加内存配额
2. 优化淘汰策略
3. 减少工作集大小
4. 扩展集群节点

症状：
- 写入延迟高
- 队列积压
- 磁盘使用率高

解决：
1. 使用SSD
2. 调整压缩策略
3. 增加节点
4. 优化写入模式

症状：
- 查询响应慢
- CPU使用率高

解决：
1. 创建合适的索引
2. 优化查询语句
3. 使用覆盖索引
4. 分页查询