Tubocharge your e-commerce site with Cache and Search Technologies

Size: px

Start display at page:

Download "Tubocharge your e-commerce site with Cache and Search Technologies"

稳山
4 years ago
Views:

1 深入探讨 ElastiCache 最佳实践与常见使用场景郑进佳, AWS 解决方案架构师 Kenny Zheng, Solution Architect, Web Services

2 主要内容 ElastiCache 概述零宕机在线扩展 ElastiCache 集群 ElastiCache 安全与加密 ElastiCache 使用场景最佳实践

3 µs 是新的 ms ElastiCache

4 内存 --- 键值 --- 存储支持 Redis Memcached 高性能全托管 ; 零管理 ElastiCache 高可用, 高可靠强化版

5 数据结构低高 ElastiCache and DynamoDB Accelerator (DAX) DynamoDB HDFS RDS S3 CloudSearch and Elasticsearch Service Glacier 高低低请求频率延迟数据量低低低

6 Redis 概览极速大多数命令延迟 <1 ms 内存数据结构服务器开源性能更强 ~200 命令 + Lua 脚本持久性实用数据结构 Strings, lists, hashes, sets, sorted sets, bitmaps, and HyperLogLogs 高可用复制原子操作支持事务简单

7 更多新功能运行 Lua 脚本地理位置查询 Pub/Sub 模式

8 REDIS:6379> SMEMBERS features 1) Easy to deploy & monitor REDIS:6379> hget feature:details deploy-monitor ElastiCache AWS CloudFormation AWS CLI and SDKs AWS Management Console AWS CloudTrail AWS Config CloudWatch alarm SNS AWS Lambda Notification

REDIS:6379> SMEMBERS features 2) Enhanced Redis Engine REDIS:6379> hget

the risk of increased swap usage during syncs and snapshots Dynamic

memory is close to being exhausted Smoother failovers Clusters recover

9 REDIS:6379> SMEMBERS features 2) Enhanced Redis Engine REDIS:6379> hget feature:details enhancements ElastiCache Optimized Swap Memory Mitigate the risk of increased swap usage during syncs and snapshots Dynamic write throttling Improved output buffer management when the node s memory is close to being exhausted Smoother failovers Clusters recover faster as replicas avoid flushing their data to do a full re-sync with the primary

10 REDIS:6379> 1) Kevin McGehee 2) Qu Chen 3) Rajib Dugar SMEMBERS ec-team:open-source:contributors REDIS:6379> hgetall ec-team:open-source:contributions ElastiCache 1) PSYNC2 (Redis 4.0) 2) 3) BGSAVE (Redis 3.2) 4) 5) MIGRATE (Redis 3.0) 6) 7) MASTER TIMEOUT (Redis 2.8) 8) 9) INCREASE 2Billion+ KEYS in a DATASET (Redis 2.8) 10) (from 11) also: 4114, 4250, 3926, 3899

11 Redis 拓扑禁用 Cluster Mode Max Storage 407 GiB Primary Endpoint I Primary Keyspace 纵向扩展 Slot 0 Slot 1 Slot 复制节点启用 Cluster Mode Max Storage 6+ TiB Configuration Endpoint 1 15 主 / 分片 Keyspace 横向扩展 Slot Slot Slot 复制节点

12 Redis 集群模式启用 vs 禁用功能启用禁用故障转移 sec (Non-DNS) 故障转移风险 Writes affected partial dataset (less risk with more partitions) Reads available 性能 Scales with cluster size (90 nodes 15 primaries replicas per shard) 最大连接数 Primaries (65,000 x 15 = 975,000) Replicas (65,000 x 75 = 4,875,000) ~1.5 min (DNS-based) 存储 6+ TiB 407 GB Writes affected on entire dataset Reads available 6 nodes (1 primary replicas) Primary: 65,000 Replicas: (65,000 x 5 = 325,000) 成本案例 : 假设需要使用 175 GB Smaller nodes but more $$ 9 x cache.r3.xlarge ($0.455hr) = $4.095 hr GB Larger nodes less $ 1 X cache.r3.8xlarge = $3.640, 237 GB

13 进一步探讨集群启用模式

14 Redis 集群 : 自动客户端分片哈希分片 / 集群每个键的分片是 CRC16( 键 ) 模每个槽分配到集群的每个分片开发者需要使用 Redis 集探知客户端客户端被重定向到新的分片智能客户端存储映射关系图 S5 S1 S2 Shard S1 = slots Shard S2 = slots Shard S3 = slots Shard S4 = slots Shard S5 = slots S4 S3 Client

Redis 集群架构举例 : 每集群 3 分片 2 只读副本 Redis 集群多可用区一个集群可包含 1 到 15 个分片 Redis 集群 slots 0 5454 slots 5455 10909 slots 0 5454 slots 0 5454 slots

15 Redis 集群架构举例 : 每集群 3 分片 2 只读副本 Redis 集群多可用区一个集群可包含 1 到 15 个分片 Redis 集群 slots slots slots slots slots slots slots slots slots Availability Zone A Availability Zone B Availability Zone C

16 Redis 集群架构分片每个分片包含一个主节点和最多五个复制节点 Redis 集群 Primary Replica Replica slots slots slots slots slots slots slots slots slots Availability Zone A Availability Zone B Availability Zone C

17 Redis 集群架构分片每个分片包含一个主节点和最多五个复制节点 Redis 集群 Replica Primary Replica slots slots slots slots slots slots slots slots slots Availability Zone A Availability Zone B Availability Zone C

18 Redis 集群架构分片每个分片包含一个主节点和最多五个复制节点 Redis 集群 slots slots slots slots slots slots Replica Replica Primary slots slots slots Availability Zone A Availability Zone B Availability Zone C

19 场景 1: 单主节点失效 Redis 集群 slots slots slots slots slots slots slots slots slots Availability Zone A Availability Zone B Availability Zone C

20 场景 1: 单主节点失效故障缓解 : 1. 自动故障检测和复制节点提升 (~15 30 s) 2. 修复故障节点 Redis 集群 slots slots slots slots slots slots slots slots slots Availability Zone A Availability Zone B Availability Zone C

21 场景 2: 大多数主节点故障 Redis 集群 slots slots slots slots slots slots slots slots slots Availability Zone A Availability Zone B Availability Zone C

22 场景 2: 大多数主节点故障故障缓解 : ElastiCache 的 Redis 强化自动故障检测和复制节点提升修复故障节点 Redis 集群 slots slots slots slots slots slots slots slots slots Availability Zone A Availability Zone B Availability Zone C

23 通过备份与恢复来扩容 Pro tip: DR strategy enable CRR on S3 bucket triggering AWS Lambda function to hydrate destination cluster Downtime rdb New writes not in snapshot 3 Shards 5 Shards Step 1 Step 2 Step 3 Step 4 aws elasticache create-snapshot --replication-group-id redisclusterid --snapshot-name sname aws elasticache copy-snapshot --source-snapshot-name sname --target-snapshot-name sname --target-bucket s3ucketname aws elasticache create-replication-group --replication-group-id NewRedisClusterID --snapshot-arns arn:aws:s3:::bucketname/redisbackup-0001.rdb, etc. Once the new cluster is up, update your app with new ElastiCache endpoint, then terminate old cluster

24 零宕机时间 --- 在线分区

25 在线分区零宕机 Shard 1 Shard 2 Shard Simple API aws elasticache modify-replication-group-shard-configuration --replication-group-id rep-group-id --apply-immediately --node-group-count 5 Scale In Out

26 在线分区零宕机 : 向外扩展 Shard 1 Shard 2 Shard , , reads/writes 无应用中断跨分片的统一分布 Shard 4 Shard , ,

27 在线分片零宕机 : 向内扩展 Shard 1 Shard 2 Shard reads/writes 无应用中断跨分片的统一分布 Shard 4 Shard 5

28 在线分片 CW 告警触发内存高 CloudWatch AWS SNS AWS Lambda var params = { ApplyImmediately: true, NodeGroupCount: 5, ReplicationGroupId: rep-group-id, } elasticache.modifyreplicationgroupshardconfiguration(params, function(err, data) { if (err) console.log(err, err.stack); else console.log(data); }); Cluster Resized 35 Shards

29 正常状态 reads reads/ writes search reads AZ1 clients reads 缓存集群 AZ2 关系型数据 search

30 重负载 reads reads/ writes search reads AZ1 clients reads 缓存集群 AZ2 关系型数据 search

31 正常状态自动横向扩展 reads/ writes reads search reads AZ1 clients reads 缓存集群 AZ2 关系型数据 search

32 ElastiCache 安全

33 ElastiCache Redis 参考架构 REDIS:6379> hget feature:details ref-arch Availability Zone A Availability Zone B Availability Zone C

34 ElastiCache Redis 参考架构 REDIS:6379> hget feature:details ref-arch Private Subnet Private Subnet Private Subnet Availability Zone A Availability Zone B Availability Zone C

35 ElastiCache Redis 参考架构 REDIS:6379> hget feature:details ref-arch security group security group security group Private Subnet Private Subnet Private Subnet Availability Zone A Availability Zone B Availability Zone C

36 ElastiCache Redis 参考架构 REDIS:6379> hget feature:details ref-arch Encryption at REST REDIS RDB snapshot security group security group security group Private Subnet Private Subnet Private Subnet Availability Zone A ElastiCache Redis Cluster Availability Zone B Availability Zone C S3 bucket

37 ElastiCache Redis 参考架构 REDIS:6379> hget feature:details ref-arch security group security group security group Public Subnet Public Subnet Encryption In-Transit Redis AUTH Public Subnet Encryption at REST REDIS RDB snapshot security group security group security group Private Subnet Private Subnet Private Subnet Availability Zone A ElastiCache Redis Cluster Availability Zone B Availability Zone C S3 bucket

38 ElastiCache 加密与合规加密传输过程 : 加密客户端和 Redis 服务端的所有通讯包括节点之间通讯存储 : 加密存储在硬盘和 S3 上的备份全托管 : 通过 API 或者控制台创建, 自动发布和更新合规 HIPAA 合规 ElastiCache Redis 包含在 AWS Business Associate Addendum Redis 3.2.6

39 常见使用场景

41 使用场景会话管理数据库缓存 APIs (HTTP 响应 ) IOT 流数据分析 ( 赛选 / 聚合 ) Pub/sub 社交媒体 ( 情感分析 ) 独立数据库 ( 元数据存储 ) 排行榜

42 缓存关系型数据 reads/ writes Clients RDS mysql.lambda_async 对象数据 reads/writes Elastic Load Balancing EC2 ElastiCache Redis write-through S3 reads/ writes DynamoDB 非结构化数据 DDB streams

43 缓存 NoSQL 更小的 NoSQL DB 集群需求 = 低成本更快的数据检索 = 更好的性能 Clients reads Elasticsearch Cluster EC2 reads/ writes Cassandra Cluster MongoDB Cluster

44 使用 ElastiCache 缓存 NoSQL 数据库 Smaller NoSQL DB clusters needed = lower costs Faster data retrieval = better performance reads ElastiCache Redis reads ElastiCache Redis EC2 EC2 reads/ writes MongoDB Cluster reads/ writes Cassandra Cluster DBObject doc = collection.findone(); Cache serialized DBObject in Redis (good) Cache rows in Redis hash (faster/more efficient) ResultSet rs = session.execute(stmt); Cache serialized ResultSet in Redis (good) Cache rows in Redis hash (faster/more efficient)

45 数据源流式数据的的加工 / 处理 Kinesis Streams raw stream 持续的数据赛选 / 加工 AWS Lambda function 1 cleansed stream Kinesis Streams Kinesis Analytics ElastiCache (Redis) Real-time pub/sub 订阅 AWS Lambda function 2

46 数据源使用 Redis 的大数据框架 Collect Process Spark Streaming on EMR Analyze Kinesis EC2 Apache Kafka Apache Storm on EMR AWS Lambda Kinesis app Store ElastiCache S3 Spark on EMR AWS Lambda Custom app AWS IoT

47 移动社交应用 EC2 GEORADIUS Search points of interest API Gateway AWS Lambda ElastiCache Redis GEOADD Update points of interest DDB streams DynamoDB

48 数字化广告 Clients Advertisers Ad network Clickstream (shopping events) Ad slot publishers Ad slot Consumer ElastiCache Redis Ad placement (websites/apps) User visits page Publisher places ad slot for auction Winners bid ad displayed <40 ms Bidders respond with bids Ad network calls for bids

49 游戏实时排行榜需要独特性和订购的热门游戏非常方便使用 Redis 排序集合 ZADD "leaderboard" 1201 "Gollum ZADD "leaderboard" 963 "Sauron" ZADD "leaderboard" 1092 "Bilbo" ZADD "leaderboard" 1383 "Frodo ZREVRANGE "leaderboard" 0-1 1) "Frodo" 2) "Gollum" 3) "Bilbo" 4) "Sauron ZREVRANK "leaderboard" "Sauron" (integer) 3

50 ElastiCache: 最佳实践

51 集群规模最佳实践存储集群需要有足够内存推荐 : 所需内存 + 25% 保留内存 ( Redis) + 预留增长 ( 可选 10%) 使用 eviction 策略和 TTLs 来优化在达到最大内存之前, 使用 CloudWatch 告警纵向或者横向扩展集群使用内存优化节点提高成本效率 ( 支持 R4) 性能不能牺牲性能使用 Redis 基准工具测试性能基准更多 READIOPS 增加复制节点更多 WRITEIOPS 增加分片 ( 横向扩展 ) 更多 network IO 使用网络优化的实例和横向扩展使用管道来批量读写数据结构命令考虑 Big(O) 时间复杂度集群独立 ( 应用共享键空间 ) 为你的应用选择合适的策略基于工作负载和环境识别哪种类型的隔离是最适合你的隔离 : 不隔离 $ 基于应用类型隔离 $$ 全隔离 $$$

52 Redis 基准工具开源工具测试性能基准 example: src/redis-benchmark -h r3-xlarge-perf.foio use1.cache.amazonaws.com -p n d 100 Syntax: redis-benchmark -h <host> -p <port> -c 50 -n d 500 q -c <clients> Specifies the number of parallel connections (default 50). -n <requests> Specifies the number of requests (default ). -d <size> Specifies the data size of GET and SET values in bytes. -t <test1,test2> Comma-separated list of tests to perform. -q Quiet operation, displays only the result.

53 Redis 最大内存策略基于你的应用负载需求选择最大内存策略 noeviction: return errors when the memory limit has been reached and the client is trying to execute commands that might result in more memory to be used allkeys-lru: evict keys trying to remove the less recently used (LRU) keys first volatile-lru: evict keys trying to remove the less recently used (LRU) keys first, but only among keys that have an expire set allkeys-random: evict random keys to make space for the new data added volatile-random: evict random keys to make space for the new data added, but only evict keys with an expire set volatile-ttl: evict only keys with an expire set, and try to evict keys with a shorter time to live (TTL) first

54 关键的 ElastiCache CloudWatch 指标 CPUUtilization Memcached up to 90% ok Redis divide by cores (ex: 90% / 4 = 22.5%) SwapUsage low CacheMisses/CacheHits Ratio low/stable Evictions near zero Exception: Russian-doll caching CurrConnections stable 设置 CloudWatch 告警指标

55 ElastiCache 可修改的参数 Maxclients: (unchangeable) Use connection pooling timeout closes a connection after it has been idle for a given interval tcp-keepalive detects dead peers given an interval Databases: 16 (default) for non-clustered mode Logical partition Reserved-memory: 25% (default) Recommended 50% of maxmemory to use before % after ElastiCache Maxmemory-policy: The eviction policy for keys when maximum memory usage is reached Possible values: volatile-lru, allkeys-lru, volatile-random, allkeys-random, volatile-ttl, noeviction

56 缓存的诀窍明白潜在数据变更的频率为每个键设置基于适合那个频率的 TTLs 值根据应用的需求, 选择合适的移除策略根据应用目的隔离集群 ( 例如, 缓存集群, 队列, 独立数据库, 等等 ) 通过写抛出保持集群更新性能测试调整集群大小监控 HIT/MISS 命中率来告警使用故障转移的 API 来测试应用弹性

57 Thank You!

1 o o o CPU o o o o o SQL Server 2005 o CPU o o o o o SQL Server o Microsoft SQL Server 2005

1 o o o CPU o o o o o SQL Server 2005 o CPU o o o o o SQL Server o Microsoft SQL Server 2005 1 1...3 2...20 3...28 4...41 5 Windows SQL Server...47 Microsoft SQL Server 2005 DBSRV1 Microsoft SQL Server