### Scaling with PostgreSQL 9.6 and Postgres-XL Mason Sharp mason.sharp@gmail.com Huawei ## whoami • Engineer at Huawei (USA based) • Co-organizer of NYC PostgreSQL User Group • PostgreSQL-based PostgreSQL-based database clusters GridSQL • Postgres-XC • Postgres-XL ## Agenda • PostgreSQL 9.6 Scaling Features • Scaling Solutions • Use Cases • Postgres-XL ## BTW- PGConf US 2017 in New York City • Join us

## Scaling a Multi-Tenant k8s Cluster in a Telco Pablo Moncada eBPF Summit October 28, 2020 ## About MasMovil group • 4th telecom company in Spain - Provides voice and broadband services to +12M customers

2 Import Dependencies 2.2.3 Start Server 2.2.4 Use ShardingSphere-Proxy 2.3 ShardingSphere-Scaling (Experimental) 2.3.1 Rule Configuration 2.3.2 Import Dependencies 2.3.3 Start Server 2.3 Supported...56 Supported...56 4.7 Scaling...56 4.7.1 Background...56 4.7.2 Challenges...56 4.7.3 Goal...57 4.7.4 Status...57 4.7.5 Core Concept...57 Scaling Job...57 Inventory Data...57 ShardingSphere-Sidecar ..... 210 5.3.1 Introduction ..... 210 5.3.2 Comparison ..... 211 5.4 ShardingSphere-Scaling ..... 211 5.4.1 Introduction ..... 211 5.4.2 Build ..... 212 Build&Deployment ..... 212

/p9_6.jpg)  ## Scaling and orchestration Containerized development and production environment Cloud orchestration with 98f7/p24_1.jpg) ## Evaluation & conclusions ## Evaluation Kubernetes and docker for fast scaling Low price tag on operations No bottlenecks throughout the season BAM!  ## Conclusion • Documentation and procedures Prometheus • Scaling • Best practice Locust ## Thank you for the attention Any questions? ![Image](/uploads/docum

workload and system performance • Identify bottleneck operators, straggler workers, skew • Enumerate scaling actions, predict their effects, and decide which and when to apply ## Mechanism: How to apply the manner • Block and unblock computations to ensure result correctness ## Automatic Scaling Control ## The automatic scaling problem logical dataflow  metrics detect symptoms scaling controller policy decide whether to scale scaling action

Unsupported ..... 58 4.7 Scaling ..... 58 4.7.1 Background ..... 58 4.7.2 Challenges ..... 58 4.7.3 Goal ..... 59 4.7.4 Status ..... 59 4.7.5 Core Concept ..... 59 Scaling Job ..... 59 Inventory databases and users in the same log ..... 216 To split into different log files ..... 216 5.2.5 Scaling ..... 217 Introduction ..... 217 Build ..... 217 Manual ..... 221 5.3 ShardingSphere-Sidecar Provider 246 6.10.3 XADataSourceDefinition 247 6.10.4 DataSourcePropertyProvider 247 6.11 Scaling 247 6.11.1 ScalingEntry 247 6.11.2 JobCompletionDetectAlgorithm 248 6.11.3 DataConsistencyCalculateAlgorithm

2 Import Dependencies 2.2.3 Start Server 2.2.4 Use ShardingSphere-Proxy 2.3 ShardingSphere-Scaling (Experimental) 2.3.1 Rule Configuration 2.3.2 Import Dependencies 2.3.3 Start Server 2.3 Integration ..... 58 4.6 Scaling ..... 59 4.6.1 Background ..... 59 4.6.2 Challenges ..... 60 4.6.3 Goal ..... 60 4.6.4 Status ..... 60 4.6.5 Core Concept ..... 60 Scaling Job ..... 60 Inventory ShardingSphere-Sidecar ..... 211 5.3.1 Introduction ..... 211 5.3.2 Comparison ..... 211 5.4 ShardingSphere-Scaling ..... 212 5.4.1 Introduction ..... 212 5.4.2 Build ..... 212 Build&Deployment ..... 212

2 Import Dependencies 2.2.3 Start Server 2.2.4 Use ShardingSphere-Proxy 2.3 ShardingSphere-Scaling(Alpha) 2.3.1 Rule Configuration 2.3.2 Import Dependencies 2.3.3 Start Server 2.3.4 Create 3.5 Scaling ..... 79 3.5.1 Background ..... 79 3.5.2 Introduction ..... 79 3.5.3 Challenges ..... 79 3.5.4 Goal ..... 80 3.5.5 Status ..... 80 3.5.6 Core Concept ..... 80 Scaling Job . 3.2 Comparison 269 4.4 ShardingSphere-Scaling 269 4.4.1 Introduction 269 4.4.2 Build 270 Build&Deployment 270 Shutdown ShardingSphere-Scaling 271 Configuration 271 4.4.3 Manual 271

Logging …… 198 区分同一日志中的 database …… 198 区分同一日志中的 database 和用户 …… 199 拆分为不同的日志文件 …… 199 5.2.5 Scaling …… 200 简介 …… 200 运行部署 …… 200 使用手册 …… 204 5.3 ShardingSphere-Sidecar …… 216 5.3.1 简介 …… ShardingSphere-Scaling 是一个提供给用户的通用数据接入迁移及弹性伸缩的解决方案。 #### 4.7.3 目标 支持自定义分片算法，减少数据伸缩及迁移时的业务影响，提供一站式的通用弹性伸缩解决方案，是Apache ShardingSphere弹性伸缩的主要设计目标。 #### 4.7.4 状态 ShardingSphere-Scaling 从 4.1.0 版本开始向用户提供。当前处于 Sharding Scaling Rule CREATE SHARDING SCALING RULE scalingName [scalingRuleDefinition] DROP SHARDING SCALING RULE scalingName ENABLE SHARDING SCALING RULE scalingName DISABLE SHARDING SCALING RULE scalingName

gptq_marlin_24,gptq_marlin,gptq,squeezellm,compressed-tensors,bitsandbytes,None} →] [--rope-scaling ROPE_SCALING] [--rope-theta ROPE_THETA] [--enforce-eager] [--max-context-len-to-capture MAX_CONTEXT_ LORA_EXTRA_VOCAB_ →SIZE] [--lora-dtype {auto,float16,bfloat16, →float32}] [--long-lora-scaling-factors LONG_LORA_ →SCALING_FACTORS] [--max-cpu-loras MAX_CPU_LORAS] [--fully-sharded-loras] [--device {auto cache scaling factors. This should generally be supplied, when KV cache dtype is FP8. Otherwise, KV cache scaling factors default to 1.0, which may cause accuracy issues. FP8_E5M2 (without scaling) is only