Transcript

• 1. © 2015, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Greg Khairallah, Business Development Manager for Big Data and Analytics, AWS March 19, 2015 Cost and Performance Optimization in Amazon RDS
• 2. Agenda Instance and Storage Selection Monitoring Reserved Instances
• 3. Instance and Storage Selection
• 4. Instance Selection • RDS offers a wide range of instance classes • Instance types differ on CPU, RAM, network performance, EBS Optimized • Base selection on • Performance • Cost (will differ based on type and region) • Region (not all instance types available in all regions) • Engine (not all instance types available for all engines) • Elastically scale instance size up/down
• 5. Instance Selection: Current Generation Instance Type vCPU Memory (GiB) PIOPS-Optimized Network Performance Standard - Current Generation db.m3.medium 1 3.75 - Moderate db.m3.large 2 7.5 - Moderate db.m3.xlarge 4 15 Yes Moderate db.m3.2xlarge 8 30 Yes High Memory optimized - current generation db.r3.large 2 15 - Moderate db.r3.xlarge 4 30.5 Yes Moderate db.r3.2xlarge 8 61 Yes High db.r3.4xlarge 16 122 Yes High db.r3.8xlarge 32 244 - 10 Gigabit Micro instances db.t2.micro 1 1 - Low to Moderate db.t2.small 1 2 - Low to Moderate db.t2.medium 2 4 - Low to Moderate
• 6. Instance Selection: Previous Generation Instance Type vCPU Memory (GiB) PIOPS-Optimized Network Performance Standard - previous generation db.m1.small 1 1.7 - Low db.m1.medium 1 3.75 - Moderate db.m1.large 2 7.5 Yes Moderate db.m1.xlarge 4 15 Yes High Memory optimized - previous generation db.m2.xlarge 2 17.1 - Moderate db.m2.2xlarge 4 34.2 Yes Moderate db.m2.4xlarge 8 68.4 Yes High db.cr1.8xlarge 32 244 - High Micro instances db.t1.micro 1 0.613 - Very Low
• 7. Instance Selection • Measure existing application workload • Use native database engine tools – system tables, Oracle Enterprise Manager, Statspack, SQL Server Tuning Advisor • Use CloudWatch metrics • Test and scale accordingly
• 8. What is General Purpose (SSD) Storage • New EBS Volume Type - SSD Backed / Solid State Drive • New Default Volume Type on RDS • Available on all RDS DB Instance types and DB engines in all Regions • Flat, Per-Gigabyte Rate. No I/O Charges • Volume bundles Baseline IOPS • 1GB : 3 IOPS. • Scales predictably, linearly, 10:30 ,100:300, 1TB:3K • Burst Capable Volume
• 9. Sizing General Purpose (SSD) volume for databases • Calculate your required IOPS needed in steady state • Provision the GB = (Steady state IOPS) ÷ 3 • For example, customer requiring1500 steady state IOPS • Provision 1500 ÷ 3 = 500 GB • IO bursts will support • Database load or table scan operations • Spike in IO workload
• 10. Provisioned IOPS • Provisioned IOPS storage is a storage option that delivers fast, predictable, and consistent throughput performance • When fast and consistent I/O performance required • Optimized for I/O intensive, online transaction processing (OLTP) workloads that have consistent performance requirements • Important: Recommended for random reads and writes and not sequential workloads
• 11. Activities that Impact Storage Performance • Nightly backups • There is no additional charge for backup storage up to 100% of your provisioned database storage for an active DB Instance • Perform these tasks during your maintenance window • DB snapshot creation • Single-AZ to Multi-AZ modification • Read replica creation
• 12. Caveats • Allocated storage and provisioned storage can be modified • Scaling Storage operations can impact performance until new allocation is finished • Except for SQL Server • Export data • Create new DB instance with larger storage • Import data • Initially size SQL Server storage for future growth to avoid this
• 13. Burst mode – GP2 & T2 GP2 – SSD based Amazon EBS storage • 3 IOPS per GB base performance • Earn credits when usage below base • Burst to 3000+ IOPS T2 – Amazon EC2 instance with burst capability • Base performance + burst • Earn credits per hour when below base performance • Can store up to 24 hours worth of credits • Amazon CloudWatch metrics to see credits & usage
• 14. Burst mode – GP2 & T2
• 15. T2 – CPU credits
• 16. Burst mode vs. Standard vs. Provisioned IOPS 0 1000 2000 3000 4000 5000 6000 7000 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 TransactionsperSecond(TPS) Hours 100% Read - 20GB data Series1 $0.575 per hour
• 17. Burst mode vs. Standard vs. Provisioned IOPS 0 1000 2000 3000 4000 5000 6000 7000 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 TransactionsperSecond(TPS) Hours 100% Read - 20GB data Series1 Series2 $0.575 per hour $0.408 per hour
• 18. Burst mode vs. Standard vs. Provisioned IOPS 0 1000 2000 3000 4000 5000 6000 7000 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 TransactionsperSecond(TPS) Hours 100% Read - 20GB data Series1 Series2 Series3 $0.575 per hour $0.408 per hour $0.508 per hour
• 19. Burst mode vs. Standard vs. Provisioned IOPS 0 1000 2000 3000 4000 5000 6000 7000 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 TransactionsperSecond(TPS) Hours 100% Read - 20GB data Series1 Series2 Series3 Series4 $0.105 per hour $0.575 per hour $0.408 per hour $0.508 per hour
• 20. Burst mode vs. Standard vs. Provisioned IOPS 0 1000 2000 3000 4000 5000 6000 7000 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 TransactionsperSecond(TPS) Hours 100% Read - 20GB data Series1 Series2 Series3 Series4 Series5 $0.105 per hour $0.575 per hour $0.233 per hour $0.408 per hour $0.508 per hour
• 21. Scale and Performance
• 22. How to scale with Amazon RDS • Scale up or down with resizable instance types • CPU to 32 vCPUs, RAM to 244 GiB • Scale your storage up with a few clicks • Scale while online • Easy conversion from standard to Provisioned IOPS storage • Put a cache in front of Amazon RDS • Amazon ElastiCache for Memcached or Redis • Or your favorite cache, self-managed in Amazon EC2 • Offload read traffic to Read Replicas
• 23. Reads10% Reads10% Reads10% Reads10% Reads10% Reads10% Reads10% Reads10% Reads10% Writes10% Reads10% Reads10% Writes10% Reads10% Reads10% Writes10% Reads10% Reads10% Writes10% Reads10% Reads10% Writes10% Reads10% Reads10% Writes10% Reads10% Reads10% Writes10% Reads10% Reads10% Writes10% Reads10% Reads10% Writes10% Reads10% Writes10% Reads10% Reads 90% Writes10% Read to write ratio Primary Writes10% Replica1 Writes10% Replica2 Writes10% Replica3 Writes10% Replica4
• 24. Writes 20% Writes 20% Writes 20% Writes 20% Reads 80% Reads 20% Reads 20% Reads 20% Reads 20% Reads 20% Writes 20% Reads 20% Writes 20% Reads 20% Writes 20% Reads 20% Writes 20% Writes 20% Writes20% Read to write ratio – More writes Primary Replica1 Replica2 Replica3 Replica4
• 25. Series1 Series2 Series3 Series4 Series5 1 2 3 4 5 6 7 8 1 2 3 4 5 6 Scale Scale based on % Write
• 26. InnoDB cache warming 4x Load buffer pool Dump buffer pool
• 27. 0 5000 10000 15000 20000 25000 30000 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 TransactionsperSecond Time (sec) Workload with 50/50 R/W ratio Unwarmed Cache Warmed Cache 4X 9 min
• 28. To enable InnoDB cache warming…
• 29. Monitoring
• 30. CloudWatch • Monitor AWS resources in real-time • Collect and track metrics • Metrics data provide visibility into performance, storage, number of database connections, etc. • Send notifications based on the monitoring rules that you define
• 31. CloudWatch Alert Options • RDS Metrics • CPU Utilization - Track the percentage of CPU utilization • Free Storage Space – Track available storage space • Free Memory – Amount of Available RAM • Database Connections – Number of databases connections in use
• 32. CloudWatch Alert Options • RDS Metrics • IOPS – Number of IO operations completed per second • Latency – Elapsed time between the submission of an I/O and it’s submission. • Throughput – Number of bytes per second transferred to or from disk • Queue Depth – Number of I/O requests in the queue waiting to be serviced
• 33. CloudWatch Alert Options • Queue Depth – Number of I/O requests in the queue waiting to be serviced • Queue depth of 5 for every 1000 IOPS • If less, not enough I/O generated • If more, you may want to increase IOPS
• 34. CloudWatch Alert Options Metric Actions CPU Utilization Scale Compute Freeable Space Scale Storage Up Freeable Memory and Swap Usage Scale Compute Write Latency Add Provisioned IOPS DB Connections Check connection pooling
• 35. CloudWatch Alert Options
• 36. Cloudwatch Alert Options • A number of options to get alerted – SMS, SQS, HTTPS, Email-JSON, Email………
• 37. Database Parameter Groups DB parameter groups act as a container for engine configuration values that are applied to one or more DB instances.
• 38. Reduce Costs
• 39. Save money with RDS Reserved Instances Pay a low up-front fee to get a lower hourly price on database instances for a 1- or 3-year term The lower-price entitlement applies to any running instance matching its description Choose among Heavy, Medium, Light RIs Month 1 Month 2 Month 3 Month 4 Month 5 Month 6 Month 7 Month 8 Month 9 Month 10 Month 11 Month 12 On-demand 544.50 1,089.00 1,633.50 2,178.00 2,722.50 3,267.00 3,811.50 4,356.00 4,900.50 5,445.00 5,989.50 6,534.00 Heavy RI 1-yr 1,508.10 1,750.20 1,992.30 2,234.40 2,476.50 2,718.60 2,960.70 3,202.80 3,444.90 3,687.00 3,929.10 4,171.20 Start saving here
• 40. Thank You!