1 Executive Summary Database Proxies: Improves SQL read/write - PowerPoint PPT Presentation

� 1

Executive Summary Database Proxies: • • Improves SQL read/write performance and reliability • Deployment requires no application changes Demo • � 2

IMDG vs. Database Proxies Feature ProxySQL ✓ ✓ ✓ Automated Failover ✓ ✓ Query routing ✓ ✓ ✓ ✓ ✓ Database Vendor Neutral Automated Cache ✓ invalidation Reduces network latency ✓ � 3

IMDG vs. Database Proxies Amazon ElastiCache ProxySQL • May be “good enough” • Best scale & performance • Existing applications, small dev • Greenfield applications • No code changes • Requires code changes � 4

Transparent Database Proxy Click to add text

Software Package Options Application Server Application Server Application Application Vendor Database Heimdall Data Driver JDBC driver, .jar JDBC file Vendor JDBC Driver Runs as an Heimdall Data Proxy agent Any JDBC data source � 6

Database Proxy Platform APPLICATION AUTOMATED BATCH SQL CACHING FAILOVER PROCESSING READ/WRITE CONNECTION HTAP SPLITS POOLING HEIMDALL DATABASE PROXY PLATFORM Aurora, RDS, Redshift � 7

Heimdall Transparent Deployment Application Servers Amazon Aurora Heimdall Application RDS, Redshift Heimdall Data Heimdall Application Application Server Data Database Application Server Application Application Server Proxy Heimdall Application Server Central Console ElastiCache SQL Analytics Audit Logging � 8

Heimdall Centralized Deployment Database Proxy Tier Application SQL Application Server Heimdall Heimdall DB Proxy Heimdall DB Proxy DB Proxy Application Heimdall Application Server Central Console � 9

Use Cases

Caching and Read/Write Splits SQL SQL Heimdall DB Application Proxy Write Local Cache App VM 1 L2 Cache Read 1 App VM 2 Local Cache Read 2 SQL SQL Heimdall DB Application Proxy � 11

How Caching Works Uses real-time analysis and statistics on: • Query frequency and variability • Relative performance of Cache vs. Database Provides: • Auto-cache only if there is a performance benefit • Cache recommendations and benefits � 12

Batch Processing Data DML Generator Request START TRANSACTION; #1 DML 1; DML 2; DML 3; DML 4; #2 #4 COMMIT; Queue 8 7 6 5 4 3 2 1 Exceptions are logged, removed from batch, and Benefits: • Lower CPU overhead due to fewer commits #3 transaction restarted 4 3 2 1 • Improved application response time Batch Size 4 • Improved DML scale � 13

HTAP: Intelligent OLTP/OLAP Routing Heimdall Data Data SQL Traffic Manager Amazon Redshift Consumers Analytics OLTP Materialized View management � 14

SQL Analytics Very cacheable. 700 µ s per query � 15

IMDG vs. Database Proxies Amazon ElastiCache ProxySQL • Good enough • Best scale & performance • Existing applications, small dev • Greenfield applications • No code changes • Requires code changes � 16

� 17