Gear6 Solution Brief: Database Applications

Introduction

Businesses strive for predictable and scalable performance from databases to establish a consistent platform for enterprise applications. As businesses grow, demands on the databases increase and require higher levels of performance. Delivering increased performance cost-effectively has been a challenge for database, computing, and storage vendors alike. Centralized storage caching represents an innovative approach to address scalable database performance while complementing existing infrastructure.

The Challenges

A large number of enterprise applications rely on relational databases that have On Line Transaction Processing (OLTP), On Line Analytical Processing (OLAP), or a mix of both. Relational databases often have instances running on different servers, such as with Oracle Real Application Clusters, that simultaneously process queries to the repository’s data and log files. Each of these instances can have hundreds or thousands of database client connections accessing it. Ultimately this leads to a high number of concurrent access points to a single data set.

Disk-based storage systems that rely on adding spindles to compensate for low throughput partially address enterprise I/O requirements, but these systems cannot solve the latency requirements critical to OLTP and OLAP environments. As a result databases can experience the following scenarios:

• Slow database response, including database contention or deadlock
• Poor application performance
• Excessive tuning required just to maintain minimum operation

Gear6 Solution for Database Applications

Gear6 addresses the need for database performance with centralized storage caching, an approach that complements existing disk-based storage solutions with scalable, high-capacity caching appliances. This solution offers a number of benefits:

• Serving data from low latency memory as opposed to disk reduces the access time from milliseconds to microseconds, accelerating databases and enabling faster application processing
• Scalable caching appliances serve data in parallel through multiple cache channels, eliminating disk contention during read-intensive transactional or query access
• Efficient caching algorithms dynamically place frequently accessed data in cache reducing the administrative overhead to sustain peak load operations
• Throughput and access time achieved through caching eliminates the need to over-provision disk spindles
• Read intensive database workloads, with their random access patterns, can be dramatically accelerated with caching