Unified Redis Platform (URP)

Motivation

With the increase of Cloud Adoption, cost optimization has become a concern for organizations moving to the Cloud. Upon conducting a cost analysis of our Azure Subscriptions, we realized that Azure Caches for Redis were among the highest sources of expenditure. In our ecosystem of Micro-services, each service was using its own isolated Redis Cache for each environment. This is the same of many services and ecosystem across the industry. Furthermore, if you need to satisfy your customer demands of BCDR (Business Continuity Disaster Recovery), you might create replica Redis Caches in alternate regions. In most cases, we analyzed that these Caches were not used up to their full potential (both in terms of load and storage). Our analysis concluded that we were using about 1GB of the P1 Redis Caches. A single P1 cache can store 6GB of data. A single P1 Redis Cache costs around $412 per month. With three micro-services and each service is utilizing two Redis Caches were spending about $2,412 per month.

Such an analysis motivated us to look for opportunities to merge the existing Redis Caches into a single cluster shared by the existing applications. Still, it needed to ensure that each application had proper isolation.

This platform allows multiple application to re-use a single or a cluster of Redis Caches. The platform provides proper isolation required by various apps to operate (get/edit/delete/update cached objects) on a single cluster without interfering (key overrides) with each other.

Capabilities

Consolidate multiple Azure Cache for Redis instances to a single cluster
Automatic key name-spacing for providing isolation between multiple applications
Disaster Recovery by keeping Azure Redis Cache in multiple regions in a single cluster
Multiple read-regions
Geo-replication of data (add/delete/update) across multiple regions
Support for all operations available in StackExchange.Redis library. You can use all methods and data types from StackExchange.Redis without any code change.
Support for auto-retires and timeouts in case of errors
Automatic lazy and singleton connection for preventing connection exhaustion issues
Secure connection using application-specific private Keys
Telemetry support for Redis Key usage, performance reports and failures
Management console for managing keys from backend