August 2014 Summaries

Many developers initially use Memcache for caching data in relational databases to improve performance. However, when the volume of traffic exceeds the capacity of a single Memcache instance, strategies like sharding are often employed. Sharding can lead to issues such as unpredictable cache sizes, difficulty in managing multiple instances, lack of automated failover, and absence of data persistence. Aerospike offers an alternative approach by shifting to a single cluster architecture without the need for caching layers or separate connections to relational databases. This provides benefits like automatic failover, replication of data, predictable performance, lower costs using SSDs, and simpler application development and operations.

The text discusses the challenges and limitations of using Memcache as a caching solution for large-scale applications, such as those at Snapdeal. The authors highlight issues with sharding, scaling, and data management, which can lead to problems like cache size, hardware costs, sharding complexity, failover, persistence, and development complications. In contrast, the Aerospike cluster provides automatic and immediate failover, replication of data, predictable performance, lower cost using SSDs, and simpler application development, making it a more attractive alternative for replacing or augmenting Memcache and relational databases.