Cost, Performance, Cache: Boosting Database Workloads Affordably with Amazon ElastiCache

Speed is at the forefront of technological innovation; users and customers of any online service have increasingly low tolerance for latency issues. Often, users will abandon a website if it takes too long to load—even if that load time is delayed by mere seconds. The natural response is to increase database performance; with those improvements, however, the costs quickly become detrimental.

DBTA recently hosted a webinar, sponsored by AWS, titled “Optimize Cost and Boost Performance of Your Database Workloads With Amazon ElastiCache”, outlining the ways in which enterprises can boost performance of databases without breaking the bank. Speakers Damon LaCaille and Steven Hancz, both senior in-memory databases specialist solutions architects at AWS Databases, highlight Amazon ElastiCache as the solution to this performance/cost balance issue.

LaCaille delves into the performance advantages that ElastiCache can provide its users, pairing the service with relational databases. According to LaCaille, microseconds are the new milliseconds; enterprises need to ensure microsecond engine responses from their in-memory data stores to keep up with customer’s demands for speed. LaCaille highlights relational database integration with ElastiCache, emphasizing a strategy called “lazy loading.” “Lazy loading” refers to the process of asking ElastiCache for a query result first, then if it doesn’t provide the result, asking the relational database for the query result; once the result is acquired, it is stored back into ElastiCache for a better chance of reading from cache the next time that query occurs, delivering increased performance.

Regarding cost efficiency, Hancz provides several examples of database types coupled with ElastiCache to greatly reduce cost. Highlighted in the webinar is ensuring proper size for resources, avoiding overpay for unused attributes; utilizing the proper resources and configurations for the job at hand, enabling optimal workload execution; and not using more resources than what is required. ElastiCache addresses these pillars through a cost-effective strategy, implementing key policies:

  • Only pay a per instance fee.
  • No separate IO charge.
  • Cost effective for many RDBMS implementations.
  • Cheaper than scaling the RDS.

Hancz details examples of ElastiCache savings opportunities, such as Oracle, SQL, and Aurora.

AutoScaling provides a plethora of savings opportunities within ElastiCache, according to Hancz. Users can automatically size ElastiCache to suit their workload needs without wasting any resources—users do not pay for unused capacity. Further, users can either implement predefined rules or Amazon CloudWatch metrics for horizontal scaling both in and out, or schedule-based scaling. Data tiering also supplies users with savings opportunities, as its instances have memory and CPU but are attached to an SSD drive for increased memory capacity by a factor of 4-5x, without increasing cost; this will best benefit users who have both hot data and warm data and for workloads that access up to 20% of their dataset regularly.

Amazon ElastiCache aims to supercharge performance without massive increases to cost, suiting any enterprise needing to reduce expenditures in regard to their data.

You can view an archived version of this webinar here.