amazon_aw_logo_2010.jpg
While it sounds a bit like a freakish superhero, Amazon's AWS ElastiCache could bring big benefits to those running data across its systems.

The Power of Cache

Any PC user has long been aware of the benefits of cache, a little memory in the hard drive or on the processor has boosted computer performance and made things run a lot faster as data is stored ahead of time, ready for rapid delivery.

Now Amazon is moving this action to the cloud with ElastiCache, which has launched into public beta in the US East (Virginia) region, with wider roll-outs expected soon. The service sits between user web requests and your databases, speeding up the response to user requests. It allows users to run a cloud-based in-memory cache that can be scaled and monitored just like other AWS services.

The result is faster performance from your web apps as you access data from memory rather than disk. The Memcached-protocol-compatible service (Memcached provides a global hash table of serialized objects that can be spread across multiple machines for quick access) simplifies the management of the cache, boosting intensive tasks like major compute services including media, gaming, massive-user social services and so on, as a Scandinavian Troy McClure explains:

Demand for On Demand?

The amount of RAM devoted to the cache can be expanded or shrunk by adding or deleting nodes on the cluster. Pricing depends on the size of the Cache Nodes and type of use:

Standard Cache Node Type (On-Demand)

  • Small -$0.095
  • Large -$0.38
  • Extra Large - $0.76

High-Memory Cache Node Type (On-Demand)

  • Extra Large - $0.56
  • Double Extra Large - $1.12
  • Quadruple Extra Large - $2.24

High-CPU Cache Node Type (On-Demand)

  • Extra Large - $0.76

While it sounds like a big win for AWS users, Amazon will need a relationship with content management vendors who provide services for third parties to allow them access, along with Memcached support to get the most from it. Metrics are displayed on Amazon CloudWatch and there are several safeguards built in to protect against database overload or cache failure.