Caching Overview

Caching helps applications perform dramatically faster and cost significantly less at scale

What is Caching?

In computing, a cache is a high-speed data storage layer which stores a subset of data, typically transient in nature, so that future requests for that data are served up faster than is possible by accessing the data’s primary storage location. Caching allows you to efficiently reuse previously retrieved or computed data.

How does Caching work?

The data in a cache is generally stored in fast access hardware such as RAM (Random-access memory) and may also be used in correlation with a software component. A cache's primary purpose is to increase data retrieval performance by reducing the need to access the underlying slower storage layer.

Trading off capacity for speed, a cache typically stores a subset of data transiently, in contrast to databases whose data is usually complete and durable.

Caching Overview

Due to the high request rates or IOPS (Input/Output operations per second) supported by RAM and In-Memory engines, caching results in improved data retrieval performance and reduces cost at scale. To support the same scale with traditional databases and disk-based hardware, additional resources would be required. These additional resources drive up cost and still fail to achieve the low latency performance provided by an In-Memory cache.

Caches can be applied and leveraged throughout various layers of technology including Operating Systems, Networking layers including Content Delivery Networks (CDN) and DNS, web applications, and Databases. You can use caching to significantly reduce latency and improve IOPS for many read-heavy application workloads, such as Q&A portals, gaming, media sharing, and social networking. Cached information can include the results of database queries, computationally intensive calculations, API requests/responses and web artifacts such as HTML, JavaScript, and image files. Compute-intensive workloads that manipulate data sets, such as recommendation engines and high-performance computing simulations also benefit from an In-Memory data layer acting as a cache. In these applications, very large data sets must be accessed in real-time across clusters of machines that can span hundreds of nodes. Due to the speed of the underlying hardware, manipulating this data in a disk-based store is a significant bottleneck for these applications.

In a distributed computing environment, a dedicated caching layer enables systems and applications to run independently from the cache with their own lifecycles without the risk of affecting the cache. The cache serves as a central layer that can be accessed from disparate systems with its own lifecycle and architectural topology. This is especially relevant in a system where application nodes can be dynamically scaled in and out. If the cache is resident on the same node as the application or systems utilizing it, scaling may affect the integrity of the cache. In addition, when local caches are used, they only benefit the local application consuming the data. In a distributed caching environment, the data can span multiple cache servers and be stored in a central location for the benefit of all the consumers of that data.

When implementing a cache layer, it’s important to understand the validity of the data being cached. A successful cache results in a high hit rate which means the data was present when fetched. A cache miss occurs when the data fetched was not present in the cache. Controls such as TTLs (Time to live) can be applied to expire the data accordingly. Another consideration may be whether or not the cache environment needs to be Highly Available, which can be satisfied by In-Memory engines such as Redis. In some cases, an In-Memory layer can be used as a standalone data storage layer in contrast to caching data from a primary location. In this scenario, it’s important to define an appropriate RTO (Recovery Time Objective--the time it takes to recover from an outage) and RPO (Recovery Point Objective--the last point or transaction captured in the recovery) on the data resident in the In-Memory engine to determine whether or not this is suitable. Design strategies and characteristics of different In-Memory engines can be applied to meet most RTO and RPO requirements.

Layer Client-Side DNS Web App Database
Use Case

Accelerate retrieval of web content from websites (browser or device)

Domain to IP Resolution Accelerate retrieval of web content from web/app servers. Manage Web Sessions (server side) Accelerate application performance and data access Reduce latency associated with database query requests
Technologies HTTP Cache Headers, Browsers DNS Servers HTTP Cache Headers, CDNs, Reverse Proxies, Web Accelerators, Key/Value Stores Key/Value data stores, Local caches Database buffers, Key/Value data stores
Solutions Browser Specific Amazon Route 53 Amazon CloudFrontElastiCache for RedisElastiCache for MemcachedPartner Solutions Application Frameworks, ElastiCache for RedisElastiCache for MemcachedPartner Solutions  ElastiCache for RedisElastiCache for Memcached

Caching with Amazon ElastiCache

Amazon ElastiCache is a web service that makes it easy to deploy, operate, and scale an in-memory data store or cache in the cloud. The service improves the performance of web applications by allowing you to retrieve information from fast, managed, in-memory data stores, instead of relying entirely on slower disk-based databases. Learn how you can implement an effective caching strategy with this technical whitepaper on in-memory caching.

diagram

Benefits of Caching

Because memory is orders of magnitude faster than disk (magnetic or SSD), reading data from in-memory cache is extremely fast (sub-millisecond). This significantly faster data access improves the overall performance of the application.

A single cache instance can provide hundreds of thousands of IOPS (Input/output operations per second), potentially replacing a number of database instances, thus driving the total cost down. This is especially significant if the primary database charges per throughput. In those cases the price savings could be dozens of percentage points.

A common challenge in modern applications is dealing with times of spikes in application usage. Examples include social apps during the Super Bowl or election day, eCommerce websites during Black Friday, etc. Increased load on the database results in higher latencies to get data, making the overall application performance unpredictable. By utilizing a high throughput in-memory cache this issue can be mitigated.

In many applications, it is likely that a small subset of data, such as a celebrity profile or popular product, will be accessed more frequently than the rest. This can result in hot spots in your database and may require overprovisioning of database resources based on the throughput requirements for the most frequently used data. Storing common keys in an in-memory cache mitigates the need to overprovision while providing fast and predictable performance for the most commonly accessed data.

In addition to lower latency, in-memory systems also offer much higher request rates (IOPS) relative to a comparable disk-based database. A single instance used as a distributed side-cache can serve hundreds of thousands of requests per second.