Microsecond Latency at Scale: Amazon ElastiCache for Valkey Explained

Key Takeaways

during Prime day 2024 Amazon elasticache handled more than a quadrillion requests in a single day with Peak Performance exceeding 1 trillion requests per minute that's the kind of scale that keeps developers up at night hi everyone I'm Brooke today we're diving into Amazon elasticache which recently announced support for valky if you're building high performance applications or dealing with traffic spikes you'll want to pay attention to this let's break down Amazon elastic hash falky into three key areas first up technical specs elastic has for valky supports valky version 7.2 an open- Source high performance key value data store it's important to note that valky is stewarded by the Linux foundation and supported by over 40 companies including AWS now let's talk about what makes elasticas shalk stand out first it supports both serverless and node-based deployments giving you flexibility in your architecture performance-wise it's impressive we're talking about scaling to 500 million requests per second on a single node-based cluster with microsc read and write latency reliability is also a key Focus elastic falky offers 99.99% availability automatic scaling and high availability with multi-az deployments for those of you concerned about Disaster Recovery it also supports cross region replication ensuring business continuity two performance and cost benefits when it comes to Performance and cost elastic has falki brings some significant improvements to the table on the pricing front elastic has servus feli is priced 33% lower than other supported engines if you're looking at node-based clusters you'll see 20% less cost than other supported engines for those of you just starting out or working on smaller projects here's some great news the minimum data storage for serverless is now just 100 megabytes down from 1 gab with redus OS you can get started for as low as $6 per month with serverless three developer experience now let's talk about what this means for you as a developer the good news is that elastic has falky maintains API compatibility with redus OSS in plain English this means your existing code should work without modifications there are a few other developer friendly features worth mentioning first you can migrate from elastic hash for redus OSS with zero downtime AWS has also introduced valky Glide an open- Source client available for Java Python and node.js with go support in the works this client is designed to improve reliability especially during unplanned outages let's walk through setting up and using an elastic cash for valky instance this process involves four main steps first we'll create a serverless cache this command creates a valky compatible cache in your default VPC using the default Security Group it automatically scales based on demand so you don't need to worry about node sizes or cluster configurations next we'll set up an ec2 instance in the same VPC as our cache for this demo we'll assume that's already done three now we need to install and set up the valky CLI utility in our ec2 instance note that we're building with TLS support which is required for elastic hash serverless for valky four finally let's connect to our cache using the valky CLI now that we're connected let's let's create a hash object representing a car and retrieve specific fields and this shows you vy's compatibility with existing reddis OSS commands your current reddis OSS based applications should work without changes elastic valki combines the familiarity of redus OSS with performance improvements that can make a real difference in production environments whether you're handling massive traffic spikes like Prime day or building real-time analytics systems it's worth considering for your caching layup if you're currently using elasticache for Reedus OSS you can upgrade to elasticache for valky with no downtime plus if you're using reserved nodes you'll retain your existing discounted rates across all node sizes within the same family give it a try and let me know in the comments how it performs for your use cases don't forget to like And subscribe and I'll see you next time