Amazon MemoryDB for Redis – A Redis-Compatible, Durable Database(aws.amazon.com)
aws.amazon.com
Amazon MemoryDB for Redis – A Redis-Compatible, Durable Database
https://aws.amazon.com/blogs/aws/introducing-amazon-memorydb-for-redis-a-redis-compatible-durable-in-memory-database-service/
13 comments
In the launch podcast, the question was asked about specific features. They didn't mention the ones you listed, but the response was that it was 100% compatible with v6, so I suspect so.
https://aws.amazon.com/podcasts/467-introducing-amazon-memor...
https://aws.amazon.com/podcasts/467-introducing-amazon-memor...
I'd much rather use OSS like Aerospike or ScyllaDB for this workload and still retain some agency over my deployment.
Getting a system like this right is super hard, and not being OSS, you are at the mercy of the cloud provider to fix issues.
Aerospike recently released their own Redis protocol proxy for Redis [1]. Scylla has their own nascent Redis api driver [2], would love to get folks feedback on it.
[1] https://github.com/aerospike/skyhook
[2] https://siddharthc.medium.com/redis-on-nvme-with-scylladb-5e...
Getting a system like this right is super hard, and not being OSS, you are at the mercy of the cloud provider to fix issues.
Aerospike recently released their own Redis protocol proxy for Redis [1]. Scylla has their own nascent Redis api driver [2], would love to get folks feedback on it.
[1] https://github.com/aerospike/skyhook
[2] https://siddharthc.medium.com/redis-on-nvme-with-scylladb-5e...
I'm really impressed with Seastar [1], the library ScyllaDB wrote to solve for writing non-blocking, asynchronous code in C++. It's also used by Redpanda [2], a Kafka-compatible service in the same vein as MemoryDB/Skyhook.
[1] https://github.com/scylladb/seastar
[2] https://github.com/vectorizedio/redpanda
[1] https://github.com/scylladb/seastar
[2] https://github.com/vectorizedio/redpanda
The arch of ScyllaDB blew my mind gotta say. Great talk [1] by Avi Kivity, co-founder of Scylla that goes over the design and architecture.
Glommio [2] (believe it used to be called Scipio) is a thread-per-core framework for Rust modeled after Seastar.
There isn't much in the hn hive mind about it, this one [3] stands out.
[1] https://www.youtube.com/watch?v=0S6i9BmuF8U
[2] https://github.com/DataDog/glommio
[3] https://news.ycombinator.com/item?id=24976533
Glommio [2] (believe it used to be called Scipio) is a thread-per-core framework for Rust modeled after Seastar.
There isn't much in the hn hive mind about it, this one [3] stands out.
[1] https://www.youtube.com/watch?v=0S6i9BmuF8U
[2] https://github.com/DataDog/glommio
[3] https://news.ycombinator.com/item?id=24976533
Great recommendation, thank you.
This is pretty cool. I'm hoping to see a stress test write up.
Is this the project that triggered Antirez to change the license on Redis?
Disclosure: I work for AWS where I build infrastructure services
The license to Redis never changed. See his blog post [1] around the time when his _employer_ announced some license changes on modular extensions to Redis. See also threads on HN [2].
[1] http://antirez.com/news/120
[2] https://news.ycombinator.com/item?id=17818923
The license to Redis never changed. See his blog post [1] around the time when his _employer_ announced some license changes on modular extensions to Redis. See also threads on HN [2].
[1] http://antirez.com/news/120
[2] https://news.ycombinator.com/item?id=17818923
When will it pass Jepsen?
How is this better than Hadoop?
Redis and workalikes are in-memory key-value datastores, more appropriate for real-time transactional workloads. They serve entirely different purposes than Hadoop, which is a data analytics platform. Hadoop is targeted towards large "slow" data processing jobs, using a function called "map reduce." Redis and its workalikes are targeted to quick atomic CRUD operations (create, reads, updates and deletes).
It's like asking what's the difference between a truck and a Ferrari. They both have engines and steering wheels, right?
But totally different workloads.
It's like asking what's the difference between a truck and a Ferrari. They both have engines and steering wheels, right?
But totally different workloads.
How is Redis different than Hadoop?
Those are the only reason I am using Redis because not other database easily let you subscribe to notification on change to a specific set of keys.