The things I like more as an "external" person given that I handle the OSS part and I'm not directly involved in the internal products development, if not as a technical advisor, are:

1. Transparent clustering. You talk to that thing like if it was a single instance and it scales to the cluster. Failover and so forth happen automatically. It'a a lot higher level than Redis Cluster (which also they support as a protocol), there are good and bad things in the different designs, but I like how well it works.

2. CRDTs store for multi master, also across far geographical area.

3. Redis on flash, using different persistent memory types. This feature originated from an early experiment I did years ago, called "diskstore", then Redis Labs worked at it many years and reached a quite cool thing.

4. All the above can be installed in your servers if you want very high memory setups that is impossible to pay for on the cloud.

5. The various modules like RedisGraph, RedisSearch and so forth are only available in the Redis Labs cloud (or you can install manually on your server). Some of those modules like RedisGraph is receiving years of development.

6. I know that the support team and everybody at Redis Labs really knows Redis. Which was one of the main reasons eventually I joined the company. So I think they are able to handle core-level issues for every user having troubles. A lot of Redis operational improvements came from Redis Labs core/support team after investigating issues with customers.

AFAIK those are the stuff I like more from the external but honestly they do a number of things I don't know very well for certain big customers.

I use the Redis service. It's way less expensive. around a 90% less expensive for me. It has all of the high end features. Fault tolerance, etc.

AWS does not compete on price or performance and the Redis servers are hosted directly in whatever cloud facility you'd like.

On final note. Since setting it up, I've never had to look at it again.

Do you use Redis on flash by chance? I'm guessing it could help reduce reliance on more expensive memory-heavy servers (and their higher cost), but I don't hear often about real experiences with the enterprise version of Redis.

I use the RedisLabs SaaS. No idea how it works and I don't want to know.

Regarding the speed difference between Flash and RAM etc, it's a very interesting topic.

Free plan of 30 concurrenct connections and 30Mb of memory is pretty cool and plenty for small projects!

so now we are down-voting people for asking useful questions?

Amazon actually doesn't run the same Redis as the rest of us, I wouldn't be surprised if they re-wrote a lot of parts of it. They probably re-wrote the storage engine ect ...

> wouldn't be surprised

> probably

Hmm, anything of substance, or is this just pure speculation?

The fact that they've done this for multiple other products built on FOSS projects is a pretty good indicator it can happen.

Aurora is a good example that's explicitly marketed that it's a different core storage engine, but they definitely make many modifications to the projects they're building on top of, but what kinds of changes really varies.

They're certainly capable of it but I believe they would talk about and differentiate it more explicitly if they had any Aurora-like special sauce.