5 ms·

Can someone speak more to this? I've also seen it stated that it's possible to use redis for durable data storage, is that really the case? What do you lose out

Can someone speak more to this? I've also seen it stated that it's possible to use redis for durable data storage, is that really the case? What do you lose out on?

tyingq 5y ago
They have two different approaches to persistence, with different tradeoffs: https://redis.io/topics/persistence https://redis.io/topics/persistence Either would have the generic disadvantages of: - Dataset can't exceed available memory - The general disadvantages of NoSQL vs SQL around queries, transactions, etc, if you're expecting relational type functionality
- Raed667 5y ago
  > Dataset can't exceed available memory From my understanding, they do support sharding
  - tyingq 5y ago
    Sure, but still limited to the total amount of memory of all machines. You can't page in/out from storage.
    - node-bayarea 5y ago
      You can store Terabytes of data in Redis. Just google it.
      - tyingq 5y ago
        Yes, in Terabytes of RAM, or some separate commercial offering with flash. Not sure I get your point.
- sudo-it-all 5y ago
  Redis supports clustering and sharding: https://redis.io/topics/cluster-tutorial https://redis.io/topics/cluster-tutorial There are also managed solutions that implement Redis with partial flash storage. Redis Enterprise and AWS ElastiCache both do this.
Eikon 5y ago
From my experience it works pretty well, the issue is that the idea that it's temporary storage is so prevalent that I wouldn't risk it. From what I saw, people think two times before doing `truncate table` but not that much with `FLUSHALL` in redis... Also, I don't really see where the usecase would be as a permanent datastore compared to something like postgres.
remote-dev 5y ago
I have found this to be useful: https://github.com/fpereiro/backendlore https://github.com/fpereiro/backendlore It is more about writing backends generally, but in it he uses Redis as a primary database and walks through durability and scalability.

Can someone speak more to this? I've also seen it stated that it's possible to use redis for durable data storage, is that really the case? What do you lose out on?

tyingq 5y ago

They have two different approaches to persistence, with different tradeoffs: https://redis.io/topics/persistence https://redis.io/topics/persistence Either would have the generic disadvantages of: - Dataset can't exceed available memory - The general disadvantages of NoSQL vs SQL around queries, transactions, etc, if you're expecting relational type functionality

Raed667 5y ago
> Dataset can't exceed available memory From my understanding, they do support sharding
- tyingq 5y ago
  Sure, but still limited to the total amount of memory of all machines. You can't page in/out from storage.
  - node-bayarea 5y ago
    You can store Terabytes of data in Redis. Just google it.
    - tyingq 5y ago
      Yes, in Terabytes of RAM, or some separate commercial offering with flash. Not sure I get your point.
sudo-it-all 5y ago
Redis supports clustering and sharding: https://redis.io/topics/cluster-tutorial https://redis.io/topics/cluster-tutorial There are also managed solutions that implement Redis with partial flash storage. Redis Enterprise and AWS ElastiCache both do this.

Eikon 5y ago

From my experience it works pretty well, the issue is that the idea that it's temporary storage is so prevalent that I wouldn't risk it. From what I saw, people think two times before doing `truncate table` but not that much with `FLUSHALL` in redis... Also, I don't really see where the usecase would be as a permanent datastore compared to something like postgres.

remote-dev 5y ago

I have found this to be useful: https://github.com/fpereiro/backendlore https://github.com/fpereiro/backendlore It is more about writing backends generally, but in it he uses Redis as a primary database and walks through durability and scalability.