[GitHub] [incubator-kvrocks] GoodManWEN edited a discussion: Need advice on production deployment

[GitBox <gi...@apache.org>]

GitHub user GoodManWEN edited a discussion: Need advice on production deployment

Hi everyone, I came across this amazing library some time ago and I'm trying to use it in a production environment soon. I've got some questions that I haven't found answered in the documentation, so I'm going to seek help from the community. 

My usage scenarios are:
1. Approximately 500G of data in total (uncompressed)
2. Two existing machines with 64GB RAM and 2TB storage
3. The usage scenarios are almost all random reads and writes, with a read/write volume of approximately 100/1
4. The service have a relatively high number of qps but not to the limit of what a single machine can load. (Peaks can reach 300-500 requests per second)
5. We want to optimise the service with the aim of minimising response latency.
6. We want the data service take some memory for caching to increase performance if possible.
7. We would like to see a more robust solution for securing data, things such as crash safe policy.
8. As we have two machines and one is currently able to handle peak loads, although we want to set up a 1-master-1-slave cluster, we only want one of the two to provide service. The other one we only want to do for backup purposes. For example, we synchronise every 6 hours, which prevents our data from suffering an unacceptable crash in the event of an accident.

Suggestions on any of the above points are very welcome. Because we know that although this service has a similar API to redis, the architecture is on a completely different underlying layer. In many cases, this makes it difficult to apply our experience. Also because the community is not yet rich in information, there are many questions we cannot fully confirm the answers to. 

P.S. Can anyone tell me where I can find documents related to the multi-threaded read and write design of this library. Since the original single-threaded model of Redis is completely changed, I'm afraid we'll have different expectations for dirty reads/writes cases.

GitHub link: https://github.com/apache/incubator-kvrocks/discussions/1276

----
This is an automatically sent email for issues@kvrocks.apache.org.
To unsubscribe, please send an email to: issues-unsubscribe@kvrocks.apache.org