0

I am currently studying on MariaDB for HA and data sharding, but there are a few points that confuse me, please helps me out with it.

  1. Suppose I wanna implement data sharding to my Galera Clustering, I think both MaxScale and Spider can do this. So, what exactly is the difference between MaxScale and Spider? If the functions they provide overlap, can I choose only one of them to implement it? Or can I combine these two solutions to make the different improvement?

  2. As I mentioned above, I wanna to implement the data sharding to improve the writing speed of the system. But the main goal(or advantage) of Galera Clustering is to support multi-master replication so that it can guarantee the high availability, isn't it? So when I shard the data into cluster, the data inconsistency must be happened right? Is there always a trade-off to determine whether efficiency or availability? And what're the alternatives on in this situation?

Thank you guys very much!

ycs18153
  • 13
  • 3
  • Galera isn't a sharding solution, only a HA one. Don't try to do both at once. Get one concept down first at a time. Don't prematurely optimize. Define your writing speed problem ahead of jumping to a solution. "wanna" just sounds childish. Welcome to SO. [dba stackexchange](https://dba.stackexchange.com/) however is the site for database questions. – danblack Jul 12 '21 at 10:45
  • Do you really need sharding? That is, will you have thousands of _writes per second_ or _billions_ or rows of data? – Rick James Jul 13 '21 at 18:21
  • To shard, you would need multiple Galera _clusters_, not just multiple _nodes_. – Rick James Jul 13 '21 at 18:22

0 Answers0