Replica copy of the partition

Help
1

Hello Team,
I see the following output when I run show partition. Does this mean for each partiton i have a slave copy and when master copy goes the slave will be given a role of master?

Adding to it, I have 2 leaf node with 4 CPU each and based on the documentation i see that partition per core and that means i need to have 8 partitions in total but the following output shows me 16 partitions per node.
memsql> show partitions ;
±--------±-------------±-----±-------±-------+
| Ordinal | Host | Port | Role | Locked |
±--------±-------------±-----±-------±-------+
| 0 | 10.69.50.225 | 3306 | Master | 0 |
| 0 | 10.69.50.226 | 3306 | Slave | 0 |
| 1 | 10.69.50.226 | 3306 | Slave | 0 |
| 1 | 10.69.50.225 | 3306 | Master | 0 |
| 2 | 10.69.50.225 | 3306 | Master | 0 |
| 2 | 10.69.50.226 | 3306 | Slave | 0 |
| 3 | 10.69.50.226 | 3306 | Slave | 0 |
| 3 | 10.69.50.225 | 3306 | Master | 0 |
| 4 | 10.69.50.225 | 3306 | Master | 0 |
| 4 | 10.69.50.226 | 3306 | Slave | 0 |
| 5 | 10.69.50.226 | 3306 | Slave | 0 |
| 5 | 10.69.50.225 | 3306 | Master | 0 |
| 6 | 10.69.50.225 | 3306 | Master | 0 |
| 6 | 10.69.50.226 | 3306 | Slave | 0 |
| 7 | 10.69.50.226 | 3306 | Slave | 0 |
| 7 | 10.69.50.225 | 3306 | Master | 0 |
| 8 | 10.69.50.225 | 3306 | Slave | 0 |
| 8 | 10.69.50.226 | 3306 | Master | 0 |
| 9 | 10.69.50.226 | 3306 | Master | 0 |
| 9 | 10.69.50.225 | 3306 | Slave | 0 |
| 10 | 10.69.50.225 | 3306 | Slave | 0 |
| 10 | 10.69.50.226 | 3306 | Master | 0 |
| 11 | 10.69.50.226 | 3306 | Master | 0 |
| 11 | 10.69.50.225 | 3306 | Slave | 0 |
| 12 | 10.69.50.225 | 3306 | Slave | 0 |
| 12 | 10.69.50.226 | 3306 | Master | 0 |
| 13 | 10.69.50.226 | 3306 | Master | 0 |
| 13 | 10.69.50.225 | 3306 | Slave | 0 |
| 14 | 10.69.50.225 | 3306 | Slave | 0 |
| 14 | 10.69.50.226 | 3306 | Master | 0 |
| 15 | 10.69.50.226 | 3306 | Master | 0 |
| 15 | 10.69.50.225 | 3306 | Slave | 0 |
±--------±-------------±-----±-------±-------+

2

The answer to your first question is yes.

The default number of partitions should be fine for you unless you have a very demanding workload.

Our query processor, under default settings, uses one core per partition. If there are more cores than partitions, the cores will process one partition after the other until they are done. If you want the fastest possible performance for running one full-scan-type query, then make sure you have at least one partition per core.

CREATE DATABASE lets you specify exactly how many partitions you want if you want to control it.

If you have super-high concurrency and you want to avoid one query taking up all the resources, you can use one partition for every two or four cores, or use resource governor.