| 1 |
Kashani wrote: |
| 2 |
> http://dev.mysql.com/doc/mysql/en/MySQL_Cluster_DB_Definition.html under |
| 3 |
> DataMemory |
| 4 |
|
| 5 |
I maybe wrong, but I didn't get that from reading that page. |
| 6 |
|
| 7 |
The way I understand it, it allocates that amount of memory to hold |
| 8 |
table records that are used in transactions and certain queries. Almost |
| 9 |
like a cache, except the cluster uses it for it's "partition" of the table. |
| 10 |
|
| 11 |
Says just two things about how big it needs to be: |
| 12 |
|
| 13 |
[1] Thus it is necessary also to allocate memory to be able to handle |
| 14 |
the largest transactions which are performed in the cluster. |
| 15 |
|
| 16 |
[2] The default DataMemory size is 80MB. The minimum size is 1MB. There |
| 17 |
is no maximum size, but in reality the maximum size has to be adapted so |
| 18 |
that the process doesn't start swapping when using the maximum size of |
| 19 |
the memory. |
| 20 |
|
| 21 |
Further reading produced this: |
| 22 |
|
| 23 |
http://mysql.mirror.nedlinux.nl/tech-resources/articles/storage-engine/part_2.html |
| 24 |
|
| 25 |
While nodes keep their data entirely in memory for maximum performance, |
| 26 |
the durability requirement of ACID transactions is maintained by the |
| 27 |
cluster making sure that at least two nodes in each group have the a |
| 28 |
copy of the data. On-disk "redo" logs are maintained for recovery |
| 29 |
purposes., and each node requires sufficient disk space to store its |
| 30 |
entire fragment for cold starts. |
| 31 |
|
| 32 |
[This tells me that each node doesn't need to store the entire database, |
| 33 |
just the portion of the database that's responsible for. The more nodes |
| 34 |
you have, the less memory each node requires.] |
| 35 |
|
| 36 |
Because MySQL Cluster is main-memory based it is extremely fast. MySQL |
| 37 |
Cluster has good scalability by adding more memory and/or more nodes to |
| 38 |
a configuration. |
Archives