summaryrefslogtreecommitdiff
path: root/issues/systems/mariadb/ProbeSetData.gmi
diff options
context:
space:
mode:
Diffstat (limited to 'issues/systems/mariadb/ProbeSetData.gmi')
-rw-r--r--issues/systems/mariadb/ProbeSetData.gmi2
1 files changed, 2 insertions, 0 deletions
diff --git a/issues/systems/mariadb/ProbeSetData.gmi b/issues/systems/mariadb/ProbeSetData.gmi
index 9fceea9..f3ac37c 100644
--- a/issues/systems/mariadb/ProbeSetData.gmi
+++ b/issues/systems/mariadb/ProbeSetData.gmi
@@ -369,9 +369,11 @@ Now the table is 58Gb without indices. Convert to innodb and add indices
```
CREATE INDEX id_index ON ProbeSetData(Id);
+CREATE INDEX strainid_index ON ProbeSetData(StrainId);
ALTER TABLE ProbeSetData ENGINE = InnoDB;
```
+The conversion to InnoDB, annoyingly, leads to an exceedingly large file. I think we'll keep the myisam version for ProbeSetData. It does not do text comparison, so we should see no slow downs mixing with InnoDB. From this exercise I know we can reduce the size a little. Also ProbeSet can probably be made smaller by combining id+strainid where id is dataset id and perhaps even reducing the value size by compressing it into a smaller float.
## Notes