-
Notifications
You must be signed in to change notification settings - Fork 472
Open
Labels
C-doc-improvementO-externalOrigin: Issue comes from external users.Origin: Issue comes from external users.P-3Low priority; nice-to-haveLow priority; nice-to-haveT-incorrect-or-unclear-infono-issue-activity
Description
Exalate commented:
On https://www.cockroachlabs.com/docs/dev/column-families.html the example contains a table with one column as the primary key, the id
:
CREATE TABLE test (
id INT PRIMARY KEY,
last_accessed TIMESTAMP,
data BYTES,
FAMILY f1 (id, last_accessed),
FAMILY f2 (data)
);
The example then mixes never updated fields (id
) with the frequently updated one last_accessed
in one column family, putting the never modified data
into f2
.
For the first glance, it would be logical to group static (id
and data
) columns into one family and frequently modified ones to the other:
CREATE TABLE test (
id INT PRIMARY KEY,
last_accessed TIMESTAMP,
data BYTES,
FAMILY f1 (id, data),
FAMILY f2 (last_accessed)
);
So could you please extend the current doc, explain and add more details about:
- why putting
id
andlast_accessed
into one CF is recommended - what if there are more PK columns, should those be in the frequently modified CF in this case as well?
Jira Issue: DOC-462
Metadata
Metadata
Assignees
Labels
C-doc-improvementO-externalOrigin: Issue comes from external users.Origin: Issue comes from external users.P-3Low priority; nice-to-haveLow priority; nice-to-haveT-incorrect-or-unclear-infono-issue-activity