When a query returns many rows, it would be inefficient to return them as a single response message. Instead, the driver breaks the results into pages which get returned as they are needed.
The fetch size specifies how many rows will be returned at once by Cassandra (in other words, it's the size of each page).
You can set a default fetch size globally for a Cluster instance:
// At initialization:
Cluster cluster = Cluster.builder()
.addContactPoint("127.0.0.1")
.withQueryOptions(new QueryOptions().setFetchSize(2000))
.build();
// Or at runtime:
cluster.getConfiguration().getQueryOptions().setFetchSize(2000);The fetch size can also be set on a statement:
Statement statement = new SimpleStatement("your query");
statement.setFetchSize(2000);If the fetch size is set on a statement, it will take precedence; otherwise, the cluster-wide value (which defaults to 5000) will be used.
Note that setting a fetch size doesn't mean that Cassandra will always return the exact number of rows, it is possible that it returns slightly more or less results.
The fetch size limits the number of results that are returned in one page; if you iterate past that, the driver will run background queries to fetch subsequent pages. Here's an example with a fetch size of 20:
client Session Cassandra
--+--------------+---------------------------------+-----
|execute(query)| |
|------------->| |
| | query rows 1 to 20 |
| |-------------------------------->|
| | |
| |create |
| |------>ResultSet |
| | |
+-----+--------+-----------------+-+ |
|For i in 1..20| | | |
+--------------+ | | |
| | get next row | | |
| |------------------------->| | |
| | row i | | |
| |<-------------------------| | |
| | | | |
+-----+--------------------------+-+ |
| | |
| | |
| get next row | |
|------------------------->| |
| | query rows 21 to 40 |
| |-------------------->|
| row 21 | |
|<------------------------ | |
By default, the background fetch happens at the last moment, when there are no more "local" rows available. If you need finer control, the ResultSet interface provides the following methods:
getAvailableWithoutFetching()andisFullyFetched()to check the current state;fetchMoreResults()to force a page fetch.
Here's how you could use these methods to pre-fetch the next page in advance, in order to avoid the performance hit at the end of each page:
ResultSet rs = session.execute("your query");
for (Row row : rs) {
if (rs.getAvailableWithoutFetching() == 100 && !rs.isFullyFetched())
rs.fetchMoreResults(); // this is asynchronous
// Process the row ...
System.out.println(row);
}If you use paging with the async API, you'll also want to use those methods to avoid triggering synchronous fetches unintentionally; see async paging.
Sometimes it is convenient to save the paging state in order to restore it later. For example, consider a stateless web service that displays a list of results with a link to the next page. When the user clicks that link, we want to run the exact same query, except that the iteration should start where we stopped on the previous page.
To do so, the driver exposes a PagingState object that represents where we were in the result set when the last page was fetched:
ResultSet resultSet = session.execute("your query");
// iterate the result set...
PagingState pagingState = resultSet.getExecutionInfo().getPagingState();This object can be serialized to a String or a byte array:
String string = pagingState.toString();
byte[] bytes = pagingState.toBytes();This serialized form can be saved in some form of persistent storage to
be reused later. In our web service example, we would probably save the
string version as a query parameter in the URL to the next page
(http://myservice.com/results?page=<...>). When that value is
retrieved later, we can deserialize it and reinject it in a statement:
PagingState pagingState = PagingState.fromString(string);
Statement st = new SimpleStatement("your query");
st.setPagingState(pagingState);
ResultSet rs = session.execute(st);Note that the paging state can only be reused with the exact same statement (same query string, same parameters). Also, it is an opaque value that is only meant to be collected, stored an re-used. If you try to modify its contents or reuse it with a different statement, the driver will raise an error.
Putting it all together, here's a more comprehensive example implementation for our web service:
final int RESULTS_PER_PAGE = 100;
Statement st = new SimpleStatement("your query");
st.setFetchSize(RESULTS_PER_PAGE);
String requestedPage = extractPagingStateStringFromURL();
// This will be absent for the first page
if (requestedPage != null) {
st.setPagingState(
PagingState.fromString(requestedPage));
}
ResultSet rs = session.execute(st);
PagingState nextPage = rs.getExecutionInfo().getPagingState();
// Note that we don't rely on RESULTS_PER_PAGE, since Cassandra might
// have not respected it, or we might be at the end of the result set
int remaining = rs.getAvailableWithoutFetching();
for (Row row : rs) {
renderInResponse(row);
if (--remaining == 0) {
break;
}
}
// This will be null if there are no more pages
if (nextPage != null) {
renderNextPageLink(nextPage.toString());
}Due to internal implementation details, PagingState instances are not
portable across native protocol versions. This
could become a problem in the following scenario:
- you're using the driver 2.0.x and Cassandra 2.0.x, and therefore native protocol v2;
- a user bookmarks a link to your web service that contains a serialized paging state;
- you upgrade your server stack to use the driver 2.1.x and Cassandra 2.1.x, so you're now using protocol v3;
- the user tries to reload their bookmark, but the paging state was serialized with protocol v2, so trying to reuse it will fail.
If this is not acceptable for you, you might want to consider the unsafe API described in the next section.
Note: because of CASSANDRA-10880, paging states are also incompatible between Cassandra 2.2 and 3.0, even if they're both using native protocol v4. This will manifest as the following error:
com.datastax.driver.core.exceptions.ProtocolError: An unexpected protocol error occurred on host xxx.
This is a bug in this library, please report: Invalid value for the paging state
The Cassandra documentation recommends staying on protocol v3 during an upgrade between these two versions:
Clients must use the native protocol version 3 when upgrading from 2.2.X as
the native protocol version 4 is not compatible between 2.2.X and 3.Y. See
https://www.mail-archive.com/[email protected]/msg45381.html for details.
As an alternative to the standard API, there are two methods that
manipulate a raw byte[] instead of a PagingState object:
These low-level methods perform no validation on their arguments; therefore nothing protects you from reusing a paging state that was generated from a different statement, or altered in any way. This could result in sending a corrupt paging state to Cassandra, with unpredictable consequences (ranging from wrong results to a query failure).
There are two situations where you might want to use the unsafe API:
- you never expose the paging state to end users and you are confident that it won't get altered;
- you want portability across protocol versions and/or you prefer implementing your own validation logic (for example, signing the raw state with a private key).
Saving the paging state works well when you only let the user move from one page to the next. But it doesn't allow random jumps (like "go directly to page 10"), because you can't fetch a page unless you have the paging state of the previous one. Such a feature would require offset queries, but they are not natively supported by Cassandra (see CASSANDRA-6511). The rationale is that offset queries are inherently inefficient (the performance will always be linear in the number of rows skipped), so the Cassandra team doesn't want to encourage their use.
If you really want offset queries, you can emulate them client-side. You'll still get linear performance, but maybe that's acceptable for your use case. For example, if each page holds 10 rows and you show at most 20 pages, this means you'll fetch at most 190 extra rows, which doesn't sound like a big deal.
For example, if the page size is 10, the fetch size is 50, and the user asks for page 12 (rows 110 to 119):
- execute the statement a first time (the result set contains rows 0 to 49, but you're not going to use them, only the paging state);
- execute the statement a second time with the paging state from the first query;
- execute the statement a third time with the paging state from the second query. The result set now contains rows 100 to 149;
- skip the first 10 rows of the iterator. Read the next 10 rows and discard the remaining ones.
You'll want to experiment with the fetch size to find the best balance: too small means many background queries; too big means bigger messages and too many unneeded rows returned (we picked 50 above for the sake of example, but it's probably too small -- the default is 5000).
Again, offset queries are inefficient by nature. Emulating them client-side is a compromise when you think you can get away with the performance hit. We recommend that you:
- test your code at scale with the expected query patterns, to make sure that your assumptions are correct;
- set a hard limit on the highest possible page number, to prevent malicious users from triggering queries that would skip a huge amount of rows.