Does Sql batch query execution involve multiple exhange of data between server and client? - sql

Question

Batch query execution from what I have read from multiple sources online have been stating that it enables grouping multiple statements together and executing it at once, thereby eliminating multiple back and forth communication.
Some sources that claim this are:
https://www.tutorialspoint.com/jdbc/jdbc-batch-processing.htm#:~:text=Batch%20Processing%20allows%20you%20to,communication%20overhead%2C%20thereby%20improving%20performance.
http://tutorials.jenkov.com/jdbc/batchupdate.html
https://www.baeldung.com/jdbc-batch-processing
All of these talk about single network trip etc, however going through the source code of H2 or Sqlite, it does look like its executing one by one. Albeit with autocommit disabled.
Eg: Sqlite
final synchronized int[] executeBatch(long stmt, int count, Object[] vals, boolean autoCommit) throws SQLException {
if (count < 1) {
throw new SQLException("count (" + count + ") < 1");
}
final int params = bind_parameter_count(stmt);
int rc;
int[] changes = new int[count];
try {
for (int i = 0; i < count; i++) {
reset(stmt);
for (int j = 0; j < params; j++) {
rc = sqlbind(stmt, j, vals[(i * params) + j]);
if (rc != SQLITE_OK) {
throwex(rc);
}
}
rc = step(stmt);
if (rc != SQLITE_DONE) {
reset(stmt);
if (rc == SQLITE_ROW) {
throw new BatchUpdateException("batch entry " + i + ": query returns results", changes);
}
throwex(rc);
}
changes[i] = changes();
}
}
finally {
ensureAutoCommit(autoCommit);
}
reset(stmt);
return changes;
}
Eg: H2
public int[] executeBatch() throws SQLException {
try {
debugCodeCall("executeBatch");
if (batchParameters == null) {
// Empty batch is allowed, see JDK-4639504 and other issues
batchParameters = Utils.newSmallArrayList();
}
batchIdentities = new MergedResult();
int size = batchParameters.size();
int[] result = new int[size];
SQLException first = null;
SQLException last = null;
checkClosedForWrite();
for (int i = 0; i < size; i++) {
Value[] set = batchParameters.get(i);
ArrayList<? extends ParameterInterface> parameters =
command.getParameters();
for (int j = 0; j < set.length; j++) {
Value value = set[j];
ParameterInterface param = parameters.get(j);
param.setValue(value, false);
}
try {
result[i] = executeUpdateInternal();
// Cannot use own implementation, it returns batch identities
ResultSet rs = super.getGeneratedKeys();
batchIdentities.add(((JdbcResultSet) rs).result);
} catch (Exception re) {
SQLException e = logAndConvert(re);
if (last == null) {
first = last = e;
} else {
last.setNextException(e);
}
result[i] = Statement.EXECUTE_FAILED;
}
}
batchParameters = null;
if (first != null) {
throw new JdbcBatchUpdateException(first, result);
}
return result;
} catch (Exception e) {
throw logAndConvert(e);
}
}
From the above code I see that there are multiple calls to the database, with each having its own result set. How does batch execution actually work?