1.0

Table Of Contents
appropriate locks on the affected rows, and afterward an explicit update statement can be execute to update the
rows.
Transaction Isolation Levels
This release of SQLFire supports the READ_COMMITTED, READ_UNCOMMITTED, and
REPEATABLE_READ transaction isolation levels. Any transaction that uses the READ_UNCOMMITTED
level is implicitly upgraded to be READ_COMMITTED. When nothing is specied, then JDBC operations use
TRANSACTION_NONE isolation which denotes the absence of a transaction. However, SQLFire still offers
certain data consistency and atomicity guarantees as described in Understanding the Data Consistency Model
on page 135.
Using Distributed Transactions in Your Applications on page 137 provides an overview of the distributed
transaction implementation and semantics.
Trigger Limitations
This release of SQLFire does not support statement triggers.
UNION, INTERSECT, and EXCEPT Operators
SQLFire does not support any query that has either nested set operators or a set operator with either a join,
function expression, SQL procedure, view, or sub-query. There is no explicit support provided for ORDER BY,
GROUP BY, or complex lters in the WHERE clause in either child of a query that uses a set operator. Also,
transactions and high availability features are not supported for queries that use a set operator.
In this context, a set operator includes any of these operators: UNION DISTINCT, UNION, UNION ALL,
INTERSECT DISTINCT, INTERSECT, INTERSECT ALL, EXCEPT DISTINCT, EXCEPT, or EXCEPT ALL.
VIEW Limitations
SQLFire does not support views that involve grouping, aggregate, distinct, or join operations on a partitioned
table.
Query Capabilities and Limitations
SQLFire queries have a unique set of capabilities and limitations that are inherent to the distributed database
design.
A general prerequisite for multi-table queries is that all of the partitioned tables involved in the query must reside
in the same server groups. The partitioned tables' server groups must be a subset of the server groups for any
replicated tables in the query. Essentially, this means that SQLFire peers that host partitioned tables data stores
should also host replicated tables. Conguring server groups in this manner enables SQLFire to service equijoins
that involve partitioned and replicated tables locally on the data stores.
Certain types of equijoin queries and inner queries are not supported depending on the table types involved
(replicated or partitioned). Updates are not supported on the partitioning or primary key columns.
The following table summarizes SQLFire support for different types of queries.
NotesSupport levelQuery type
SQLFire does not support subqueries having
more than one level of nesting.
LimitedNested subqueries
Basic queries that involve only a single table
(either partitioned or replicated) work without
Supported with some exceptionsSingle-table queries
any limitations. This includes queries with
655
vFabric SQLFire Limitations