28.15 Batched Invocations

Oneway and datagram invocations are normally sent as a separate message, that is, the Ice run time sends the oneway or datagram invocation to the server immediately, as soon as the client makes the call. If a client sends a number of oneway or datagram invocations in succession, the client-side run time traps into the OS kernel for each message, which is expensive. In addition, each message is sent with its own message header (see Chapter 34), that is, for n messages, the bandwidth for n message headers is consumed. In situations where a client sends a number of oneway or datagram invocations, the additional overhead can be considerable.

To avoid the overhead of sending many small messages, you can send oneway and datagram invocations in a batch: instead of being sent as a separate message, a batch invocation is placed into a client-side buffer by the Ice run time. Successive batch invocations are added to the buffer and accumulated on the client side until they are flushed, either explicitly by the client or automatically by the Ice run time.

28.15.1 Proxy Methods

The relevant APIs are part of the proxy interface:

namespace IceProxy {
    namespace Ice {
        class Object : /* ... */ {
        public:
            Ice::ObjectPrx ice_batchOneway() const;
            Ice::ObjectPrx ice_batchDatagram() const;
            void ice_flushBatchRequests();
            // ...
        };
    }
}

The ice_batchOneway and ice_batchDatagram methods convert a proxy to a batch proxy. Once you obtain a batch proxy, messages sent via that proxy are buffered in the client-side run time instead of being sent immediately. Once the client has invoked one or more operations on batch proxies, it can call ice_flushBatchRequests to explicitly flush the batched invocations. This causes the batched messages to be sent "in bulk", preceded by a single message header. On the server side, batched messages are dispatched by a single thread, in the order in which they were written into the batch. This means that messages from a single batch cannot appear to be reordered in the server. Moreover, either all messages in a batch are delivered or none of them. (This is true even for batched datagrams.)

An asynchronous version of ice_flushBatchRequests is also available; see Section 29.3.7 for more information.

28.15.2 Automatic Flushing

The default behavior of the Ice run time, as governed by the configuration property Ice.BatchAutoFlush (see Appendix C), automatically flushes batched invocations as soon as a batched request causes the accumulated message to exceed the maximum allowable size. When this occurs, the Ice run time immediately flushes the existing batch of requests and begins a new batch with this latest request as its first element.

For batched oneway invocations, the maximum message size is established by the property Ice.MessageSizeMax, which defaults to 1MB. In the case of batched datagram invocations, the maximum message size is the smaller of the system’s maximum size for datagram packets and the value of Ice.MessageSizeMax.

A client that sends batch requests cannot determine the size of the message that the Ice run time is accumulating for it; automatic flushing is enabled by default as a convenience for clients that unknowingly exceed the maximum message size. A client that requires more deterministic behavior should flush batched requests explicitly at regular intervals.

28.15.3 Batched Datagrams

For batched datagram invocations, you need to keep in mind that, if the data for the invocations in a batch substantially exceeds the PDU size of the network, it becomes increasingly likely for an individual UDP packet to get lost due to fragmentation. In turn, loss of even a single packet causes the entire batch to be lost. For this reason, batched datagram invocations are most suitable for simple interfaces with a number of operations that each set an attribute of the target object (or interfaces with similar semantics). (Batched oneway invocations do not suffer from this risk because they are sent over stream-oriented transports, so individual packets cannot be lost.)

28.15.4 Compression

Batched invocations are more efficient if you also enable compression for the transport: many isolated and small messages are unlikely to compress well, whereas batched messages are likely to provide better compression because the compression algorithm has more data to work with.¹

28.15.5 Active Connection Management

As for oneway invocations (see page 842), you should disable server-side active connection management (ACM) when using batched invocations over TCP/IP or SSL. With server-side ACM enabled, it is possible for a server to close the connection at the wrong moment and not process a batch (with no indication being returned to the client that the batch was lost).

Regardless of whether you used batched messages or not, you should enable compression only on lower-speed links. For high-speed LAN connections, the CPU time spent doing the compression and decompression is typically longer than the time it takes to just transmit the uncompressed data.