The use of SocketOpPriority session parameter is based on a variety of ways of doing network I/O (input/output):

As well as various ways to perform an I/O operation:

Dedicated threads allow lower response time while using a pool provides higher throughput. So the optimal combination of "how to do I/O" and "what performs I/O (current thread/thread pool/dedicated thread)" choices depends on the application.

FIX Antenna allows to pick such a combination using non-blocking I/O on a per session basis and SocketOpPriority is the corresponding session parameter.

Valid values of SocketOpPriority parameter:



Use thread pool for performing I/O for send and receive operation. The operations are distributed between the worker threads by a dispatcher thread which polls sockets and notifies worker threads when some socket is ready. In this mode FIX Antenna batches incoming/outgoing messages, i.e. several FIX messages can be processed in one send() system call. This is the default value.

Business scenario

It's recommended to use EVEN when there are many concurrent sessions ("many" means much more than the count of cores used), i.e. application demands a high throughput rather than a low response time. Allocating dedicated per session threads is not an option in such cases, so using a pool makes sense.



Tries to send the message from the current thread, if that fails (would block) delegates sending to a dedicated per session thread. Uses a combination of busy polling and readiness notifications. Intended for applications that need low response time (e.g. to send an order as soon as possible).


Use a dedicated per session thread to handle receiving. It uses I/O with busy polling and switches to readiness notifications if no data has been received during a certain configurable time interval. Intended for applications that need low response time (e.g. to receive and react as soon as possible).


Use dedicated per session threads to send and receive messages;

Business scenario

Aggressive strategies of I/O are recommended to use when minimum latency is needed. But it can reduce throughput. 


Try sending from the current thread, if this would block, does the same thing as "EVEN", i.e. queues the message and performs the operation in a worker thread when the corresponding socket becomes ready. The dispatcher thread handles readiness notifications and notifies workers whenever some socket has data to send/receive.

Business scenario

It's recommended to use DIRECT_SEND when there are many concurrent sessions ("many" means much more than the count of cores used), but on the other hand, an application needs a low(er) response time rather than a high throughput.