Summary

Here's a review of the most important points of this chapter:

• Whereas synchronous drawbridges are based on component technologies, asynchronous drawbridges are based on message queues.

• The message queues that are used for asynchronous drawbridges are both persistent and transactional.

• The insertion of an infogram is transactionally coordinated with work being done by the donor fortress by DTC, the distributed transaction coordinator .

• The receipt of an infogram is also transactionally coordinated by DTC, this time with work being done by the recipient fortress.

• Sending of the infogram is fairly well coupled with receipt of the infogram.

• Homogeneous asynchronous drawbridges are based on a configuration in which a component veneer is placed over a message queue architecture. They are not recommended.

• Asynchronous drawbridges have four main advantages over their synchronous counterparts: they are nonblocking, they give pseudoreliability , they provide workload averaging, and they provide a poor-man's cluster architecture.

• Asynchronous drawbridges based on message queues can have performance problems resulting from the use of DTC to coordinate transactions. Sometimes these problems can be reduced if the message queue is replaced by a homegrown asynchronous system built on a database.

I hope this chapter has whetted your appetite for investigating opportunities to use asynchronous drawbridges in your enterprise architecture.