Late last year we introduced the Enterprise Edition of X-Link. The Enterprise Edition provides X-Link platform with many very versatile features that can be used to provide performance enhancements. And we've just taken advantage of these platform features again to provide better TCPip communications. The biggest change which is shipping with 20.05 Enterprise Edition is Sub-Linkages.
So what is a Sub-Linkage?
A sub-linkage is a 2nd linkage definition that handles some of the connections of a full linkage. In this particular case, the sub-linkage contains the TCPip tasks, while the main linkage contains the normal data transfer tasks.
So why is this helpful?
This is very helpful when the linkage is between a TCPip system and HTTP systems, as well as some other slow to respond systems. Most HTTP systems perform posting of data rather quickly, generally within 2 to 3 seconds. However, bill posting can be more complicated, and depending upon the contents of the bills they may take 15 to 30 seconds or even over 1 minute to post. This delay is normal.
Traditionally, it is not uncommon for vendors using TCPip to wait 30 seconds for a reply. Now many vendors are requiring a response within 10 seconds. Generally, if the sender doesn't get a response to the sent message in 10 seconds, the sender resends the message, thus putting a duplicate message in the data as well as using up more of the bandwidth of the connection. X-Link does remove any duplicate messages received.
The rub occurs when X-Link receives a bill from the TCPip sender and then starts to process the bill with the destination system. Even in low volume scenarios it is very possible for the sender to send a 2nd message to X-Link while X-Link is still processing the bill posting of the 1st message in the destination system. In this scenario it can easily take more than 10 seconds for X-Link to get back to the TCPip task to respond. The following time line shows how the response is blocked while the destination system is being updated.
By splitting the linkage up into a primary and a sub linkage, X-Link Enterprise Edition starts a second engine dedicated to the TCPip tasks. This means that any waiting for the destination system to post messages is no longer stopping the TCPip tasks from receiving and responding to messages timely. In fact, the average response time is 2.5 seconds no matter what is posting on the other linkage. The following time line shows how the Enterprise Edition with sub-linkages solves the response issue.
And this works because X-Link runs a separate engine for each linkage (or sub-linkage). This multi-engine technique provides the ability for X-Link to perform many operations simultaneously thus improving response times and resource usage. In this particular case, the first engine services the TCPip system connections at the same time as the second engine is servicing the destination system connection.
In the Application and Server Editions, the X-Link Process Manager only starts one X-Link Engine for all connections. That engine round robins the processing time of X-Link between all of the connections setup to run under X-Link. This round robin can cause message delivery delays when large numbers of connections are running in a single X-Link Engine or the destination system is slow to respond, like with HTTP systems.
This particular feature is only available in the Enterprise Edition, as it is the only edition that starts multiple engines.
Thank you. We hope the information in the entry was useful to you.