BizTalk Adapter Wizard problems….

December 23, 2004 / Comments Off on BizTalk Adapter Wizard problems….

I Have some problems with the BizTalk Adapter Wizard.


Let me explain…..I try to make a receive adapter and have found some problems with the Adapter Wizard. It has to do with the batches. On some point there is a the following lines


While (NrOfMessages < SPXAdapterProperties.BatchSize And Not Me._manageEndpoints.TerminateCalled)
  
< some code here>
End While


Ok let me explain :

SPXAdapterProperties.BatchSize is 20
NrOfMessages = 0 ( is only increased if we really receive a message but in my case we received nothing)


Then if I stop the receive location I would expect something happening to “Me._manageEndpoints.TerminateCalled“ like it was set to True but this is not the case.


I Start my receive location, but I Don’t receive any messages. I Stop the Receive location and the Adapter is stuck into this loop.


The only way of getting out of this loop is to stop BTS. But it will take ages then. Is there anybody out there that has similar problems with the Adapter Wizard


 


 


 

Problems with the BizTalk SQL Adapter……

I have some experience withe the Biztalk SQL Adapter and I am not too happy about my findings. Here are some of my findings:



  • The SQL Adapter is SLOW cause the Transaction level is Serializable

  • You CANNOT modify the SQL Schema’s

  • Schema Generation is cumbersome

Transaction level Serializable
This is a slow transaction level in SQL server. Causing a lot of unneccesary locks in the database. With one message everything works very nice. But when you submit about a 1000 messages at once each message causing some queries and some inserts you will see plenty of deadlocks and things will start to go wrong.


Schema’s
Well let’s say we have the SQL Receive function pickup some data from a SQL database… And let’s suppose we want to correlate on let’s say an order number. Ok we start the schema generation wizard…. ( don’t forget to modify the XML clause of the sproc ) and we end up with a nice schema. Let’s test the schema once ( don’t forget to change the for XML clause in the sproc again). After a while everything should be running smoothly ( although rather slow).
Then we decide we need to promote a field, or add a promoted property….. Everything falls apart. By adding the properties we also change generated SQL Adapter Schema. If the Schema is changed the SQL adapter won’t recognize the schema anymore…….So if you have to use the SQL adapter and correllate on that message as well you always have to use a map in the pipeline to map from the SQL Adapter format to a Internal format where you can define your promoted properties and correlation fields.


So if you have to do a lot of SQL stuff in your orchestrations write your own adapter.
¬


¬

One advantage of MS over Open Source

A lot of people will probably not agree with me but I think the quality of MS products is pretty good. If you want to know how serious Microsoft is about quality have a read of this article. It will give you some insight in the testings of Whidbey and als makes clear what testing on an industrial scale is.

Whidbey is just an example. I guess they have these test scenarios for every product. Can you Imagine how many testing is ging on at MS ?.

My guess is that Microsoft will move on and will continue to be a leader in the IT industry for decades to come. I don’t think the open source community can make such an investment in new products. That’s why Microsoft will continue to be a leader and the ‘Open Source field’ will be a follower at best…. If they can keep up with Microsoft’s current pace……

 

  • Recent Posts
  • Recent Comments
  • Archives
  • Categories
  • Meta