My orchestration subscribe message from BTS MsgDB, do few modification to xml and use request/response send port to send the message to external webserive via Wcf-basichttp. It update a table based on status in response. I notice that Orchestraton goes dehydrate for a sec or less before it complete, no failure or warning message during this dehydrate. I have delivery notificate = transmited in Req/Res port. I guess orch goes to dehydrate since it need to wait for response even though it gets the response very fast. Is there way to avoid orch going to dehydrate or let be active for 2min (sendport timeout is 2min) before it goes to dehydrate?

Can we do this in Hostinstance setting or some config?

Thank you for your help in advance,

Siv

Hi,

Rehyderation and  dehyderation of Orchestration instance is much more necessary  in pub sub architecture on which BizTalk is build.

I would not suggest to change anything on this part. You are try changing  The MaxReceiveInterval from adm_service(BizTalk Management DB) but it may cause you race condition for messages published to the mesagebox .

You can also try creating multiple host instances for processing ,receiving and sending messages in/out of BizTalk .If your environment is taking much more time then try running messagebox viewer to see any inconsistency in biztalk databases

Thanks

Abhishek

Hi

There are ways to modify the dehydration behavior, but it's not worth it for the trouble it can cause if not done judiciously. Why are you looking to prevent the dehydration? Are there strict latency requirements for the scenario?

Thanks
Arindam

Thank you for your reply Abhishek and Arindam,

we want our orchestration send 8 web request at a time. So we used maximum worker throat to 8 in Host Instance .NET CLR. What we observed is, 8 orchestrations are active at a time. However, these orch dehydrate for few sec and more orchestration become active to keep the 8 throats. As a result more requests are hitting the web service at a time. Is it possible to keep the orch active until it completes or for few sec?  This orche runs on its on host

You can increase the Minimum Threshold for Orchestration Throttling for your Host from the default of 1sec. But beware, this is potentially dangerous because this affects all orchestrations in that host and may lead to memory issues.

Maybe you can try and control the number of active Send Port instances using the Dispenser pattern - that's a much cleaner approach-

http://social.technet.microsoft.com/wiki/contents/articles/23924.biztalk-resource-dispenser-send-port-edition.aspx