Do we have any articles/document on Capacity planning for Boomi Interfaces for Local Atoms? For example say for 100 Interfaces developed in Boomi, how many Local Atoms should we go for based on transaction size/volume?
I believe it is independent on number of Interfaces developed in your account .But yes,it depends on Volume of Integrations.
Thanks for the link Sai!
Here is the scenario:
1. I don't want to use Boomi Cloud but deploy all my interfaces to Local Atoms.
2. There will be some interfaces (Around 5 Interfaces) which will process huge amount of transactions. - 10000 transactions in 10 minute window. Remaining 105 Interfaces will handle average amount of transactions.
Now, what will be best strategy in planning infrastructure for local Atoms. If we choose to have say 5 Atoms in PROD, we will need to create 5 Environments right ? As if i tie 5 Atoms to PROD, whatever interfaces i deploy (110) will go to all 5 Atoms, but my objective is to divide 110 Interfaces to 5 different Atoms?
Any thoughts on this?
Hi Rajinder,You need not create multiple environments for that.You can attach multiple atoms to a single environment.Note:If your process has two connectors which is deployed to an environment having 5 atoms attached ,It will count 5*2 =10 Licenses.And later your can schedule your processes to run on different atoms( Divide load accordingly)
My point is , I don’t want to deploy processes to all the Atoms in an environment, as this will consume lots of licenses In my scenario. Any workaround to this?
As you said, I could deploy processes to all atoms but not schedule them to run on all. But this doesn’t help me in the licensing problem.
Do you think , if I create environments like Prod-FTP, Prod-HTTP, Prod-SAP and associate different Atoms with them, I could work around licensing to some extent by not deploying all connectors to all Atoms.
Nope.That is again similar to attaching multiple atoms to single environment ,even though you have individual environment for each atom that would count same licensing.
You can use Molecule for such scenario's . License count will only be one and you can process high volume of data.
I think a lot depends on the complexity of the interfaces. I have that volume (interfaces running every min) running on a single moderate (8GB ram) sized atom. Our plan is to evaluate more ram/cpu, and/or move to a molecule when it gets stressed out.
Retrieving data ...