MultiBot Architecture

[VedSengupta]

Hello All,
I have a query that is bugging me.
I have a use case where the historical data suggests that data load can never exceed 200 line items [ For Example].
I am more or less confident that data load will not increase drastically. 

Should I be implementing MultiBOT architecture?
I mean what is the best practise? Should a developer always implement Multi-BOT architecture?

------------------------------
------------------------------
Best Regards,
Ved Sengupta
RPA Developer
Bangalore | INDIA
------------------------------
------------------------------

[EmersonF]

Hey @Ved Sengupta, see if help you:
Multi bot architecture | Blue Prism University​

------------------------------
Emerson Ferreira
Sr Business Analyst
Avanade Brasil
+55 (081) 98886-9544
If my answer helped you? Mark as useful!
------------------------------

[ewilson]

Hi @Ved Sengupta,

A couple questions.

1. Do you know how long it takes a single digital worker to address 1 line item?
2. Do you know approximately how often new line items are being added?

If the amount of time required to address a single line item is less than the frequency that new items are being added I'd say a single digital worker will be able to cover the data load. If the frequency of adding new items is faster than the amount of time it takes to work a single line item then you'd need to look at adding additional digital workers.

Cheers,
​

------------------------------
Eric Wilson
Director, Partner Integrations for Digital Exchange
Blue Prism
------------------------------

[VedSengupta · Accepted solution]

Hello Eric,

1. Do you know how long it takes a single digital worker to address 1 line-item ->It will be worked in intervals throughout its runtime.[For Example, an item needs to be checked for 2 hours so BOT is configured to check every 5 mins and defer and come back and in this 5 min interval BOT checks other available items].
2. Do you know approximately how often new line items are being added -> There is a fixed set of items that are loaded once a day so everyday load differs but by a margin of 5-10%.

What I want to understand is ideally what should one do when developing a use case as implementing MultiBOT will not impact if running on a single bot as the lock will be acquired successfully each time and no latency added to the process.
So, Is it recommended to implement MultiBot no matter the data load as scaling the solution in production will make it easier as no re-work will be required other than the mapping of correct credentials and some other generic stuff. The bot can simply start running.

------------------------------
------------------------------
Best Regards,
Ved Sengupta
RPA Developer
Bangalore | INDIA
------------------------------
------------------------------

[peterlacken]

Hi Ved,

if you mean use work queues for case data as MultiBOT architecture then I would say always Yes - use a work queue when you have case data to handle regardless if 1 bot is sufficient.  You have audit, retry, multi stage handling with status and analytic tagging in addition to scalable with multi bots.  There is almost no disadvantage 🙂

Peter

------------------------------
PeterLacken
Ren Røros Intelligent Automation
Peter.Lacken@rria.no
http://www.rria.no
------------------------------

[TusharVarshney]

@Ved Sengupta - from a designer perspective, I would recommend using Multi bot architecture in all use cases as this could be very helpful in case of increased volumes at any point of time given not much significant time is required at the the time development.​ There is no such best practice defined.

------------------------------
Tushar Varshney
+91-9880188366
------------------------------