Extremely Slow Performance in Flow Designer When Working With Large Flows

[JoshAtBrotech]

Hello Microsoft & the Power Automate Community,

These past couple months, my knowledge of Power Automate and my and ability to use it has expanded rapidly, allowing for our organization to take advantage of Power Automate in ways we were incapable of imagining at the beginning of 2020. It has truly been a fulfilling challenge to learn and use Power Automate.

Lately, I have been working on Flows that will automate business processes that are large in scope, requiring larger Flows with more Action Cards.

Microsoft advertises that Power Automate can handle up to 500 Action Cards in a single Flow, meaning that the product is capable of handling larger Flows.

However, I have noticed that, as the number of Action Cards increases, the performance of the Flow Designer for that specific Flow decreases.

I have a Flow that I have been working on that has approximately 200 Action Cards - so, not even half of the advertised Action Cards per Flow capacity of the Product.

However, at this point, I am essentially incapable of doing additional work on the Flow.

Clicking + New Step results in a lengthy load time, some times several minutes, before I can select which Action Card I want to add, followed by another lengthy load time before the Action Card becomes editable, sometimes with additional lengthy load times while editing, depending on the Action Card being used (Update a Row in an Excel Table, for example, requires another lengthy load time after selecting the Key Column and Key Value). 

In my internet sleuthing, I can see similar issues reported since 2017.

My question to Microsoft and the Power Automate team is, what are my real solutions here?

In the Power Platform Admin Center, our organization's Environment Capacity page shows %87.21 of Database available, & 98.24% of File available, indicating no issues there.

The issue persists across multiple browsers (Chrome, Firefox, Edge), indicating that the issue is not specific to a browser. 

The issue is not present when editing a smaller flow with fewer Action Cards, indicating that the poor performance is specific to larger Flows with more Action Cards, indicating that the issues is specific to the Product.

I do not believe that splitting the Flow into "two flows" is a real solution, as this requires some sloppy dumpage of variables into one location so they can be extracted & communicated in a "part two" of the Flow.

Are there, perhaps, settings, either specific to Flow or to a browser or to the Microsoft Firewall or Microsoft Security settings that can be adjusted to alleviate these issues?

Any insight that Microsoft or the Power Automate Community can provide is greatly appreciated.

[yashkamdar]

Hello Team,

Can you please let us know the License that you currently have for Power Automate in your organization.

The Flow performance depends a lot on the license that is associated for the flow w.r.t the tenant/organization.

Licensing plays a decisive factor in understanding the thresholds, performance limits etc.. that might come into picture.

So let me know your Flow license and then I can check if with regards to your license including a large no. of Action cards is okay or not. 

If this reply has answered your question or solved your issue, please mark this question as answered. Answered questions helps users in the future who may have the same issue or question quickly find a resolution via search. If you liked my response, please consider giving it a thumbs up. THANKS!

[JoshAtBrotech]

Hi there,

I have the Power Automate per user with attended RPA plan and the Power BI Pro licenses attached to my account.

[JoshAtBrotech]

Anyone from Microsoft able to chime in here? Still experiencing these issues today.

[dsmcwhir]

Hi @JoshAtBrotech , my situation is identical to this. I love working with power automate, aside from this, which makes troubleshooting/editing flows near impossible to get through without eating up a huge chunk of time in my day.

Did you ever find any real solution this? 

[TMerkel]

Same here. PA is extremely slow when editing bigger flows. I am considering a switch to Okta workflows for some projects. Their flow designer is very responsive.

[Anonymous]

I'm experiencing a similar issue. Running about 8 parallel processes, a few list rows from various records in series. Nothing too complicated. I would say less than 100 action cards all up.  I have experienced times simply adding a note or two to a dataverse update row action card has caused it to lag out for a couple minutes and then throw all of my text in after it all catches up.

I wonder if breaking down the flow into several smaller flows would work here. Ideally MS actually fix the underlying performance issue with the flow designer, but in the meantime I could imagine this working. 

[Imade]

Hello Everyone,

Has anyone found a solution? I'm having the same problem creating complex workflows.

Thank you in advance

Imade

[pholoabach]

Still having this issue, crazy it's 3 years later & still a problem. 

Mine are all dataverse based, but if I go beyond 2-3 apply to each in designer, it just lags like crazy. If I click a field, it takes a solid minute for the cursor to populate or the dynamic content to pop up. 

I thought for sure with the preview features / new PA update that it would improve. But it hasn't. 

[MVMar]

I'm having the same issues. 
I have a rather complex flow with approx. 200-250 cards and it's almost impossible to edit. 

For every hour I spend on modifying the flow, my actual work time is no more than 5-6 minutes. The rest is waiting time. Wait for the text/action to change, wait to save, wait to add. Awful.
Saving alone takes about 1.5 minutes. 

I have an 8 core CPU running @4.4GHz, and it's getting absolutely hammered by Edge. Spikes of well above 50% usage on all cores.

There must be something that MS can do to optimize the performance.

Otherwise it's a waste of time. 
It's nice to have everything automated, but when it takes me 2 months to create a flow, of which 1.5 months was spent waiting for the thing to load, I'm beginning to doubt the usefulness.

[pholoabach]

Hello All, Not sure if you've tried this method to clear the cache (I typically just use the browser clear). I tried this today and went in to edit my lagging flows, and they are far more responsive than they were. It's still not 100% snappy to how quickly I can move in PowerAutomate, but it's a GREAT start. Let me know if you guys find any value! 

[Imade]

@pholoabach ,

Thanks for the tip, but it's not coming from the cache. I've tried on several browsers and even changed pc always very slow. You lose a lot of time just to change 2/3 settings...

[tcurrent]

Performance has only gotten worse with the implementation of the new flow designer.

[sami_mrs]

@JoshAtBrotech, you are absolutely correct. It consumes a huge amount of time for each step editing. When I checking in performance, it shows that every step is checking flow error and warning. Also, there is no temporary save option. If anything goes wrong, it deletes everything without any backup. It is very inconvenient to edit flow when it has many actions. 

[maxpower45255]

There is a new version of the designer that makes the popup not display automatically. You have to click the 'electric shock' icon before it pops up. This feature has improved the designer performance drastically. I have just spent the last 2hrs literally waiting 20-30secs for the popup to appear each time I switch to a field. I restarted Edge and cleared the cache, but to no avail. I was ready to put a ticket in before deciding to try the new designer. 

Although I'm still not too fond of this new designer, I do like the performance improvements with this new feature. I didn't see any prior performance issues with the old power automate designer until recently. It could be that more user adoption is putting more load on the platform. Or, maybe MSFT is pulling the Apple trick to get people off of the iPhone 4? lol. Nah, I don't believe that.  

[pholoabach]

Funny you say that! my new designer is terrible, it locks up and causes more issues than the classic version. I'll be heartbroken when they take away the ability to switch between 😞 

[maxpower45255]

oh no! I haven't used it until today. So I can't say. I hope they improve the UI though. It's so hard to read with all the rounded border lines. I think the main problem is it's stretching too tall. Before, the actions were shorter in height and more compact. The new layout makes them narrower and taller. I don't know, maybe I'll get over it after using it for a few months.     

[D_Clarke]

The Performance is bad when opening the dynamic content on the new version and it sometimes makes things disappear.

[D_Clarke]

I do tend to create a solution with the main flow calling child flows to do things that keeps the flows a bit smaller.

It does mean you have to send parameters to the child flow in place of variables. 

To save the effort of defining all the variables again in a child flow I use a text input to do a lot of work, i.e. I have a compose action before I call the Child flow and that contains a string that resembles a dummy json file but with the quotes replaced by a **.  

e.g

**Var1**:**Value1**,

**Var2**:**Value2**,

**VarX**:**ValueX**

(NB You have to replace any quotes in the values and surround any * from the values with something like a dash to prevent it failing)

This then is read by the child flow as an text input and the first step is to use Parse Json (Say called I)with the content as:

Concat('{',Replace(triggerBody()['text'],'**','"'),'}')

and the schema:

{

    "type": "object",

    "properties": {

        "Var1": {

            "type": "string"

        },

        "Var2": {

            "type": "string"

        },

        "VarX": {

            "type": "string"

        }

    }

}

 I then call the values body('I')?('Var1') where needed and populate files or other variables as needed in the child flow.  Any changes to the variable values input can be passed back to the main flow using the opposite process.  Using a compose to gather all the changed ones into a ** delimited string and pass that into the output step. The Main flow can then use a parse json to read that and apply the values to update variables or better still if the design is right use be used as the input for sequential child flows that will run depending on what happens in the  previous child flows.

[VictorIvanidze]

Hi @D_Clarke,

how your "solution" can speed up the flow designer?