All language subtitles for 006 Performance Analyzer_en

Tutor: Up next, let's talk about the Performance Analyzer

which is another tool that's native to Power BI.

And the Performance Analyzer is a tool that records all

of the behind the scenes actions

that are happening within your report.

When filters are changed,

a visual is updated, cross filtering occurs,

DAX measures are calculated, et cetera.

Basically, anytime something is updated within the report

the performance Analyzer records it

and will then display it for you.

So you can almost think of this like Excel's macro recorder.

The Performance Analyzer tracks the time

in milliseconds for each action

and then groups those actions into distinct buckets,

query load time, the visual load time, other tasks

and then you'll have a specific line item

for direct query visuals

if a direct query connection is present.

So looking at this Performance Analyzer image on screen

we can see here the output of each of these steps.

After the Performance Analyzer is finished recording

all of those recorded steps will show

up with an associated time, they'll be named

and they'll be in this pane here.

And by expanding an individual query

like our card query that we have highlighted,

you'll be able to see all of these different buckets.

So let's talk about each of these categories.

First, we've got the DAX query, and this shows the amount

of time that it takes for the visual to send the query

to the internal engine for the engine to process and run it

and then to return the result.

So basically this is how long it takes the DAX engines

to process your code.

And for the DAX query specifically

this is where some external tools,

like DAX Studio are really useful for optimizations.

Next up we have the visual display bucket

and this bucket shows the amount

of time it takes for the visual you've selected to render.

In basic terms, it's simply just the time it takes

for that visual to kind of draw itself on screen.

Now, keep in mind that if the visual contains web-based

or geocoded images, those will also be included

in this duration as well.

And then last we've got our other bucket.

In the other bucket

it's really just all of the other tasks that the engine is

performing that aren't related to either creating the visual

or returning the results of the specific DAX query.

So the other bucket includes activities

like preparing the query, waiting for other visuals

on the page to complete their queries,

plus other background tasks that the engines are performing.

All right, so let's head over to Power BI

and we're gonna explore

the Performance Analyzer for ourselves.

All right, so back in our AdventureWorks report here.

And first things first, I'm gonna collapse these panes,

I'm gonna remove formatting.

We'll clean this up a little bit,

and then I'm gonna select the white space here just

so we don't have a specific visual selected.

So if we come up to the optimized pane

we're gonna click on Performance Analyzer.

And this launches a brand new Performance Analyzer pane.

And we'll change the size of this a little bit.

You can see

that we have one option here right now is start recording.

So if we click start recording, nothing's happened yet.

So there's two ways to start generating this list of items.

We can either click this refresh visuals button

and this is going to refresh all of the visuals on the page

or you could start interacting with the report

and that will also trigger this

Performance Analyzer list to populate.

So let's start off here

and we're gonna click Refresh Visuals.

So what this does is Power BI goes through

and it's going to refresh all

of the different visuals within the page

and it's gonna return the results in milliseconds, right?

And you can see we've got a bunch of different values here.

And if we expand one of these

we can start at the top of the list here.

Like with our left nav background

you can see that the visual display took 604 milliseconds

and then our other bucket took 453 milliseconds.

Because this is more or less an image

there's no DAX associated with this, right?

So it makes sense that we're not seeing that DAX bucket.

Same thing with our next shape here.

We've got this shape that's highlighted

for this first kind of KPI card.

And you can see again in total it took 1,049 milliseconds

for this to initially load.

The visual display was 604

and then the other bucket was 444.

So pretty straightforward there.

Let's collapse these and let's head down to our first card.

So now, because this card actually contains a DAX query

we now have that detail.

So again, this is a long time right to wait

for this to load, 5,966 milliseconds.

But what's interesting here is the DAX query was 408,

the visual display was 23, and the other component, right

that time that we spent waiting

for everything else to kind of load

or render ahead of this took 5,535 milliseconds.

So again, you get some really interesting insights here

as you start poking around at how long load times take.

All right, so at any point here, you know, say you're happy

with the results here, you kind of understand how

things are working, click stop.

We can close out of the Performance Analyzer pane.

And then again, it's a simple is turning it back on again,

start recording, and then we can refresh all

of the visuals to get those durations

and milliseconds again.

One other interesting thing to call out here is

that you can actually export the entire list here

as a JSON file, right?

So if you click export, it's basically

gonna give you this entire output in JSON format.

Pretty cool stuff here.

And at this point you may be thinking,

"All right this is great.

Like I can see what's happening

and I can export this as a JSON file.

But what does that mean?

Like what do I do with this?"

And there are some great performance optimization tips

on learn.microsoft.com.

So things like making sure

that your data model is really well built, right?

A lot of those best practices that we covered

in the modeling section, limiting the number

of visuals on the page to only what's necessary, right?

That's gonna help decrease those other wait times.

Making sure that you're applying restrictive filters so that

only the data that needs to be loaded is actually loaded.

You know, if you're using custom visuals

check on their performance, right?

Is a custom visual loading really slowly

and dragging down the overall performance of your report?

And you can also start digging into your DAX,

right within the Power BI report as well.

So let's say that you notice one

of your DAX calculations is running really, really slow.

You know, let's jump down to, you know

this revenue trending chart here, right?

But let's say your DAX query here is running

very, very slow.

You could dig into that measure and try and determine

what types of DAX optimizations could be made.

Maybe you're using iterator functions that end

up being slower than you anticipated

or maybe you're using a bunch

of different logical functions that are slowing things down.

So there's some instances like that where

you could kind of discern or understand like,

"Hey, if I update these functions that I'm using

within a measure that may impact the optimization."

But if you're really looking for some deeper optimizations

and understanding into exactly what's happening,

there are some great third party tools, like Dax Studio

and we're gonna talk about some of those quickly, next.