ASF 030: Benni De Jagere interview

ASF 030: Benni De Jagere interview


Benni De Jagere is a Senior Data Insights Consultant with a strong focus on the Microsoft (BI) Stack.
On a daily basis, he turns (large amounts) of coffee into insights for customers, and references witty British comedy way too often. Overly enthusiastic about anything data related, he’s trying hard to keep up with all things new and shiny.
Rumour has it that he’s also involved with a ragtag band of data enthusiasts, enjoying themselves whilst organising cool community things. They go by the name of … dataMinds!

This talk has taken place during SQL Saturday 2020 in Mechelen, Belgium on 7 March 2020 (Saturday).
Bear in mind that we were in a moment when the corona-virus just started spreading across Europe and 5 days earlier Microsoft made the decision that this year’s MVP Summit will be an online-only / virtual event.

Interviewer: Kamil Nowinski (T).

Audio version

Don’t you have time to read? You can listen to this as a podcast! Wherever you are, whatever you use. Just use the player directly from this site (above), find it on Spreaker, Apple Podcasts, Spotify (new!) or simply download MP3. Enjoy!


KN: Hello everyone, today we are sitting in SQL Saturday in Belgium. But the city we’re sitting in is… Mechelen?

BJ: Mechelen. The pronunciation is quite OK.

KN: OK, I’m trying my best. And we are sitting with Benni. Could you introduce yourself?

BJ: Yes, so I am Benni De Jagere. I am a Belgian Data Platform consultant, doing all sorts of things in the Microsoft Azure Data Platform stack, and on top of that I try to do some things in the community with a user group called dataMinds.

KN: So I know there is the dataMinds user group and there’s also a conference called dataMinds Connect.

BJ: Correct. dataMinds does a few things each year but the pinnacle of our events is dataMinds Connect. It’s a two-day conference where we do a pre-con track, which means a day full of sessions by one speaker, going in-depth into a topic ranging from introduction to Databricks and going deeper or Power BI, how to govern it, and all sorts of things. The second day is a conference day where there are six tracks with six sessions each of 60 minutes, so six is a number that comes back a lot. And it is 60-minute sessions by all sorts of different speakers, like we have at SQL Saturdays for instance, but a bit bigger.

KN: And more commercial?

BJ: Not more commercial. It is still a user group that drives it, the only difference is that they effectively need to pay to attend, unlike SQL Saturday.

KN: That’s what I meant, OK, by commercial I mean that it’s not a free event.

BJ: It’s not a free event, but it is not for profit-driven, meaning that every penny that we keep afterwards we put into different things like today, SQL Saturday.

KN: So it’s dataMinds Connect. And when does it happen this year?

BJ: Well, dataMinds Connect will take place on Monday, October 12 and Tuesday, October 13.

KN: How many people do you expect?

BJ: Well, last year we had 500 people on the second day, the conference day, and we had 380 people on the pre-con day. And we’re aiming at 50 more each day so 430 on Monday and then 550 on Tuesday.

KN: That means this is quite a big conference.

BJ: It’s getting there. The thing is that we don’t want to grow too much at once, because we quite like the venue we’re at right now. It’s a nice building, the insides are good, we love working with those people. And if we grow too fast too soon, the venue cannot handle us anymore and then we have to go venue shopping, which is a hassle.

KN: OK. What kind of speakers do you have at the conference?

BJ: It ranges all sorts of things. It goes from people we get from Microsoft, for instance, last year we had Kasper de Jonge from the Power BI CAT Team coming over, Patrick LeBlanc, Adam Saxton, two MVP experts like Ben Weissman, Enrico van de Laar, more experienced speakers as well, and we also try to actively promote getting Belgian speakers, because we are still a user group, so we actively want to get Belgian people there as well. So there is a fair portion of Belgian people, but one of the things that we put an emphasis on, that we try really hard, is the track that we call the newcomer track. And this is a track that we’ve borrowed the idea from DataGrillen, from William Durkin and Ben Weissman. We loved the idea that they did there and it is effectively just people that have never presented at a bigger conference. Inexperienced is a bad term but they are not experienced speakers at all. So we paired them up with a buddy and they help each other, they write up the story, they help them with the slide design, do some dry runs of the talk, and on the day itself they are inside the room as well, making sure that they have a friendly face to look at, that they are not alone in the big room. And we’ve done it for two years past and it is a big success, I’d like to say. Because there are people that have presented in a newcomer track, they’ve effectively felt the rush of presenting in public and they’re doing it still to date.

KN: Fantastic, a great achievement.

BJ: It is. We saw the same thing happening with the DataGrillen track and then Data Scotland does it as well and it is fun to see new people coming up, because we don’t want to see the same people every single time again. Some diversity is good and that’s what we’re actively trying to achieve.

KN: Seeing the same people all the time might be a little bit boring for attendees.

BJ: It’s fun because you get to know those people more in a deeper way as well, but seeing new people present is also good fun, because they have some different insights and they have a different way of talking. And the main thing is as well is that one of the confusions that we had about the newcomer track is that it’s all going to be basic sessions, it’s all going to be introductory level sessions, so it’s going to be the junior consultants presenting. And we’ve actively negated that confusion over the past years they are all people that are experienced in what they do, they know what you’re talking about, they’re not going level 400 or 500 expert level deep dives…

KN But also not 100, yeah?

BJ: No, They do 300 solid sessions, meaning that they know what they talk about and they do a good job with that.

KN: They’re very well prepared because they want to do it very well, especially at their first session.

BJ: They are overly prepared, yeah. But it is also good fun to see the mentors or buddies, or whatever you want to call them, because they get a lot of good fun out of it as well. Because they sacrifice free time for it, they are well aware of it, but they help launch a new speaker as well, which for them is amazing as well.

KN: You mentioned last year you had Kasper, Adam, Patrick. That’s when the dinner happened, the call happened that I saw yesterday on Twitter?

BJ: Yeah, the infamous Teams call, yeah.

KN: Some people maybe, if they use Twitter, they know what we’re talking about.

BJ: Yeah, but it is good fun. A speaker dinner is intended for all people and all speakers to unwind a bit, to get to know each other and drink if they want to. And it is good fun.

KN: It’s not only work, it’s also fun.

OK, tell me exactly what you do for a living.

BJ: I am a consultant and at this point, the last few years I am a consultant that does a lot of shorter assignments together. Meaning that I don’t have a client where I’m working full time for, I’m doing a few other clients that are shorter assignments or one day a week and then doing those kinds of things.

KN: Oh, really short assignment.

BJ: It is running for a longer time but it is one day a week that I spend time for them. And it ranges from doing Power BI development, report developments for a pharmaceutical company, it is helping a public company manage their Power BI premium capacity and helping them tune their Power BI data sets, it is helping another company architect their Azure Data Platform landscape for the coming years. So it’s all sorts of things for each other. It’s fun, it’s challenging but it’s demanding as well because you need to stay on top of your game of all the different things.

KN: So it’s mainly Power BI but not only?

BJ: The majority of my time goes to Azure Data Platform and landscaping architecture things. But the thing I like to do a lot at this point right now is Power BI. That’s what I enjoy most at this point.

KN: So for the last year you’ve been mostly engaged in Power BI?

BJ: Yeah, all sessions that I’ve presented the last two years of all have been Power BI.

KN: You know why I was asking about that? You know where I’m going?

BJ: I don’t know.

KN: So quite recently, this month, last Sunday, you have been awarded the MVP title. How did you feel?

BJ: Well, surprised. I can genuinely say that I was surprised because I had reopened my nomination process not that long ago and I hadn’t even gotten a word “we’re working on your application”. So I was genuinely surprised but then the second you read the email, because I was doing things on Sunday in another one of my spare time hobbies, it felt good because it’s a form of recognition for things that you do, and it’s nice to be appreciated.

KN: So after that Gianluca mentioned on Twitter that he is no longer the tallest MVP among the members, yes?

BJ: Yes, I’m not sure on how many centimetres I’m taller than him but Gianluca will always be the one with the best taste for coffee. No one will ever take that away from him.

KN: So tell me how tall you are.

BJ: Well, in centimetres it’s 2 meters 3 centimetres, and I’m told that in British size it’s 6 foot 7.

KN: I’m still using the metrics.

BJ: Metrics is better, it makes more sense. 203 centimetres.

KN: Next time, when I’m speaking to Gianluca, I will ask him how tall he is. OK, so tell me in terms of MVP: how does this title, this recognition helps you in your work, in your commitment to the community. How does it work for you?

BJ: Well, the thing is that one of the reasons I gave to Microsoft as well is that as a user group, having someone with an active MVP title helps a lot in making connections with other people and with Microsoft, because it’s easier to make that communication for the first time with them. So that is something that will definitely be helpful for the user group that we’re doing. And I think for me personally well it’s a nice benefit that I get some Azure credits to play around with. That’s always welcome. But I think it’ll allow me to go deeper into some of the topics from Azure Data Platform that I may have not been going into before and then start digging into those as well because those product group interactions that you can do, I think those will be valuable for me as well.

KN: Are you struggling to read through all those emails that you’re getting? Especially in the Power BI area there’s plenty of them.

BJ: For now it is the SQL Server that’s been most active for me. But the second I saw it was emails coming in, I made a rule in Outlook that shifts them all to a separate folder that I read when I can. But it is a lot of emails yeah, and I’m not gonna read them all.

KN: Oh, definitely. That’s almost impossible.

How do you prepare for a speech or if you’re preparing your session?

BJ: I do not like to be over-prepared. I still want some… improvisation is not the proper word but I still want some fluidity in my sessions. So the main thing that I want to prepare for myself always is making sure that my demos are working, that they do what I want them to do, and the first time I’m doing a session, I make sure that I’m very well prepared for it, but then all the repeats that I do of a session, making sure that I can get the general flow of the session, that I know what the storyline is and that I can build up on that storyline throughout the session. I’m not that kind of person that rehearses every single bullet or every single visual that I have in a PowerPoint, because that limits me in the way that I present. It just doesn’t work for me.

KN: So do you change your slide or demo at the last moment like last day or something like that?

BJ: No.

KN: Never?

BJ: No.

KN: So when do you freeze the things?

BJ: I haven’t done it yet because the sessions that I’ve been presenting have been quite stable as a topic. But yeah, I think it would depend if it’s a breaking change to something, you’d have to change it. But other than that I try to freeze it out two days before, so that my mind has some time to mull over some things, to get some talking points in my head, and then two days before is my set date.

KN: Yeah, I’m mostly talking about a situation when you rehearse the session and a lot of ideas come to your mind and “maybe, I’ll add this and this, maybe I’ll change the order of the slides or maybe I’ll add something to the demo”.

BJ: The thing is that I do a session about Power BI dataflows as well and I’ve had it a few times already that they’ve done some changes that break some things that I noticed an hour before the session, that I’m doing a final check of a demo and “ahh, they changed this”. And then I’m glad that I’m doing it an hour beforehand and not noticing it during the session. Because I’ve had that experience as well, not in community presentations but at customer presentations. It’s like “the buttons should be there in the corner but it’s gone so they moved it somewhere – please allow me to find it”.

KN: So, about this newcomer track: I know, we all know that is quite a challenge to encourage new people to start speaking. So how can you encourage people to start doing that?

BJ: Well, I’ve found that the way that you can get people to convince them is by face-to-face interaction, just explaining to them one on one what it is, what they can do, what they need to prepare for it, and what it’s going to be like. Like the general approach and sending an email to 20 people, like: “hey, do you want to do a newcomer track?”, that typically does not get that much response. And it is engaging with those people one-on-one that helps the most, I think. And we’ve tried it over the past years and it’s worked out quite well.

KN: Because there’s live interaction, you can answer for the new questions or explain how it works.

BJ: For instance we had someone present a user group session for us in February, and during the introduction to that talk I explained dataMinds Connect what we were going to do and the newcomer track as well. And then afterwards he said like “yeah, I actually like to do that, the newcomer track. Can I still do that because I’ve done this user group session?”. And yes, you can still do that. The only limitation we put on is the fact that you cannot have presented at another technical conference or another bigger event. So anything like customer presentations or user group sessions – they don’t matter.

KN: And we know that the IT market specifically needs a lot of people and a lot of specialists. So what kind of hints would you like to give young people to start working on the IT market? How to start, basically.

BJ: I’m a firm believer in functional, non-technical soft skills. I think that those are the things that help people advance more firmly in an IT career. But it is quite clear that data and data-related jobs are going to be key in the future. It is quite clear. And something that I still regret of not having done more in the past is having a more statistical, mathematical base for allowing me to do those more data science kind of things. Analytics, like the things we do in business intelligence, those I can do perfectly but I’m not the kind of person that’s going to draw out neural networks or those kinds of things. So if you were to talk about data engineer and data scientist, I’m clearly on the data engineer side. And then, I would have loved to be doing more of a data science side as well but you have to make decisions sometimes and you have to choose what you’re going to do. And for me right now I’ve said: I’m going to go deeper into data engineering and then I’ll leave data science up to other people, because data scientists need me as well because they need their data to be prepared properly.

KN: Yeah, absolutely. There’s still a lot of cooperation.

BJ: The thing is that, focus on those mathematical skills, the statistical base, that’s going to be good for anything you do with data. And then, work on data manipulation, is it SQL, is it Python, whatever, but work on those kinds of skills. And I think that they will get you very far in an IT data-related career.

KN: Yeah, also transformation data. Now you can do it in Power BI as well with dataflows. How do you think it works? I mean, you are much closer to Power BI and dataflows.

BJ: I hate it and I love it. The concept of dataflows is excellent. I really love the concept. The thing is that for it to properly work in a way that I myself as a data engineer, an ETL guy, for that to work you’re pretty much stuck to premium. So dataflows is not a premium product. If you talk to any of the Power BI CAT Team members they’ll hammer that into you: Data Flow is not premium only. But you need some of the key features, like in Lake computing and then linked entities, and they require premium capacity. And they allow you to make a better daily lineage across your different entities and for it to work the way it should be. But dataflows in general, I’m positioning it these days with customers like, for instance if they have some master data that needs to be managed or we have an Excel file with a mapping or something. That’s where I’m positioning dataflows. I’m like: OK, use dataflows, land it somewhere in Data Lake Storage and then we’ll pick it up with Data Factory or Databricks to use that mapping file in our own thing. So you’re effectively just making them the responsible ones for their own master data or mapping files.

KN: They know the data. They know the quality of the data.

BJ: Yeah, we just tell them: “have it there by that hour of night and then we’ll pick it up and do the thing that we need to do with it”. And that works quite well. And then, the other way I’ve positioned it is allowing certain key figures in an organization that have some technical skills, I’m not saying IT users but people that have been working with Power BI, people that know how to work those things, have them build small prototyping entities.

KN: Yeah, I wanted to ask about POC [Proof of Concept].

BJ: Not that much POC but it can be used for that as well, but also for, let’s say that you have a Data Warehouse or Data Lake and there is something that’s missing in there, according to business. And your data engineers don’t have capacity for it, they cannot build it. I have used it in the past to have those people build out a small thing for themselves, have them build it the way it should be, the way it should be working and then, after they’ve had some collaboration with IT, the way it should be and how it effectively should be working. Just transition that and do change of ownership to IT and then they incorporate that into the entire Data Lake flow. So that’s how I’ve positioned it in the past as well. And that worked pretty well, but those key users had decent knowledge of what they were doing, and I do think that that is key for a setup like that to work. But it is an exciting tool, it is showing potential, but the pro version has limits if you want to build an entire ETL flow for that. But on the other hand, you shouldn’t be building an entire ETL flow in dataflows. For many reasons. We’re all data professionals, we’re all BI engineers and we know that some things need to be done properly instead of in a self service tool. But there are options to do it.

KN: Performance-wise as well?

BJ: Performance as well. For now incremental refresh only works in premium in dataflows. So you always have to load the entire data set, which doesn’t always work that well. And some sources like CSVs have not that good of a performance if you load them in with dataflows. So there’s clear differences, but yeah, like I mentioned, it is a good tool for business people to build prototypes in or build gaps into some things, but you’re not building enterprise data lakes with dataflows. That’s not the way you should position it.

KN: Yeah, that’s what I thought. It’s perfect to build something at the beginning, it’s perfect to prove that it might work with that idea, with that set of transformations, but for the bigger solution, you need to just transform the solution basically and use different tools. What about the Common Data Model in Power BI? That’s a cool feature, yeah?

BJ: It is cool. I want to do more with it, though. For now it’s just landing things in there, making sure that it’s just properly formed, that it’s been usable, but the main thing in there as well is that the data is stored in CSV.

KN: Do you know why?

BJ: I don’t know why. That’s one of the product group interactions that I want to do in the future. But it is stored in CSV files, which is OK, but that just underlines the previous statement that I made that you’re not going to build enterprise-like stores and their data lakes in there. But it is a cool feature, especially one of the things that I’m going to be working on in the next few months with a few of my Dynamics 365 colleagues is using data that we land from dataflows into Common Data Model to be picked up by Common Data Service by Dynamics 365 and the other way around, so that we can tie those two together. That’s one of the things that I’ll be doing in the next months.

KN: Oh, pretty interesting.

BJ: Yeah, I want to try it.

KN: But still, under the hood, there’s Azure Data Lake Storage, yeah? So it’s very effective, very performant in terms of read and write, so even CSV shouldn’t be so bad. Obviously, there’s some other different formats…

BJ: If I can get Parquet files as a Common Data Model storage, I’d take it with both hands.

KN: Maybe in the next version we will have an option.

BJ: We can only hope.

KN: What about your work-life balance. How can you cope with that?

BJ: I don’t have a balance I think. The last few months have been wild for me. But I’m a consultant and in general the agreements that I have with customers and my employer is get it done, and if you can do it in 7 hours it’s okay, if you can do it in 9 hours, well… So I can do some balancing and some flexible things there, for instance when I need to leave at 4:30 to make it to a venue for a user group meeting in time. They’re not that fussy about it. I just need to make sure that the job is done. But other than that – we do quite a lot of user group meetings throughout the year as well. We do SQL Saturday, we do Global AI Bootcamps and AI Nights, we do a two-day conference, so there’s a lot of time that gets into that as well. But it is fun to do because otherwise I would not be doing it. But other than that, I do try at least two or three times a week to go out to karate, try to practice.

KN: Yeah, exactly, that’s my next question. I’ve seen on your Twitter profile…

BJ: Full-contact karate, yeah. It is an entirely different thing. But also it’s a sport that requires a lot of respect for each other, and I really appreciate that. For instance back in the day, when I still used to compete, I don’t do it anymore these days, you effectively get on the mat, you give each other a very hard time. You beat each other, you kick each other. But the second it’s over, you get off the mat and you hug the other guy and you start cracking jokes. There’s such a deep form of respect for each other that it is genuinely a good thing to do. And yes, you’ll get bruises or yes you’ll have some pains afterwards but yeah it is still a fun sport to do.

KN: I’m just wondering, you know, who might be that brave to stand in front of you and fight with you.

BJ: Well, a lot of people. They think it’s a challenge. And the general thing is that I’m quite tall and I’m not the lightest one as well, so they can work on speed to get past me. I’m not that fast, so if they’re fast and they can do Muhammad Ali like “float like a butterfly sting like a bee”, that works for them as well. But yeah, I land fewer kicks and I land fewer punches but the ones that do land, they come across well. But like I said, it is all good fun. We have these things called 50 people kumite, which means that we do 50 rounds of sparring against all people that are there. A hundred people get together, we do 50 rounds of sparring together. And it’s 50 times 2 minutes after each other that you do some sparring matches. And there’s two breaks in between. So you die 27 times at least, but it’s a way of the sport, I think. If you don’t do that kind of sport, it is difficult to understand, but for us it’s a lot of fun. But it is also a matter of respect because sparing is not beating the pulp out of each other. It is making sure that you try out something new and yeah, you hit each other but you don’t hit each other to effectively break your ribs or something.

KN: That’s interesting. So how often do you attend?

BJ: I try to get to practice two to three times a week. Sometimes it works, sometimes it won’t. If I’m travelling abroad it’s not but in general it works out quite well. But for me it’s a release mechanism to just relax, to beat a punching bag or to just let everything go.

KN: Do something physical, not just mental.

BJ: Yeah, I need it, I really do. I’ve had some periods of injury in the past, I couldn’t go train for two months and those were two months that I was really antsy and yeah.

KN: And you also mentioned that you are a coffee addict. Are you? How many coffees do you drink a day?

BJ: Yes, it depends really. I’m trying to cut back. The thing is that I drink a lot of coffee but I drink it on a low strength and I take half cups, meaning that I still drink a lot of coffee, because it forces me to get up to walk to the coffee machine, to take a lap around the office and not sit behind a desk all day. But six cups a day is not an oddity. And if I’m working towards deadlines or if it’s days like today where I haven’t slept that much, it’s more than six, definitely.

KN: It was because of yesterday’s dinner.

BJ: Well, yeah. I got back to the hotel at around midnight and I was here at 7:00 so not that much sleep.

KN: So your hobby is karate… How do you say it?

BJ: Full contact karate. Shinkyokushin.

KN: It’s a completely new area for me.

BJ: Yeah, in Europe it’s quite big. The countries like Poland and Lithuania and Bulgaria and Hungary, they’re really big in that sport. Here in Belgium it’s a hobby. The countries that I’ve mentioned before, they’ve got professional athletes that get paid to train, paid to compete. But out here in Belgium and the Netherlands and France as well it’s something we do after work. So we get home from work, we go train, so as for level, we’re not that high level as the countries like Poland in Lithuania. They are the absolute top in European rankings, but it is still Japan that is the absolute Lord and Master of the entire thing. They start training younglings from 3 to 4 years old and they train really hard a lot. And they also have professional athletes. They start training them in the techniques etc. and not starting to train them to be killing machines from age 4. But Japan is also really protective of their competitors, underage competitors. They are very heavily protected up until 18, and then from the second they get 18, it is full contact as we all do it, meaning no protection. Just groin protection.

KN: So how much do you travel across Europe or across the world?

BJ: The last months have been a bit more, because I’ve been speaking more. But I try to limit it to one European trip a month. That’s what I try to limit myself to.

KN: I also try to limit you know like up to one even per month averagely but it doesn’t always work.

BJ: Yeah, February was… I’ve said yes to a few things in Belgium that were happening, organised a few things myself, and then some other international things came along as well and I said: “yeah I do want to do that” because I got a selection to SQLBits last Sunday as well. They emailed me and said that “Do you want to come present?”, and I’m like “Yeah, well, I do”. I’m travelling already in March in April but it’s SQLBits. You’re not gonna say no to SQLBits. So I try to limit myself but it does not always work.

KN: So obviously you’re going to SQLBits?

BJ: Yeah.

KN: SQLDay this year as well.

BJ: I am going to SQLDay as well, in Poland, yeah. It will be my first time in Poland.

KN: Fantastic, great to have you there. So what do you think about what’s happening due to this corona-virus and MVP Summit in the next week?

BJ: Well, I get why they do it. Especially because the state of Washington has been hit hard in America, that’s why I really understand why they’re doing it. For me this year is not that big of a deal because I wouldn’t be able to attend anyway. They’ve notified me on March 1st and then it was a bit too short notice to get a trip to Seattle and hotels and all those things to get it sorted out, and I already have customer visits planned that week.

KN: Yeah, obviously, you have just been awarded, so it’s too late. I’m sorry, I mentioned that to you before that in my mind I thought that you had been already an MVP for years.

BJ: No, but we’ve got the dates for 2021 already and I’ve blocked a mini schedule and I want to travel there if I can, so yeah. But I definitely don’t get to go this year. For me personally it’s not that big of a deal that it’s been cancelled but I understand why they do it. Same reason why they’ve locked down campus, that they’re asking everyone to work from home. I really get it.

KN: Yeah, absolutely, it makes sense. And to be more precise, Microsoft hasn’t cancelled that event. It’s just been moved online.

BJ: I’ve yet to see how it effectively will work out but the thing is as well if I have to be at a client during the day, and if there’s a session at 3:00 a.m., I have to be reasonable and make sure that I can do my client work as well because they expect me to do some work. So I’ve yet to see what it’s going to be, what the schedule is going to be like, if there are going to be recordings or video on demand or something but I’ll see when it gets there.

KN: Recordings are quite good but it doesn’t always work after the event.

BJ: It’s not the same.

KN: It’s not easy to find the time to watch the video again.

BJ: But the thing is that the MVP Summit, the content alone is great, but the main reason you go there is the interaction with other people. It is like this very big reunion of all people coming together and that’s the main reason why I would want to attend there as well.

KN: So your session today was about Power BI troubleshooting your Power BI report performance. Could you tell us a few things about what you presented?

BJ: Well, I’ll start off with a sentence with what I didn’t present. It was not a DAX deep dive performance session. I leave that up to Marco Russo, Alberto Ferrari, Kasper, those kinds of people that are DAX experts. I go to great lengths to avoid having to write complex DAX, which means that I model things. I star schema all the things, which is the baseline of the entire session. It is fix the things in the model to avoid having to do complex things afterwards and that is the main reason how you can get your performance to be better. Because the things I run into at clients is that they dump Excel files in there and there we’re starting to do BI-directional filtering all over the place and they’re starting to write complex DAX with things that could be done probably more efficiently. So a star schema is the base of everything in Power BI. That’s my opinion on it and I’ve found many people to share that opinion. I’m aware that star schema is a difficult thing if you’ve not learned it before from a BI perspective or something, data warehousing, but I do feel that it’s a key skill to learn if you’re serious about Power BI. But other than that it is getting started with the tools that Power BI has itself, to performance analyzer, how you can see what’s effectively running slow, what can you do about it, and then using a few community tools like DAX Studio, VertiPaq Analyzer, Tabular Editor, Power BI Helper, those kind of tools to start pinpointing your way through what be improved upon. So typically what I do is when someone comes to me and says: “my report is slow, fix it, please”, then what I typically do is I try to isolate the query that runs slow, or the queries that run slow, then I open up VertiPaq Analyzer, well DAX Studio to see what’s effectively in there and how its distributed between the different engines in Power BI, open up VertiPaq Analyzer to see what the model internals are looking like, and then go from there. And that’s usually how I get started on working on those things. And if it’s effectively well-modeled and the DAX has to be fixed, then I’ll do that, but usually I go for the model first.

KN: So basically the rule is still valid, the rule that says that 80% of the problems comes from the model, the design.

BJ; Yes, the design, the data quality, the data cleansing.

KN: It’s still true even for Power BI, where the whole operation is in the memory.

BJ: The thing is that I also see things in reports that people have built that they’re fixing things in calculated columns, like for instance I want to have an age bucket. Is it age 30 to 50, is it age 50 to 70, and they start doing that in calculated columns in Power BI, which is… Yes, OK, you can do it but it would be way more efficient to have it in Power Query, where you can fix it and it’s been optimized in the model, or even if you can, put it in the Data Warehouse, just to do it as soon as possible in your process, so it can be reused throughout all the reports that you’re doing. Because if you’re doing it in Power Query alone and you have to use that same age bucket somewhere else, you have to remake it. So putting it as soon as possible in the process works. And that’s what data warehousing is about. That’s what we’ve been telling for a few years.

KN: That’s cool. And also that’s good to know that some tools exist to do that diagnosis.

BJ: There are many community tools out there, and they are of great help, yes.

KN: OK, Benni, thank you very much for this conversation. And at the end, could you tell us where we can find you, where people can find you, follow you?

BJ: Well, they can find me on Twitter and LinkedIn. It’s mainly just my first name and my last name – Benni De Jagere. But good luck on trying to spell that if you’re not Flemish.

KN: Definitely! I will put the links under the transcription. OK, so LinkedIn and Twitter. Do you still have your blog?

BJ: I still do but it’s not been updated for 5 years. That’s one of the things that I’ve been telling myself for the past year to start doing again, but work-life balance is getting in the way.

KN: I definitely understand it, you are already very, very busy doing all this stuff. So yeah, I appreciate that, OK.

BJ: It’s fun, so that’s the reason why I do it.

KN: Yeah, exactly. Thank you very much again!

BJ: Thank you!

Useful links

Benni’s profiles: Twitter | LinkedIn
Related events: dataMinds Connect | DataGrillen


Previous ADF - Continuous Integration & Deployment with Azure DevOps
Next ADF - Deployment from master branch code (JSON files)

About author

Kamil Nowinski
Kamil Nowinski 200 posts

Blogger, speaker. Data Platform MVP, MCSE. Senior Data Engineer & data geek. Member of Data Community Poland, co-organizer of SQLDay, Happy husband & father.

View all posts by this author →

You might also like

Podcast 1Comments

ASF 032: Jennifer Stirrup interview

Introduction Jennifer Stirrup is a data strategist and technologist, Microsoft Data Platform Most Valuable Professional (MVP), Microsoft Regional Director, Microsoft Certified Trainer, founder of Data Relish Ltd, diversity and technical

Podcast 1Comments

ASF 031: Paul Andrew interview

Introduction Paul Andrew. Principal consultant and architect at Altius specialising in big data solutions on the Microsoft Azure cloud platform. Data engineering competencies include Azure Data Factory, Data Lake, Databricks,

Podcast 0 Comments

ASF 002: Greg Low interview

Introduction Dr Greg Low is well known in the SQL Server community, an internationally recognized consultant, trainer and author of a number of Microsoft white papers, blog posts. Greg holds a


No Comments Yet!

You can be first to comment this post!

Leave a Reply