0:00Hey, it's Tim here. In today's video we're
0:01checking out the new Amazon S3 connector
0:04that's currently out in beta straight off
0:06the back of conference.
0:07As ever, let's get started.
0:09Okay, so we're going to try this in two
0:10places. First we're going to try it in
0:12desktop. You can skip ahead to the all
0:14cloud version where we try it in Tableau
0:16cloud.
0:17But for this one, I'm starting off in
0:18desktop because I just want to show you how
0:20to install the connector when working off
0:23Tableau desktop.
0:24So if you go to the tour service section,
0:26you can see there's a list here. This is
0:28normally the most recent connections, but
0:30you can see I've already used the S3
0:32connector.
0:33So it's available. But if it's not there,
0:34if you select more, you can go down to this
0:36list and you can see that I actually have
0:38it here.
0:39And if you're opening Tableau and you don't
0:41have the Amazon S3 connector, that's
0:43because it's not installed by default.
0:45To install it, what you want to do is you
0:46want to head to the Tableau exchange and
0:48this works for Tableau prep, Tableau server
0:50and Tableau desktop.
0:52You go to the connector section of the
0:53Tableau exchange. The Tableau exchange is
0:56just the marketplace for connectors,
0:58extensions and templates for Tableau.
1:00And then you can see the S3 connectors
1:02right here. You download it. And what this
1:04does is it downloads a file to your
1:06downloads folder or your desktop, wherever
1:08you tell it to go.
1:09And then once you've got that file, you're
1:10not done. What you need to do is you need
1:12to put it in one of these directories.
1:15It's super important you do this because
1:17Tableau looks in these folders to see
1:20additional connectors. And so you can see
1:22here it's pretty consistent.
1:23In fact, in desktop and prep, it's
1:24virtually the same place. So I actually
1:26have those locations open here on my
1:29computer.
1:30You can see my Tableau prep repository. If
1:32I go to the connectors, I have two TACO
1:33files. This is the file extension that adds
1:36additional connectors.
1:38So Google Analytics is one I've just done a
1:39video on just literally a couple of days
1:41ago.
1:42And then here you've got the S3 connector.
1:43And you notice that these have versions. I
1:45'll come back to that in one second.
1:47And the same is true with Tableau. So Table
1:49au desktop has the exact same folder
1:50structure.
1:51If you go to connectors, you can see I have
1:53the two files. And so once you've put it
1:55there, you open Tableau.
1:57If you've already got it open, close it,
1:59open it again, it will load these
2:00connectors and you'll see it in this list
2:03that loads up here along with the other
2:05data connectors, even though it lives on
2:06the Tableau exchange.
2:08So this is how you get this file on here.
2:12Now before I hop into the product, it's
2:13just worth noting that this is a beta.
2:14And obviously what that means is Tableau
2:15will keep updating this. I'm not aware of a
2:18mechanism that can update these TACO files
2:21as these get updated.
2:23So I think what you have to do is you
2:24bookmark this page and come back and you
2:25have to check the version you're running
2:27versus the version that's on here.
2:29I hope in the future Tableau does roll out
2:31some sort of update capability.
2:33So when you're updating desktop, are you
2:35doing any patches? It checks these
2:37connectors and downloads the latest version
2:39for you.
2:40But because you're typically putting a file
2:41on a very sensitive place, this is probably
2:43going to have to be a manual operation or
2:45there is a sort of command line
2:48installation mechanism for Tableau.
2:50And if you're working in enterprise, it
2:51should be possible for your IT admins to
2:53put a specific file in this directory.
2:56So have it done that way if that's going to
2:57be something that you're going to use
2:58across the organization.
3:00Now, once you've got everything set up, you
3:02can go ahead to Tableau prep or Tableau
3:04desktop. I'll show you Tableau prep briefly
3:05.
3:06If I go to the plus sign, you'll see that
3:08if I actually just type in S3, you see that
3:10it comes up right there.
3:11And if I click on that, it opens up the
3:12connector. I'm going to show you how to
3:14connect here in desktop because that's what
3:15most people are using.
3:17So we'll go ahead and click on the
3:18connector and I'll show you how to set this
3:20up.
3:21When it loads for the first time, it
3:22actually loads a web window.
3:24And you can see here it has a timer saying
3:26elapsed time, eight seconds executing query
3:29. And then for me, a few seconds later, a
3:31web window shows up.
3:33Now, when this was happening in those eight
3:34seconds, the first time I run this, I was
3:36actually clicking around thinking, hey,
3:38what's going on?
3:39And the simple act of clicking on Tableau
3:41actually sent that window to the background
3:44.
3:45So you might be looking at a screen like
3:46this. The time is going up and you're
3:48wondering why is this taking so long?
3:50Well, it's because it's behind the window.
3:52So just make sure you look out for that.
3:55That caught me out for about a couple of
3:56minutes and I realized, oh, wait, this is
3:57probably behind the window.
3:59Anyway, now we're here. It wants four
4:01things. The essentially the details around
4:04your Amazon S3 bucket.
4:05If you're not familiar with Amazon S3, it
4:07stands for S3 stands for simple storage
4:10system, essentially simple storage system.
4:12I hope it's simple storage system. If it's
4:14not correct me in the notes below.
4:15I used to be certified in AWS. I need to
4:17get back on that, obviously. But it wants
4:20four things.
4:21The bucket region, the bucket name. These
4:23are two pretty standard things to get if
4:24you're working with S3.
4:26Then it has access key ID and a secret
4:28access key. These are things to authent
4:30icate you so it knows that you have access
4:32to look at this.
4:33So you should have this already set up and
4:34you should know these details if you need
4:36to use this connector.
4:38If you don't know these details, find them
4:40first, then you can use this connector.
4:42So I'm going to set this up and then we're
4:43going to skip past this because I don't
4:44want to share my details so you can all go
4:47and use up my AWS credit.
4:49So let's go ahead and copy my details. OK,
4:52so I've got all my details entered.
4:53You simply go ahead and hit sign in. And
4:55what this should do is it should load up
4:57the next step.
4:58Now, the next step shows you what's in your
5:00bucket. Essentially, the first step is it
5:02scans your bucket.
5:03And you can see here it says it found 60
5:05files.
5:06You can actually dismiss that if you know
5:07what you're going to find. You can actually
5:09get rid of it just here on the top right.
5:11And all you have left is essentially
5:12directory structure.
5:14And the reason I've got a few files here is
5:15because I wanted to see what this connector
5:17could do, because you can upload virtually
5:19anything to S3.
5:20So I wanted to see, hey, can this S3
5:21connector connect to a range of different
5:23files?
5:24And so the two things I know it can connect
5:26to are CSV and Excel files.
5:29It can't connect to hypers. So this is not
5:30going to be a way to upload a hyper to S3
5:33and then have this connector read that
5:35hyper in a sort of consistent way.
5:37And that's not currently possible. You're
5:38still going to be better off publishing the
5:40hyper to Tableau Cloud or Tableau Server to
5:42use that in a more sort of flexible way.
5:45That said, I've had a few bugs opening
5:47Excel files. I actually haven't been able
5:49to get it to work.
5:50It might be that my version of Tableau is
5:52not actually supported on my version of Mac
5:54OS.
5:55I'm using Mac OS Ventura. 23.1 does not
5:58support that. 23.2 will support that. So
6:01hopefully this will be more stable.
6:03So I'll go ahead and I'm actually going to
6:04use a CSV file. This file I'm going to use
6:06is actually this one, London Fire Brigade
6:09incident data for the last three years.
6:12And so we'll go ahead and connect that. You
6:13'll see something really interesting with
6:15CSV files.
6:16When you do that, it goes and queries the
6:17whole bucket. And just like on desktop, you
6:20can see that it's not only just brought
6:21back the file we selected, but it's
6:23actually brought back all the CSV's in the
6:26whole directory, including the ones that
6:29were inside of this other directory.
6:31So if you have an S3 bucket where lots of
6:33people are putting files, maybe every
6:35single day there's a transaction or maybe
6:37every single day there is something going
6:38on, you should be able to use that set up
6:41to sort of bring them all in.
6:44I'm not sure you can tell this connector to
6:46continuously scan the entire bucket for
6:48everything and bring it all in.
6:50It doesn't seem to support some way of
6:52saying that. And I'd be interested to know
6:54if there maybe there's a secret sort of
6:55back end where you can query this with SQL.
6:58But you don't have that option here to
7:00create custom SQL that reads the S3 bucket.
7:03There's nothing like that. But you've just
7:05got a basic set up here.
7:07And you know what you could do is if you
7:08wanted to, you could actually select
7:10multiple files and you can go ahead and
7:12select new union.
7:14And if you drag that in, you then get the
7:15option to essentially specify what you want
7:17. And you can drag all these files in and it
7:19's capable of reading that as well.
7:21But for that, this is still in beta and I
7:23've not been able to sort of consistently
7:26get that to work. So I'm just going to show
7:28you one file type so we can keep this video
7:30nice and snappy.
7:31We'll bring in the London Fire Brigade
7:34incident data.
7:36And, you know, I'm constantly sort of
7:38having to battle me being on a Mac and also
7:41using a beta product to unstable things.
7:44And sometimes you experience sort of
7:45interface issues. Sometimes you experience
7:48bugs.
7:49Now, the thing to note about this file is
7:50180 megabytes. And don't forget, I'm using
7:53this on my desktop machine.
7:55So if we think about the network journey,
7:57my desktop machine is sending off a query
7:59to Amazon S3 in US East one.
8:01That's the region that my bucket is in. It
8:04's querying that data. And don't forget, it
8:06's doing it live. It's doing this live to
8:08the data set.
8:09So if you were to add, you know, update
8:12that file, I'm assuming it should be able
8:14to read that update in almost real time.
8:18And it's able to sort of go and access that
8:20data could take an extract.
8:22So it's local. But when we go ahead and
8:23connect to sheet one, we get everything
8:25that was in that file.
8:27Now, this is an annoying file because it's
8:29a CSV had no column headers.
8:31But I can reassure you that all the data is
8:33indeed here. And you can pretty much see
8:36that everything is working.
8:38There's a few. There's a lot of metadata
8:39missing from this. Everything is coming as
8:41text.
8:42It's clearly not. But you know, the way
8:44this was working, you'd hope that over time
8:48,
8:48this S3 connector gets a little bit smarter
8:50and can automatically identify some of
8:51these column types.
8:53There's definitely a column or two here
8:54that you'd hope Tableau would be able to
8:56identify.
8:57For example, you've got the cities. The
9:00London is a city here.
9:01You've got some towns here that it should
9:02be able to identify as geographical fields.
9:05But it might be that the way the S3
9:06connector is querying the file, it's just
9:08not able to deduce that metadata
9:10automatically.
9:11Like desktop does sometimes with CSPs. So
9:13that's something to watch out for.
9:15But this is in beta. This is not done. So
9:17that's still yet to come. And hopefully
9:18Tableau resolves that.
9:20So now here I am connected to a file live
9:22in S3. You can actually just look at this.
9:24That's going ahead. Let's go ahead and just
9:26do a basic chart. Just say this is working
9:28live.
9:29So we'll go ahead and get column 12. S
9:32column 12 is the postcode, the outcode.
9:35So we can go ahead and bring that in. And
9:37if we do that, we can then also go get the
9:40column.
9:41Sorry, the row count, which is just down
9:44here. And we can see how many rows for each
9:46data set.
9:47And that should give us an idea of the
9:49spread. So it seems that most of the Fire
9:51Brigade incidents were in Croydon.
9:54Again, I'm reading this in real time from
9:55Amazon S3 just here in Tableau Desktop. So
9:58this works really, really nicely.
10:00You can sort of get this to work and get it
10:02to work. In fact, let's just try and do
10:05something.
10:05Let's just try and say to Tableau that this
10:07has a geographic role. It's a postcode and
10:11it's probably going to complain.
10:13I'm not going to I'm not going to lie to
10:14you. I think it's going to complain because
10:16it doesn't have any country information,
10:18but it's actually been able to handle it.
10:19You can see it's actually been able to sort
10:20of generate that.
10:22So we can let's look at the locations. It's
10:26got a yeah, that is UK Kingdom.
10:30It doesn't know these four postcodes, so
10:32that's fine. You know, you could probably
10:35correct these.
10:37There's probably something wrong with these
10:38particular postcodes. I think it's probably
10:40just easy to and easy 3P.
10:45This this looks like a typo. So we could we
10:46could probably sort of do some data
10:48cleansing to these to make them work.
10:51W1A you don't normally in UK don't normally
10:54have an outcode that ends with a letter.
10:57Normally they end in a number. So these are
10:58definitely issues.
11:00And BF1 might just be another typo from a
11:01different sort of set up. So at least that
11:04we know is working.
11:06So we can go ahead, get a background. Let's
11:09see what we're going to do here. Let's get
11:10let's do streets.
11:13Let's do this one. This is a nice map. So
11:15it kind of looks a little bit more familiar
11:16.
11:17Give that a bit of time to load. And there
11:19we go. We can kind of start to see what's
11:21going on.
11:22And then what we can do to give us a sense
11:24of what the London Fire Brigade is facing,
11:28we can put that on size.
11:30We can make it red and we can just make it
11:33sort of stand out a little bit.
11:36Let's get my face out of the way here. Let
11:40's just make this a little bit larger and
11:42then bump this up a little bit.
11:45Yeah, there we go. So that's starting to
11:46look nice. Now we can start to see where
11:49the London Fire Brigade is getting called
11:50out in London.
11:51Very quick, nice and easy file in Amazon S3
11:52. Perfect.
11:54Now, Tableau also talk about this connector
11:56working in a cloud only set up.
11:58And so you've kind of seen the pace and the
12:00speed that we've had going here.
12:02Now we're going to do the exact same thing
12:03in Tableau Cloud.
12:05So give me one second to open up Tableau
12:07Cloud and we can set that up and we can
12:09almost start in exactly the same place with
12:11a new connector.
12:12Let's take a look. OK, we're here in Table
12:14au Cloud through the magic of production.
12:17We're going to go straight into this. Let's
12:19go ahead, select a new workbook in Tableau
12:21Cloud.
12:23For clarity, my Tableau Cloud instance is
12:2610 AX.
12:27I think this is the Arizona Amazon instance
12:29that Tableau Cloud runs in.
12:3210 AX is Arizona. My AWS region is US East.
12:38So although this might not be in the exact
12:40same region, it will definitely be
12:43traveling inside of what they call the AWS
12:45backbone network.
12:46So rather than traversing the Internet, it
12:48's going to be traversing the special stuff
12:51that exists between Amazon data centers.
12:54So we'll go ahead to the connectors tab, we
12:56'll select Amazon S3 and we'll go ahead and
12:59hop in.
13:00And you'll notice you get the same web
13:01window.
13:02And one thing I found super interesting
13:03with all the new connectors, it's
13:05definitely clear that Tableau is building
13:07sort of unified interfaces that work both
13:09on the web and inside of the desktop setup.
13:13So it feels to me like desktop is heading
13:15towards the web more and more to the point
13:17where I think what we'll have in the end is
13:19essentially a container which contains the
13:21web experience, but allows you to go
13:23offline.
13:24So let's go ahead and put the same
13:26information we need to put in here.
13:29So let's go ahead and fill this out.
13:32So we fill this out, we're going to hit
13:35sign in.
13:36And we get the same window. So we've got
13:37this window. I don't know if it's possible
13:39to resize. It doesn't seem like that's the
13:41case.
13:42If I do a browser resize, it doesn't help
13:44either. So I wish I could sort of resize
13:46this.
13:47It feels a bit constrained, but I'll get
13:48rid of that notification. We can get it all
13:50in one screen.
13:52Now you can see the exact same setup. And I
13:53have to say, this experience is already
13:55smoother because I had a Mac version that
13:57was all very weird and smooth.
13:59Because I'm using a browser and everything
14:01supported, it just works.
14:03So dare I say the web experience is
14:04actually slightly better than the Mac
14:07experience.
14:08And so if we go ahead and hit connect, what
14:11I'm hoping to get a sense of is the speed.
14:13And it's definitely faster. It's definitely
14:15zippier.
14:16And this is sort of the power of the cloud
14:17if you borrow sort of Tableau's terminology
14:20here, because everything is basically
14:22happening in the same place.
14:24There's virtually no real transfer of
14:25anything. And all I'm really doing in the
14:27browser is just getting a window into what
14:29's happening, which requires considerably
14:31less information to sort of operate.
14:33In fact, it probably takes longer for my
14:35click to register between my mouse and the
14:37server in America than it does to process
14:40the request and do all the stuff required.
14:43So let's go ahead and bring that into the
14:45view. It's doing exactly the same thing.
14:48You'll notice that Tableau web interface
14:50has a slightly different setup.
14:52But this is kind of nice that we're able to
14:53do all of this in the browser. We'll wait
14:55for this to load. I want to see if it's
14:56actually faster. I don't get a sense of it
14:59feeling faster.
15:01So it might just be that, hey, my internet
15:03connection and the one between the data
15:05centers are actually not that different.
15:08But it's hard to believe that.
15:09So what we'll do is we go ahead and get
15:10column 12. Now, can I change the geographic
15:13role to postcode? Yes, we can.
15:17And in this, we double clicked and we built
15:19a map. Now, the difference here is, let's
15:23set that as postcode.
15:25And what's not going on here is it's not
15:28because this Tableau cloud instance is in
15:31America.
15:32I think it's defaulting the geography here
15:37to a different country. And so what I can I
15:41can't really change it.
15:43I think it thinks these are zip codes
15:45rather than postcodes. But if I go to map,
15:50I'm not really able to specify the country.
15:52And I'm not also seeing the information I
15:54need to see down here. It is on the map.
15:57We can see here that it's all detailed out,
15:59but I'm just not seeing anything that I
16:01need to.
16:02So let's go down and see if we can force it
16:04to render something, putting the count of
16:07items on size.
16:09And yeah, we don't see sort of the null
16:10indicator here. You can just see that there
16:13are.
16:14It says this. It's weird. There's a legend,
16:16but there's just nothing to select. So I
16:19don't know what I'm doing wrong.
16:21It might be me. It might be the product.
16:23But just to sort of clarify that we are
16:25doing the right thing.
16:27We're going to column 12. We're getting all
16:29the outcodes. We're going to the count.
16:32We're getting all the columns. And what we
16:34're going to do is we're going to sort this.
16:36And this is essentially what I drew in
16:38desktop. In fact, I still have that
16:40available.
16:41Yes, we do. That's basically what I drew.
16:43And the difference is here.
16:46I'm not able to see this little message
16:47that says for unknown. Get this window to
16:50pop up and just that's just not what I'm
16:52getting here.
16:53So if I go to shoot one here, you'll see
16:55that, you know, it's just not loading that
16:58screen up now.
17:00I wonder this is one of the things you have
17:02to be sort of careful with if there is a
17:05way to get it to show.
17:07And I'm just not getting it. And one thing
17:10I will triple check.
17:12And this is something you do have to check.
17:14Let me just disable the ad blocker because
17:17the Tableau cloud setup is really smart.
17:20Even though the page refreshed, it just
17:21caught the session again so I can carry on
17:23as I am.
17:24And if I whatever reason, close this
17:25browser and came back, it would actually
17:28sort of save a draft for me and it would
17:30open up this instance.
17:31I don't have to worry about losing my work.
17:34But yeah, I can't get the exact same same
17:36thing drawn.
17:37But I am using the exact same data. So, you
17:39know, that's the sort of cloud experience
17:42sometimes.
17:43There's still a little bit of work to do in
17:44terms of feature parity. Things are getting
17:46there, but they're not quite there.
17:49And some of those small things you need to
17:50kind of get you across the finish line
17:51haven't quite made it over.
17:53But the connection experience was, I think,
17:55much faster. The interface was a little bit
17:58zippier.
17:59I just shame about that final mile. Right.
18:02So, yeah, that's pretty much it for the
18:04Amazon S3 connector.
18:05It's probably worth, you know, following
18:07this connector out of beta and then maybe
18:10maybe in a few months time once it's out of
18:11beta and it's got its final sort of feature
18:14set.
18:14And we've got some instructions on how it
18:15works. We can actually try and have a look
18:17at that.
18:18But I thought I'd do this video just now
18:19just to highlight to people that it's
18:20available. You should be using it because
18:22it's in beta.
18:24I would say probably not great for
18:25production. I've had issues with Excel
18:27files, for example.
18:28It might just be my Mac and the fact that
18:30that's not supported in Mac OS Ventura. But
18:33nonetheless, you should try it. Do your own
18:35testing and come to your own conclusion.
18:37Thanks for watching and I'll see you in the
18:38next one.
18:39Thanks for watching.
18:40You
18:50[ Silence ]
Tableau | Data prep | Tool strategy