Scalability isn't limited to applications in production. No matter what you do, you want to ensure that it is a scalable approach. Short of it, the exact same activity is going to take a lot longer. Besides, what's interesting is that making even tiny changes to your approach tends to go a long way in enhancing overall productivity.
In this podcast, I'll share the approach we took to port hundreds of videos from one platform to another, and why we were able to accomplish that in shy of 10 hours.
Transcript
Krish Palaniappan (00:00.854)
Hey there, in this podcast, I'm gonna talk about work that Varun and I did the last couple of days. We spent about, I wanna say like, it took about 10 hours on this activity. And I'm gonna speak to it and just tell you what it was all about and what some learnings might be from what we actually did, right? So we published, we ported over our podcast, technical podcast from a provider. We didn't migrate because we have it in multiple providers, but.
to Spotify, right? From the source to Spotify. Because Spotify does support video podcasting, but the way it does it slightly different. In other words, you couldn't use other providers necessarily, distributors, so to speak, to publish like, you know, podcast platforms to publish to Spotify if it's a video podcast. You can do it just as easily for very easily for audio podcasts, not so much for video.
At least it's not supported by a lot of the other providers, right? However, the integration works. So that's the gist of it. So what we needed to do, because you know, our podcast has been on Apple for a while. And if you haven't checked it out, you should definitely check it out. Go to like apple.snowpile.com and it'll redirect you to the Apple podcast. But you know, the Apple podcast is a lot more popular or a lot more popular in the US.
I want to say I think their subscription is like 25, 28 million, something like that. But you know, in the rest of the world, it's not the case, right? It's not so popular. Plus I've had folks tell me that they've had streaming issues, listening to Apple podcasts from other parts of the world, like for India, for instance. So we decided to also support our podcast on Spotify. I mean, there's no reason not to. They have like a hundred million subscribers or something of that nature, right? They're huge, as we all know.
So we had to do this. So we tried to look for options to ideally, if there were no issues when it came to integration, we would have had to do nothing, right? We would have just been able to publish it through our existing podcast platform, the host, podcast host or whatever the right terminology is to Spotify, just like we do to Apple podcasts and some other providers, right? But that wasn't the case, like I mentioned earlier. So we...
Krish Palaniappan (02:18.814)
long story short, we had, we understood, we realized that we had to do this part manually. We couldn't do it any other way. At least we couldn't think of it at that point. And I don't know if there's another way to support this. If there was, do let me know. That'll be awesome to know. Anyway, so coming back to what I was trying to say, we have about 260 some podcasts on Apple and the length of each of those podcasts typically tends to mean it varies from anywhere between maybe eight to 10 minutes on an average.
If you drew a median, the median would be like about nine to 10 minutes, but it starts a little bit lower. Most of them are on the eight to 10 minute mark. And then we have some that are 20, 30 minutes and perhaps even longer lengthier. So that speaks to the size of those files. And these are podcasts typically recorded on an iPhone or on my studio. So they are pretty big files. This is what needed to happen. Let's say we have 200 or 250 podcasts. You need to take each of them, the video.
the thumbnail, if there is one for that particular podcast. Apple doesn't support it, but luckily Spotify does. So what that also meant was we did not have thumbnails, specific thumbnails for each of those podcasts. So we have to create, this is the work that needed to be done. For each and every one of those 250 or whatever number of videos, the podcast, sorry, we needed the file, the video file. So we need to download that file from our existing podcasting platform.
whatever time it takes to download those files, you know, some of them are like 100 megs. There are others that are like up to like three to four gigs a file, right? So you can imagine the time it could possibly take for those downloads, regardless of the internet speeds, right? So we have to download the file, then we need the title, we need to copy it, we need the description, we need the thumbnail, there is one, but like I said, there wasn't one, so we had to create a new thumbnail for each of those podcasts. And I think that's the least.
Those are four attributes and Spotify does ask you whether it needs to be a trailer, whether it has explicit content and a few other questions like toggle radio buttons, right? So we need to make those selections. So that's the time it would typically take. Those are the list of things that you need done for a single podcast to be ported over. Now you have to do this 250 or 270 times. And while we did that, we also did a bit of triaging if you will, right? Say, you know, I started doing that. We started doing the podcast like a couple of years ago.
Krish Palaniappan (04:45.93)
So the first one was nearly not as good as the most recent one, which isn't the best either, right? There's always room for improvement, but relatively the first one was much poorer, right? In quality, in audio, maybe in video. It's not terrible, but it's not great. So we made a decision on whether or not we're going to port over each and every one of those podcasts. If it met our criteria, if it was like a seven or eight in our minds, we decided to port that over, otherwise not.
So we had to listen to it or watch it for a few seconds as well. So that's part of this equation. Now watching list, watching it for like eight to 10 seconds, making a decision, downloading the file, copying the title, copying the description next, uh, and then making the other selections on Spotify that had to be repeated 250 or 270 times. Right. Now, if you do the math, you know, depending on how you go about doing this, uh, this could be several days worth of work, right? Could be a couple of weeks with the work.
or it could be many days worth of work. But we were like, we are always rushed to doing whatever it is that we're doing. We give ourselves a lot less at time than is needed. And we try to push ourselves even within that tiny window that we've allocated to ourselves, right? So now coming to the crux of this podcast, right? How do you do this? Now, if you did one or two and it took like five minutes and you're gonna be like, okay, if you do the math, if taking care of each one of them took about five minutes.
200 let's go with an even number right would take a thousand minutes correct a thousand minutes would be 15 hours right 15 or 16 hours that's if you never lost concentration if you're Mechanical and you kept doing it one after the other and some of this could take possibly even longer than five minutes give or take But we don't want to spend like two days or three days on it And if you broke this down it would have taken a lot more days So here is what we did. I mean some of it could have held it if it were visual
But here's the podcast, so at least this screen, just if I scroll here, you can see the number of these podcasts, right? So I can keep scrolling, and I'm scrolling at very quick pace. I can keep scrolling, I can keep scrolling, I can keep scrolling, and you can see it goes on and on and on, right? So also we took the opportunity to rename a little bit, modify the descriptions a tiny bit here and there. So how did we go about doing this? Now the-
Krish Palaniappan (07:04.058)
as lame as it might sound or as silly as this next statement might sound, trust me, it makes all the difference, most of the difference. How you position the windows on a desktop, how many windows you have and what size those windows are makes a ton of difference. And let me give you an example. You're going to have to do a bit of visualization or actually I can draw something here, right? It's going to be great. I draw terribly, but let's say this, assume that this whole thing is a desktop. I want to draw this.
and then you want to position the windows. Now what we did was we put our existing platforms window here with a single tab and the challenge here is these platforms, whether it's Spotify or other platforms, they don't support, you know, pagination very well. So every time you do something, it puts you back on page one and then you have to go back and back and forth all the time. So a lot of their functionality or lack of functionality added to, you know, a bit of pain here as well, right?
So every time we did something, you had to right click and open it on a new tab. Because if you clicked into a podcast and try to do something and port it over, and then you said, okay, and you close this, and you go back, it'll actually take you to page one. And we have like, we had like 15 pages, right? So we had to remember the page we were on and the number of podcasts that were on that page and the specific podcast we were acting upon on any given page on that platform. And they are not numbered on that platform. So a lot of the UI issues.
of those platforms including Spotify caused a lot more headache as well. Anyway, so this green box is the existing platform. I'm going to draw another window here. These are like Chrome sessions, right? These happen to be Chrome sessions. The one at the bottom was Spotify. On Spotify, we decided to have multiple tabs. We started with one, but the problem was after you uploaded something on Spotify, it took a while, not a while, a decent amount of time, a reasonable amount of time. But it...
was unreasonable in the context of the number of things we were trying to do, for it to actually process the uploaded video and publish it and create a preview. So we had to wait and we didn't wanna wait. So we created multiple windows, right? And then we created multiple windows, we had to remember which page we were on here, like page one, page two, page three, and within those pages, which item were we acting upon? Was it this item or was it this item, right? So now we download it.
Krish Palaniappan (09:26.186)
What we did was we downloaded all of the videos first because we knew that was gonna take time. So we went through page after page, right? Page one, page two, page three, all the way to page 15. Downloaded all the videos, got that taken care of. And then we said, okay, let's deal with the title because we have to copy the buffer, right? If you created a document for each of these and then had someone ported over, it'll take like many, many weeks to do this. So we had to copy it and paste it right away. So we would take...
the title from here and then paste it on the podcast here. Then we would actually take the description and paste it here, right? So we only have one item in the buffer. And then we make all the selections that were needed on Spotify. And then we would wait for it to be published because it had to process it before we could publish it. So while it was processing it, we would actually not wait on it, actually switch gears to a second tab. We'll act on the second tab, then the third tab.
So at the back of our minds, we had to remember which page we were on this existing, on this platform, the source platform, where we were on, on the target and make sure that we didn't do these mistakes. Sure, we did some mistakes here and there, but since we did it together and weren't very good at catching these issues and anomalies, he was able to call out for things where there was a mismatch. So we were able to pair on this activity and get all of this done. I mean, I'm not kidding, right? All of this done.
to the extent that you see here creating the thumbnails, uploading it and adding the title and description within I want to say like it felt like forever but it probably took like between eight to 10 hours right maybe like 11 give or take roundabouts there. What I was trying to share here is this is not programming activity, this is not development but there is still a method of the madness how you go about solving problems efficiently.
trying to push your limits. Like for instance, we started doing just a single tab and then you know, you get impatient and you want to push yourself a bit more and say, I want to get this done sooner. So I think Steve Jobs, I'm not Steve Jobs, sorry, I'm quoting Steve Jobs like because I quoted him in the last podcast, I believe. I think it was like Bill Gates, if I'm not wrong. And again, internet attributes a lot of these codes to the wrong people, but you know, forgetting who said it. I think the code itself means something. It says, you know, I believe
Krish Palaniappan (11:48.122)
it was him or they say that it was him who said that it's better to hire lazy people because they're going to find better ways to solve problems. It doesn't sound very positive, but I think there is some sense to it, right? There's some value in that statement because if we were not lazy, we would have found the old fashioned way to solving this problem. It could have taken many days and be completely acceptable because it's a large number of podcasts that needed to be ported over. But if you did that, then you'll be able to do
that much lesser on the next thing that you need done. And in your startup, you have like a million things that you need done. Prospecting being one of them, which we have to find optimal ways to do, but that's for another podcast.
#snowpal aws.snowpal.com learn.snowpal.com
(Audio) Porting over 250 podcasts from one platform to another