adityapatadia

We don't have a single DBA or DevOps or SRE. MongoDB is really that simple.

Almost all statements about MongoDB are wrong.

> You know exactly what your app needs to do, up-front

No one does. Mongodb still perfectly fits.

> You know exactly what your access patterns will be, up-front

This one also no one knows when they start. We successfully scaled MongoDB from a few users a day to millions of queries an hour.

> You have a known need to scale to really large sizes of data

This is exactly a great point. When data size goes to a billion rows, Postgres is tough. MongoDB just works without issue.

> You are okay giving up some level of consistency

This is said for ages about MongoDB. Today, it provides very good consistency.

> This is because this sort of database is basically a giant distributed hash map.

Putting MongoDB in category of Dynamo is a big mistake. It's NOT a giant distributed hash map.

> Arbitrary questions like "How many users signed up in the last month" can be trivially answered by writing a SQL query, perhaps on a read-replica if you are worried about running an expensive query on the same machine that is dealing with customer traffic. It's just outside the scope of this kind of database. You need to be ETL-ing your data out to handle it.

This shows the author has no idea how MongoDB aggregation works.

I don't want fresh grads to use SQL just because they learn relations (and consistency and constraints and what not). It's perfectly fine to start on MongoDB and make it the primary DB.

AVIF and JXL can be upto 30% smaller than webp. E-commerce and news media companies have ton of images and 30% is a huge savings for them.

Founder of Gumlet.com here (we serve more than 2 billion images per day).

Here are the reasons image processing on VPS is hard.

- Reliability and uptime is a huge concern. If service goes down, images can show broken - When you build it yourself, you pay for image processing, bandwidth AND developer time. image processing at multiple terabytes and half a million image processed is not trivial ( inputs can be crazy and output needs to be modified as per the requesting platform) - Images form a big part of Google web vitals so slow image processing can hurt LCP score. - Newer formats like AVIF and JXL are increasingly complex. They need way more processing power than plain old JPEG.

This seems like a really painful incident. Do give a try to Gumlet.com (I am the founder) and you will get a way easier and better stream experience.

I would propose another solution. Use min.io to create object storage. It has all the advantages of RAID-1 and also gives S3-compatible API.

This is only profitable if the data volume is huge though. For 5-10 TB, one is always best served by cloud storage providers like S3, R2 or Wasabi.

S3 costs seem low at start but their data transfer is very expensive at volume (~10 cent per GB)

It's almost always best to use some sort of image processing service (like Gumlet.com) to optimise images and then pay lower per GB bandwidth costs.

Yes, I figured out the Starbucks (or any other fast food chain) thing after staying there for a while. It's however really a sad scene where people deny visiting washrooms especially when you need it the most.

Plus it paints a wrong picture of a country because it's technically people asking you a few bucks for a basic necessity and the government has failed.

This is in stark contrast with developed countries like Singapore (where I visit a lot). Singapore makes sure a public toilet is available within a few hundred meters and boy are those clean!

I faced when I visited New York. I had to search Google Maps and it showed a public toilet in Rockefeller Center. I walked a few miles to reach but it was a real pain.

- Added a bunch of functions like volume adjustments. - New build system makes it easy to build. - Code structure is completely changed and tests are updated (it's still API compatible)

The original project was dead since 7+ years and we have revived it with new features, modern technologies and brand new build system.

- There are all sorts of toxins on the ships. Crude oil which drives the ship is one component but the government asks to get rid of it first as long as the ship arrives. Next big one is asbestos which is abundant and not removed by the government. It lies around. I am sure there would be more of them

- Yeah olives in brine in a sealed food tin. We did not eat a lot of fish back then but there were food tins of all sorts on those ships. I personally only picked olives.

- A visit could be arranged if you knew anyone who managed the shipbreaking. If the ship is not broken yet, they show you how to climb the ship (mostly vertical steel ladders). If the breaking has started, they don't allow you to climb the ship but then you can roam around the site and inspect all the goods removed from the ship and buy it at whatever price u feel.

- Local population was not that educated. People did not take the businesses as badly as we treat them online. They feel it's a normal industry like anything else and gives employment so they are mostly fine with it. If some worker loses their life (a few do every year) the families are compensated to the tune of $1000 to $2000 and life goes on. (sad I know)

There is a ton of food on board when ships arrive. Remember, the ships have crews and they are working when they arrive. In fact you could find all sorts of food items and kitchen equipments that you would find in a professional kitchen (a lot of restaurants in our area just got the kitchen equipment for cheap).

I lived near Alang (India) and visited the shipbreaking operations a lot. For me as a kid, the best part was these ships had olives (in brine) and we could only get them from Alang.

Happy to answer any questions you guys might have.

I have used iCloud as my personal cloud storage for 5 years now.

Gradually upgraded storage as I needed and now on 2TB tier. I actually never ever faced any problem on iCloud. I never used Dropbox but used Google drive as personal storage in past (and still use at work).

Am I the only one here thinking they should have just used MongoDB and be done with it?

I know over simplified approach but majority of problem would be solved.

Shameless plug here. In case someone does not want to build any of this and still stream video, you can check https://www.gumlet.com

Do give it a try! In our tests we found that sizes are smaller and encoding of HEIF takes very very long time. Remember, we provide bunch of other things like image resize and crop etc so we always end up encoding image again. Encoding in HEIF is very slow.

Yes!

We also support HEIF conversion with `format=heif` parameter. It's not automatic though.

Out of total requests we have served in last 1 minute, this was percent of response was JPEG-XL. This is across all of our customer base.

It still needs `f_jxl` in parameter. Automatic support has still not launched as per this article's last paragraph: https://cloudinary.com/blog/jpeg-xl-how-it-started-how-its-g...

If you are curious, here is how JPEG-XL adoption looks like after 42 hours of iOS 17 launch: https://twitter.com/adityapatadia/status/1704407864326939043

Founder here. We are first cloud service to do it "automatically" based on client support. Most services don't have this and those who have it, don't do it automatically and requires code change.

What I don’t get is why Tailscale is on top news for adding extra auth mechanism.

Seems similar to ours. We only used bigquery and it seems they are not getting those money.

We at https://www.gumlet.com use bigquery a lot, but now it seems we will have to move to click house. Can’t trust google with anything. First Kubernetes price intro, then multi region storage price increase along with inter region price introduction and now this.

I would trust AWS but not Google.

Rest in peace sir

Your actions have powered billions of dreamers and saved trillions of hours for the mankind.

We recently contacted their sales team with intention to re-work pricing. We did not mention that in first mail and got immediate response. When we mentioned we want to negotiate pricing, noone has responded in months now.

I am surprised the forum rich of tech people is not yet containerised. A simple helm chart would make deployment to AWS (or any other kubernetes) a 5 min job.