Imagine an e-commerce portal which sells physical goods. Some special days like Black Friday, its sales go up drastically. So, the owners should have additional hardware power to keep the systems operational. When the Black Friday ends, the site owner should have to shut down the extra hardware to save from the server operating costs.
Thanks to the advancements of the cloud technologies, servers are now capable of automatic scaling. When there is an increase in the traffic, the system adds additional processing power to itself. Then, the system scales down itself when the traffic load normalizes.
The situation is the same if we talk about IoT. As a manufacturer, you produce IoT devices and send them to the distributors. When your customers activate theirs, there will be a continuous data stream to your servers, which lasts years. Using auto-scalable IoT infrastructure saves you some budget and makes your system always available.
Besides, there is a difference in the protocols used by applications and IoT devices. Imagine, you're looking for #maker hashtag on Instagram. There will be a large data transfer to your mobile phone only for a few minutes. But an IoT device will send small amounts of data continuously. So, the server and security configuration should have configured specially for small and constant data streams. Otherwise, the firewall may block the traffic after a while, or you may encounter unexpected errors.
To sum up, IoT cloud platforms does everything for you and protects you from bad surprises. Some of them require infrastructure knowledge, some not, but when planning an IoT system, always consider using IoT clouds.