Keeping files in object storage

A delivery firm asks drivers to take a photo at every door. That is ten thousand photos a day.

The photos are kept for a year, in case a customer complains. They will not fit on a laptop, and a server’s disk would fill in a week. Where should they go?

Object storage keeps whole files. You make a bucket, then put files into it. Each file is an object, and you find it by its key.

A key looks like a path with slashes. But there are no real folders. The slashes are just part of the name. The provider’s screens show each prefix as a folder, so it is easier to read.

# A tiny pretend bucket: each key points to a file's contents.
bucket = {}

def put(key, data):
    bucket[key] = data

def list_keys(prefix):
    # Object storage lists the keys that start with a prefix.
    found = []
    for key in bucket:
        if key.startswith(prefix):
            found.append(key)
    return found

put("photos/2026/10/order-1042.jpg", "photo bytes")
put("photos/2026/09/order-0977.jpg", "photo bytes")
put("invoices/2026/10/order-1042.pdf", "invoice bytes")

print(list_keys("photos/"))
print(list_keys("photos/2026/10/"))

# Put an invoice for order 1043, then list the prefix "invoices/".

There are three common places to keep data in the cloud.

Object storage suits whole files that are written once and read many times: photos, PDFs, backups, logs. It grows as far as you need.

Block storage is a disk for one server. It suits a program that changes small parts of a file all the time.

A database suits records you search and update, such as orders.

Object storage is priced in a few parts. You pay for each gigabyte you keep, each month. You pay a little for each request, such as an upload. You often pay for data sent out to the internet.

Old files that nobody opens can move to a cheaper, slower storage class (a price level for rarely used files). Many teams set a rule that does this after 90 days.