Economic uncertainty has meant trimming personnel, spending freezes, and marketing cutbacks. Yet in the age of Big Data, perhaps one of the most obvious places to streamline spending went unnoticed: the significant cost of data egress.
According to IDC, the Global DataSphere reached 64 zettabytes in 2020. The same report adds that the amount of data created in the last three years is more than the amount of data created in the last 30 years. All that data lives somewhere, and for many organizations, it lives in the cloud — especially cloud-based object storage offered by providers like Amazon S3, Google Cloud Storage, and Azure Blob Storage.
For many organizations, eliminating the costs of reading stored data could satisfy necessary budget cuts and even enable business growth.
Data used to live on company-controlled servers in company-owned facilities, but today, the cloud, and in particular cloud object storage, has made data storage practically unlimited. Organizations save everything, because they can, from user data to logs to extensive data backups.
So, with the exception of organizations that face extremely strict data regulations or operate air-gapped networks, data is primarily stored remotely in the cloud – by a third party.
The cost to operate on-premise data storage includes maintaining server racks, from electricity to hardware updates, and personnel. And, retrieving stored data is as simple as accessing the server where it was saved. Forward-looking, as an organization anticipates future growth, they buy more hardware to scale.
Cloud, on the other hand, scales on demand, both up and down as the business requires – making it much more appealing. And, when data is stored by a cloud service provider, data storage becomes a service. This shift from CapEx to OpEx has created an opportunity for providers to take advantage of and charge customers exorbitant fees to access their data. These data egress fees have grown to levels completely disproportionate to the costs incurred by the cloud providers for bandwidth.
Essentially, object storage providers are charging high data egress fees because they can.
Let’s look at an example of a business that stores 1,000 TB of data in S3 and reads one-fifth or 200TB of that data per month.
The cost structure for Amazon S3 is as follows:
Data Storage Costs (per Month per GB) | Data Transfer Costs (per Month per GB) |
---|---|
First 50 TB - $0.023 | First 10 TB - $0.09 |
Next 450 TB - $0.022 | Next 40 TB - $0.085 |
Next 500 TB - $0.021 | Next 50 TB - $0.023 |
Using the cost structure above, the total cost for storing this customer’s data is $21,550, then reading one-fifth of that data is another $13,800, for a grand total of $35,350 per month.
Over a year, the monthly data transfer cost alone adds up to $165,600 — over five years, $828,000; over 10 years, $1,656,000.
This small example demonstrates the opportunity cost of storing data with providers who operate on this model.
Of course, sending data over a network is not free. Providers must pay for transit bandwidth — the right to send data via the networks they peer with. But typically, bandwidth is paid for as a monthly fee, not based on a pay-for-what-you-use model. That monthly fee is usually based on the number of Mbps that the network uses at its peak capacity, not the total amount of data transferred through the network.
To simplify: storage providers pay for capacity, not volume. But, they charge their customers for volume, not capacity. Perhaps unsurprisingly, their capacity-based costs are almost exponentially lower than the volume-based costs charged to their customers.
This is especially the case if — as is often true — these bandwidth transit fees are vastly reduced, or waived altogether. And in fact, the majority of hosting providers in the industry either substantially discount or entirely waive egress fees when sending traffic from their network to a peer.
And if the hosting provider is not paying for the bandwidth costs, those savings should be passed on to the customers as well. Under this model, Cloudflare offers a zero-egress cloud object storage service with Cloudflare R2. Customers are charged only for the storage itself, not for reading the data. R2 is also S3 compatible, eliminating vendor lock-in and making data portable.
As organizations look to trim costs, egress fees should come under heavy scrutiny. Cloudflare agrees — and eliminates egress fees altogether.
This article is part of a series on the latest trends and topics impacting today’s technology decision-makers.
Learn how Cloudflare R2 can help impact your business bottom line in the Say Goodbye to Egress Fees eBook.
Get the eBook!
After reading this article you will be able to understand:
The growth of cloud object storage
The accompanying cost of data egress
Why egress fees are not necessary
A model for zero-egress object storage