Cloudflare’s recent announcement for their S3 competitor, cleverly named R2, has been a major point of discussion in the infrastructure community. While there have been many opinions on what this announcement means for the infrastructure market, there has been less conversation about what this means for customers — and how to assess the opportunity cost of moving a workload from S3 to R2. The day after the announcement, we immediately had customers reaching out with interest and two very specific questions:
- I saw Cloudflare’s announcement about their S3 competitor, R2. I’m interested in understanding the value in migrating some workloads and data from AWS to Cloudflare. Can you help me understand the opportunity cost of making the switch?
- I know that this isn’t a core business for Cloudflare. Before I invest in making this transition, I want to be reasonably confident that this is a business that Cloudflare is taking seriously and will continue to invest in. I don’t want to transition to a service that will be mothballed in a year. Can you help me build confidence in the market opportunity for Cloudflare and that they are incentivized to take it seriously?
We will elaborate on the first question later, but our customers’ second question came as a surprise to us. One of the advantages of having thousands of users utilizing Vantage to help manage and understand their AWS costs is that we have access to a unique dataset, which we use to extrapolate the size of the market opportunity for Cloudflare. Assuming the market opportunity is large enough for Cloudflare, we can help our customer build trust that Cloudflare is properly incentivized to invest their resources into building a service that will endure as long as Amazon S3 has.
After crunching the numbers, it became clear that there is $1B in annual revenue up for grabs.
Below, we’ll detail how we got that figure.
25,000 S3 Buckets: What the Data Shows
In order to do a market opportunity estimation, we need to analyze some data. The end goal is to get an idea of how much AWS earns annually for its S3 object storage service and the percentage that Cloudflare’s R2 has a reasonable shot at capturing.
In order to answer this, we decided to sample a random set of 25,000 S3 Buckets from users who have connected AWS accounts to Vantage. This will be our core dataset to estimate how much S3 makes up of total AWS costs in aggregate across this sample set. From there, we can extrapolate the annual revenue that AWS makes from S3.
Our methodology is below:
- Randomly select 25,000 S3 buckets
- Of those 25,000 S3 buckets, pull the S3 & total costs for each associated account in September
- Sum the total S3 storage costs across all of the accounts
- Sum the total AWS costs of those customers for the associated accounts
With this data in hand, we found that S3 storage accounts for ~3.19% of AWS costs in September.
A Billion Dollar Storage Opportunity
Based on the cost of S3 across our customer base, we can extrapolate the amount of revenue that AWS makes on storage. Amazon breaks out AWS revenue as part of their quarterly filings. According to the latest quarterly filing, AWS is at a $59 billion dollar annualized run-rate.
If S3 storage costs are 3.19% of total AWS costs in the sample, then we can estimate that S3 storage makes up $1.26 billion in revenue for AWS (3.19% of $59B).
Before we call it a day, we should remember that not all Amazon S3 storage costs are at-risk of being displaced by Cloudflare R2. As primer: Amazon S3 offers a variety of object storage types — each with their own performance and pricing. We wrote in detail about S3 storage types in a previous blog post. To truly get a sense of the potential cost savings impact of moving data storage to R2 versus S3, we must first examine the breakdown of storage classes — as not all storage classes are candidates for a shift to R2. For example, it isn’t expected that objects stored in Amazon S3 Glacier will ever move to Cloudflare R2 because the price just isn’t competitive enough.
There are 3 specific storage costs that Cloudflare R2 boasts a better price than Amazon’s S3: Standard Storage, Intelligent Tiering - Frequent Access, and Reduced Redundancy. These 3 storage tiers are the only S3 tiers for which it would make sense to evaluate Cloudflare R2 against. Fortunately, Standard Storage makes up the majority of all S3 costs. If you add up all 3 tiers… a whopping ~68% of all costs are up for grabs by Cloudflare R2.
The table below breaks out all of the S3 storage tiers in detail and shows the proportion that each tier contributes to the estimated $1.26 billion in revenue that AWS earns from S3.
A breakdown of the rough run-rate for S3 storage classes and their respective opportunities for migrating to Cloudflare R2.
For storage alone, the potential opportunity cost based on AWS Q2 financials is roughly $837M — but with Q4 typically being the largest quarter for AWS, coupled with the fact that AWS has been reliably delivering 30% year-over-year growth, it’s expected that AWS’s full year run-rate would push this opportunity to roughly $1B.
Going back to the original question… is this a large enough opportunity for Cloudflare to invest considerable resources into the offering? Looking at Cloudflare’s revenue for the first half of the year, we can estimate that their annual run rate is closing in on $600M. Given that the potential opportunity for R2 from AWS alone is nearly double their current run rate, it feels safe to say that this is a huge opportunity for Cloudflare and warrants substantial investment.
What Does This Mean For Me?
Now that we’ve addressed the concern about whether this is an opportunity that Cloudflare is serious about, let’s return to the first question that our customers raised.
Can you help me understand the opportunity cost of making the switch?
Our answer to this question is an enthusiastic yes - but first we need Cloudflare R2 to be generally available.
In our market analysis above, we determined that Amazon S3 earns roughly $1.25B in revenue for storing S3 data and Cloudflare’s R2 storage comes with a ~35% discount from the list price of S3 storage. If 100% of current S3 workloads moved from S3 to R2, customers would collectively save $425M annually. Not an insubstantial amount of savings. Obviously your organization’s exact savings will depend on the profile of your S3 Storage. Vantage is currently working on showing cross-provider cost savings personalized to your AWS account’s usage on a per-S3 bucket basis for when R2 launches.
But before we declare this a win, let’s dig into the one-time tax that customers will have to shoulder in order to realize these savings.
Before Migrating, Consider The Cost of Leaving Hotel California
Cloudflare’s R2 service is offering no egress fees which makes the cost savings calculation simple. The complication comes from the fact that AWS does charge egress fees, so any customer who wants to move data from S3 to R2 will incur a one-time fee. We’ll show how this fee comes close to the savings you’ll get by changing providers.
Cloudflare has the following quote which perfectly summarizes a S3 transit story:
Another oddity of AWS’s pricing is that they charge for data transferred out of their network but not for data transferred into their network. If the only time you’ve paid for bandwidth is with your residential Internet connection, then this may make some sense. Because of some technical limitations of the cable network, download bandwidth is typically higher than upload bandwidth on cable modem connections. But that’s not how wholesale bandwidth is bought or sold.
Taking the extrapolated annualized revenue run-rate numbers from the previous table and translating them into equivalent storage amounts allows us to calculate the cost considerations of migrating away from AWS.
The table below takes the annualized run-rates from above and breaks them down to equivalent storage by dividing them by per-month storage rates for each respective S3 storage class. Once the storage is calculated, we calculated the cost of migrating out from AWS (via CloudFront) to Cloudflare using North America per-GB rates. Here are the total costs of leaving AWS for each storage class:
.svg)
The cost of transferring data out via CloudFront using North America per-GB rates.
If every current AWS user left right now, the total cost of migrating data out of AWS would be ~$1.3B in egress fees in total. Just to leave. If we’re only considering the Cloudflare R2 price-competitive storage equivalents (Standard, Intelligent Tiering Frequent Access and Reduced Redundancy), it would be roughly $400M to migrate.
The point being: when the time comes, do not ignore the cost of migration when you look to move from AWS to R2 and calculate the transfer rates accordingly.
The $NET Effect
We wrote this post to demonstrate, with actual customer usage data, the potential cost savings for both Cloudflare and AWS customers in light of R2’s arrival. While the data above is biased toward a random sampling of Vantage users, we made sure to choose a sample size large enough to model and extrapolate these numbers in order to shed light on both the market opportunity for Cloudflare and the cost implications for making a shift.
As with anything infrastructure related, you should carefully analyze your company’s use of S3 and consider the potential benefits of migrating to another provider like Cloudflare. As mentioned previously, Vantage will soon be adding a programmatic cost recommendation to show savings of moving from Amazon S3 to Cloudflare R2, in addition to cost breakdowns provided for your current storage patterns. If you or your company is interested in utilizing this feature, you can sign up for free here: https://console.vantage.sh/signup
