If you aren’t already using the cloud, you’ve probably considered it. The benefits are well-documented, and have been for years. But as more and more developers turn to the cloud to help modernize existing applications and build innovative offerings, organizations are only now coming to grips with the implications of these new cost structures. We’ve talked about what cache hit rate means for performance, but one of the topics not well covered is the cost of a cache miss. A cache miss results in a callback to your origin — or in this case, a public cloud provider like Microsoft Azure. The resulting outbound data transfer to Fastly is metered and billed by Azure, which causes you to pay twice to serve the content to your end user: once to Azure to get your data off the public cloud, and once to Fastly to deliver it to your end user.
Then there’s also the performance cost (or latency) of serving an audience that reaches your content or application from outside of your cloud provider region. We’ve heard from developers and businesses alike that these two challenges can be major pain points.
That’s why we’re very excited to announce that we’ve expanded our partnership with Microsoft to address both of these problems. Leveraging Microsoft’s new ExpressRoute Direct Local product, we’re provisioning select Fastly points of presence (POPs) with direct, local connectivity to the respective Azure region. As part of Fastly purchasing this product, Microsoft includes outbound data transfer fees for all data over these connections. Which means that starting today, when your traffic from Azure to Fastly transfers over these high-availability circuits, Fastly includes those benefits in our standard pricing to our joint customers. And that can result in some serious cost savings for you.
Wait…what does outbound data transfer even mean?
Cloud providers use the term data transfer when referring to data moving from the storage or compute service to other cloud services or to the internet. Companies typically incur outbound data transfer fees when edge cloud platforms like Fastly fetch content from a cloud-based origin, bring that data out of the public cloud network, move it to an edge cloud POP, and eventually deliver it to an end user.
Outbound data transfer fees are usually 4-5x more expensive than storage fees on a per GB basis, meaning it is much more expensive to serve a video or an HTML asset out of the public cloud than it is to store that same data inside it. And these charges can vary month to month as your cache hit ratios and content libraries change over time, making it a variable (and potentially very expensive) component of your monthly cloud bill.
We started addressing this problem over three years ago, when we were the first partner to work with Google Cloud Platform on their CDN Interconnect program, which offers joint customers a 50% discount on outbound data transfer fees. Since then, we’ve also added products like Media Shield, and worked directly with customers to optimize how they leverage platform features like request collapsing and origin shield — all of which effectively reduce a customer’s cloud costs during a cache miss callback to the public cloud.
With outbound data transfer included in your Fastly services over our Azure ExpressRoute connections, we’re taking these concepts to the next level by unlocking the benefits of this private interconnection: your inbound and outbound data transfer goes over the Azure ExpressRoute connections Fastly has provisioned, and instead of it being metered on your Azure bill, Fastly carries the traffic instead. So regardless of how your cache hit ratios change, your CFO can sleep soundly, and you can continue delivering great web experiences.
How it works
We’ve worked closely with Microsoft Azure to build a deep integration between our platforms that will directly benefit our joint customers. We started with Azure’s new ExpressRoute Direct Local connections in strategic regions that connect our networks over optimized connections. Next, we carefully mapped Fastly POPs to Azure service regions to handle origin offload. Finally, we configured routing policies to send traffic over these links when applications send or request data between Azure’s public cloud and Fastly’s edge cloud.
Moving forward, when you need to transfer data from Azure to Fastly (or vice versa), the data can be transferred over these direct connections instead of being subjected to standard request routing methods over the public internet. This drastically reduces latency, enabling you to deliver better experiences to your end users.
And since the data transfer from Azure compute and storage in select regions to Fastly goes over the ExpressRoute connections Fastly has bought, we bear the cost of that data transfer. This means that customers can also reduce their cloud costs — because no one should have to choose between lowering costs and delivering exceptional user experiences. Which is why we’ve worked with Azure to remove the barrier, and help you do both with minimal effort on your part.
If you’re already using Fastly and Microsoft Azure in the selected regions, then you already benefit from this integration at no additional cost. (If you’re not, then you can test up to $50 traffic for free on Fastly, or sign up for 12 months of free services on Azure). You’ll just need to make sure shielding is turned on in the locations where we’ve integrated with Azure’s ExpressRoute Direct Local and that you choose the right region when setting up Azure. More information about how to set this up can be found in our documentation.
Fastly + Azure = better together
Helping you make your delivery more predictable and cost efficient is just the newest aspect of our ongoing work with Microsoft. We also recently announced integrations that empower near-real-time data analytics. Because our edge cloud platform was built by developers, for developers, we understand the need for real-time visibility into your services so you can find and fix issues fast, which is why we continue to invest in these features.
By combining cost-efficient data transfer with improved visibility into your cloud services, we’re giving you the tools you need to adopt the cloud on your terms and timelines alongside Microsoft. And since we help bring predictability and control to your cloud bill, your costs look a lot more like what you’d expect if you had your origin in a traditional datacenter. That removes potential objections your team could face when making the internal business case for a cloud-migration project.
We know there’s more to consider in cloud migration though — namely, the fear that your service will go down during the transition. We can help mitigate this by acting as another layer of protection and control for your user experience. With intelligent traffic routing capabilities, we can keep your service online while you lift and shift your infrastructure – minimizing the risk associated with any operational hiccups that may occur during the migration process.
We’ve been working on solving the challenge of cost-effective data transfer from the public cloud based on feedback from customers, so we’re thrilled to roll this out to all Fastly and Azure users today. As always, we work to make these changes based on what’s working for you and what your pain points are — so continue to share your feedback in the community forums, and let us know how we can help your team and your business.
You may also like:
ButterCMS on building a globally performant + reliable API
Without realizing it, many of us are building single points of failure into our stack. Fastly customer ButterCMS requires near-100% uptime for their customers, but after multiple outages nearly crippled their business, they became “obsessed”…
DDoS attacks: how to protect + mitigate
In part one of this series, we took a look at the evolving DDoS landscape, offering a sense of what’s out there in terms of attack size and type to help better inform decisions when…
The evolving DDoS landscape
As an edge cloud platform, Fastly is in a unique position to monitor DDoS attack patterns and trends as they evolve. In this post, Jose Nazario, Sr. Director of Security Research, and Ryan Landry, Director…