Blog
How Amazon weaned itself off Oracle databases
Amazon's consumer business shut down 100% of its Oracle databases. Here’s how they did it so you can do it too.
Blog
It’s easy being green with the public cloud
Earth Day 2023 is upon us, and the race is on to reduce the telco industry’s carbon footprint. Achieving better sustainability goals not only helps Mother Earth, it also has an intrinsic benefit: it lowers operational costs. How, you may be wondering, can you upgrade to 5G (with its expected 160% increase in energy consumption) and provide the increasing bandwidth customers are demanding, while reducing your emissions and costs at the same time? I have an answer for you: it’s … wait for it …
The public cloud! (Did you see that one coming?)
What telco is doing about sustainability
The GSMA’s Climate Action Taskforce has a goal of net zero carbon emissions by 2050 (according to the Mobile Net Zero report), but some telcos aren’t waiting to get a jump on this goal: per ABI Research, industry leaders like Deutsche Telekom (DT), Telefónica, and Vodafone have goals to achieve net zero carbon emissions. Each has already accomplished 50% renewable energy; DT and Telefonica have already achieved 100%. Woo hoo!
Net zero is a great goal to have. That’s because, on average, more than 90% of a CSP’s operational carbon footprint is from electricity used to power its networks. Why so high? Well, most telco network systems today sit in on-premise data centers, built on legacy systems from the last century. In order to handle peak use, capacity is over-provisioned for compute needs, and under-utilized most of the time. But as my faithful readers know, you can get off the hardware provisioning treadmill and move to elastic scalability (and more!) by using the public cloud.
How does the public cloud reduce emissions and boost sustainability?
The public cloud can drastically reduce your carbon emissions because it gives you many ways to reduce, reuse, and recycle for your infrastructure and software needs. In this Accenture report from 2020, there’s a great chart that shows how much “CO2 intensity” each superior aspect of the public cloud is likely to save you.
For example:
| Item | Savings |
| Power and cooling efficiency reduces carbon emissions | 10-20% |
| Hardware efficiency reduces CO2 | 10-15% |
| Compute utilization reduces CO2 | 35-45% |
| Sustainable software engineering reduces CO2 | 5-10% |
| Cloud native app arch reduces CO2 | 5-10% |
| Carbon free energy and hardware circular economy reduces CO2 | 15-20% |
Obviously, the Big Three hyperscalers—Amazon Web Services (AWS), Microsoft Azure, and Google Cloud—are laser-focused on achieving maximum efficiency. As you would imagine, each has its own initiative and goals—I’ve linked them here to help you out (AWS, Microsoft Azure, Google Cloud). Of course, they are doing it for the environment; they’re all green-minded companies that get a lot of heat for their energy use. But like you, they also do for their bottom line. The more efficient their data centers are, the higher their profits. So, they’re making serious efforts to use less energy, and to use green energy wherever possible. They also have more options and deeper pockets than most telcos, so they can afford to make investments. They have aggressive sustainability goals, and they’re meeting or beating them. By using the public cloud, you get to stand on their efforts to achieve your sustainability goals, too. I guess it’s like going “OTT” on sustainability.
Below, I pulled some examples of ways to reduce energy consumption, how the public cloud helps you with reuse, and a cool example of a retail company that recycled its data center hardware.
Ways to Reduce, Reuse, and Recycle
Reducing your compute needs is the single biggest way to reduce your CO2 consumption, and this is an area where the public cloud SOARS. It allows you to purchase compute BY THE MINUTE and elasticity for scale up and scale down for “compute bursting.” Just think of all that unused capacity you’ve reserved for your network applications “just in case” something happens. With the public cloud and smartly written software applications, you can recoup all that CapEx spend and get off the treadmill of constantly having to refresh your infrastructure. Want to start reducing hardware in your on-premise data centers today? Start by getting rid of your zombie servers that aren’t doing any useful work. This article shares detection techniques to kill off these electricity-eating, carbon-belching monsters.
The public cloud is also more efficient because of the circular hardware economy that exists at the hyperscalers. You use (and pay for) only what you need at any given moment; if your application is designed to be super efficient, you don’t even use the SAME servers from one day to another. When you need more, you get more—on demand. When you need less, your bill goes down and someone else can use that capacity—and the servers. Let that sink in for a moment, because I think people still don’t get this about the public cloud. Contrast that idea with using your own captive hardware, where a lot of your equipment is reserved for specific applications, or where over-provisioned extra capacity sits idle between traffic spikes. The hardware at the hyperscalers is getting maxed out and refreshed much faster than your team can do it. The hyperscalers are squeezing every possible usable minute out of their machines. Your team doesn’t have a prayer of matching that utilization.
The other amazing thing about the public cloud is the custom chips that also help with reducing energy consumption. You’ve heard me bang this drum before—the AWS Graviton family of chips has been delivering serious cost and energy savings for years. For the Graviton2 release, AnandTech did a great comparison between Intel and AMD chips and validated the 40% performance improvement; Graviton3 (the current version) improved upon it more. To take advantage of this benefit, you’ll need to shift your workloads to Graviton machines, which will take at least a recompile and a test before you release them into the wild; AWS has even put together a handy Graviton migration “fast start” guide that promises to get your workloads migrated in as little as four hours. Users all over Twitter have reported that migration was easier than expected and confirmed the performance improvement independently. Sweet!
But my favorite TELCO example was the DoCoMo and NEC project where they shared the phase-one results of their proof-of-concept with AWS to test running NEC’s 5G core in a hybrid-cloud environment using end-to-end, cloud-native network architecture. Phase one was about testing energy efficiency and performance of Graviton2 processors. The trial demonstrated an average 72% reduction in power consumption against incumbent x86 processors. Wow, right? As Ishwar Parulkar of AWS recently told Silverlinings, “It’s a very, very cost-effective way of building redundancy and backup.” BOOM.
Already moving to the public cloud and have a bunch of zombie servers you want to jettison? Then do what Wayfair, the US home furnishings retailer, did after it decommissioned its data centers. It donated 2,000 of its discards to two university programs. It helped out other organizations that deeply needed computing resources and got a tax write-off in the process. So, what are you waiting for? It’s easy being green with the public cloud, and Happy Earth Day to you! ❤️🌎
Recent Posts
Get my FREE insider newsletter, delivered every two weeks, with curated content to help telco execs across the globe move to the public cloud.
"*" indicates required fields
