At Google Cloud, we believe that the protection of our customers’ sensitive data is paramount, and encryption is a powerful mechanism to help achieve this goal. For years, we have supported encryption in transit when our customers ingest their data to bring it to the cloud. We’ve also long supported encryption at rest, for all customer content stored in Google Cloud.

To complete the full data protection lifecycle, we can protect customer data when it’s processed through our Confidential Computing portfolio. Confidential Computing products from Google Cloud protect data in use by performing computation in a hardware isolated environment that is encrypted with keys managed by the processor and unavailable to the operator. These isolated environments help prevent unauthorized access or modification of applications and data while in use, thereby increasing the security assurances for organizations that manage sensitive and regulated data in public cloud infrastructure.

Secure isolation has always been a critical component of our cloud infrastructure; with Confidential Computing, this isolation is cryptographically reinforced. Google Cloud’s Confidential Computing products leverage security components in AMD EPYC™ processors including AMD Secure Encrypted Virtualization (SEV) technology.

Building trust in Confidential Computing through industry collaboration

Part of our mission to bring Confidential Computing technology to more cloud workloads and services is to make sure that the hardware and software used to build these technologies is continuously reviewed and tested. We evaluate different attack vectors to help ensure Google Cloud Confidential Computing environments are protected against a broad range of attacks. As part of this evaluation, we recognize that the secure use of our services and the Internet ecosystem as a whole depends on interactions with applications, hardware, software, and services that Google doesn’t own or operate.

The Google Cloud Security team, Google Project Zero, and the AMD firmware and product security teams collaborated for several months to conduct a detailed review of the technology and firmware that powers AMD Confidential Computing technology. This review covered both Secure Encrypted Virtualization (SEV) capable CPUs, and the next generation of Secure Nested Paging (SEV-SNP) capable CPUs which protect confidential VMs against the hypervisor itself. The goal of this review was to work together and analyze the firmware and technologies AMD uses to help build Google Cloud’s Confidential Computing services to further build trust in these technologies.

This in-depth review focused on the implementation of the AMD secure processor in the third generation AMD EPYC processor family delivering SEV-SNP. SNP further improves the posture of confidential computing using technology that removes the hypervisor from the trust boundary of the guest, allowing customers to treat the Cloud Service Provider as another untrusted party. The review covered several AMD secure processor components and evaluated multiple different attack vectors. The collective group reviewed the design and source code implementation of SEV, wrote custom test code, and ran hardware security tests, attempting to identify any potential vulnerabilities that could affect this environment.

Working on this review, the security teams identified and confirmed potential issues of varying severity. AMD was diligent in fixing all applicable issues and now offers updated firmware through its OEM channels. Google Cloud’s AMD-based Confidential Computing solutions now include all the mitigations implemented during the security review.

“At Google, we believe that investing in security research outside of our own platforms is a critical step in keeping organizations across the broader ecosystem safe,” said Royal Hansen, vice president of Security Engineering at Google. “At the end of the day, we all benefit from a secure ecosystem that organizations rely on for their technology needs and that is why we’re incredibly appreciative of our strong collaboration with AMD on these efforts.”

“Together, AMD and Google Cloud are continuing to advance Confidential Computing, helping enterprises to move sensitive workloads to the cloud with high levels of privacy and security, without compromising performance,” said Mark Papermaster, AMD’s executive vice president and chief technology officer. ”Continuously investing in the security of these technologies through collaboration with the industry is critical to providing customer transformation through Confidential Computing. We’re thankful to have partnered with Google Cloud and the Google Security teams to advance our security technology and help shape future Confidential Computing innovations to come.”

Reviewing trusted execution environments for security is difficult given the closed-source firmware and proprietary hardware components. This is why research and collaborations such as this are critical to improve the security of foundational components that support the broader Internet ecosystem. AMD and Google believe that transparency helps provide further assurance to customers adopting Confidential Computing, and to that end AMD is working toward a model of open source security firmware.

With the analysis now complete and the vulnerabilities addressed, the AMD and Google security teams agree that the AMD firmware which enables Confidential Computing solutions meets an elevated security bar for customers, as the firmware design updates mitigate several bug classes and offer a way to recover from vulnerabilities. More importantly, the review also found that Confidential VMs are protected against a broad range of attacks described in the review.

Google Cloud’s Confidential Computing portfolio

The Google Cloud Confidential VMs, Dataproc Confidential Compute, and Confidential GKE Nodes have enabled high levels of security and privacy to address our customers’ data protection needs without compromising usability, performance, and scale. Our mission is to make this technology ubiquitous across the cloud. Confidential VMs run on hosts with AMD EPYC processors which feature AMD Secure Encrypted Virtualization (SEV). Incorporating SEV into Confidential VMs provide benefits and features including:

Isolation: Memory encryption keys are generated by the AMD Secure Processor during VM creation and reside solely within the AMD Secure Processor. Other VM encryption keys such as for disk encryption can be generated and managed by an external key manager or in Google Cloud HSM. Both sets of these keys are not accessible by Google Cloud, offering strong isolation.

Attestation: Confidential VMs use Virtual Trusted Platform Module (vTPM) attestation. Every time a Confidential VM boots, a launch attestation report event is generated and posted to customer cloud logging, which gives administrators the opportunity to act as necessary.

Performance: Confidential Computing offers high performance for demanding computational tasks. Enabling Confidential VM has little or no impact on most workloads.

The future of Confidential Computing and secure platforms

While there are no absolutes in computer security, collaborative research efforts help uncover security vulnerabilities that can emerge in complex environments and help to prevent Confidential Computing solutions from threats today and into the future. Ultimately, this helps us increase levels of trust for customers.

We believe Confidential Computing is an industry-wide effort that is critical for securing sensitive workloads in the cloud and are grateful to AMD for their continued collaboration on this journey.

To read the full security review, visit this page.

Acknowledgments 

We thank the many Google security team members who contributed to this ongoing security collaboration and review, including James Forshaw, Jann Horn and Mark Brand.

We are grateful for the open collaboration with AMD engineers, and wish to thank David Kaplan, Richard Relph and Nathan Nadarajah for their commitment to product security. We would also like to thank AMD leadership: Ab Nacef, Prabhu Jayanna, Hugo Romero, Andrej Zdravkovic and Mark Papermaster for their support of this joint effort.

As a Google Cloud customer, your carbon footprint is already carbon neutral: Google first achieved carbon neutrality in 2007, and has been purchasing enough solar and wind energy to match 100% of its global electricity consumption since 2017. Now, Google is targeting a new sustainability goal: operating on carbon-free energy (CFE) 24/7, everywhere, by 2030. 

We want to empower you to make more sustainable decisions and progress with us towards this 24/7 carbon-free future. Earlier this year, we published the carbon characteristics of our Google Cloud regions. Later, we introduced a simple tool to help you pick a Google Cloud region, taking variables like price, latency and sustainability into account. Our next question was: what’s the best way to surface that sustainability info when you’re actually picking a region for your cloud resources? 

Starting today, we are indicating regions with the lowest carbon impact inside Cloud Console location selectors. Available today for Cloud Run and Datastream, you’ll see it roll out to more Google Cloud offerings over time:

Regions that feature the “Lowest CO2” and the leaf have a CFE% of at least 75%, or, if the information is not available for this region yet, a grid carbon intensity of maximum 200 gCO2eq/kWh. You can read more about how we calculate these metrics in our documentation.

Before releasing this feature, we ran experiments to measure its impact: Users who were exposed to the enhanced region picker were 19% more likely to select a “low carbon” region for their Cloud Run service—a significant lift. These results show that by displaying carbon information in context of when you make the decision of picking a region, we are helping you make more sustainable decisions.

By sharing and displaying carbon information of Google Cloud regions, together we’re making tangible progress towards our goal of a carbon-free future. Learn more about Carbon free energy for Google Cloud regions.

Do you want to fire up a bunch of techies? Talk about multicloud! There is no shortage of opinions. I figured we should tackle this hot topic head-on, so I recently talked to four smart folks—Corey Quinn of Duckbill Group, Armon Dadgar of Hashicorp, Tammy Bryant Butow of Gremlin, and James Watters of VMware—about what multicloud is all about, key considerations, and why you should (or shouldn’t!) do it.

Five important insights came out of these discussions. If you’re on a multicloud journey or considering one, keep reading.

Do: Choose to do multicloud for the right reasons

Don’t do multicloud because Gartner says so, implores Corey Quinn. Before embarking on a multicloud, define a “why” focused on business value journey, says Armon Dadger. For example, you might want to use services from each public cloud because of their differentiated services, according to Tammy Bryant Butow. Armon also calls out regulatory reasons, existing business relationships, and accommodating mergers and acquisitions. On the topic of M&A, Corey points out that if you acquire a company that uses another cloud, it’s usually expensive and difficult to consolidate. It can be smarter to stay put.

Don’t: Over-engineer for workload or data portability

Thinking that you’ll build a system that moves seamlessly among the various cloud providers? Hold up, says our group of experts. Armon points out that aspects of your toolchain or architecture may be multicloud—think of some of your workflows or global network routing—but that shifting workloads or data is far from simple. Corey says that trying to engineer for “write once, run anywhere” can slow you down, and ignores the inherent uniqueness that’s part of each platform. Specifically, Corey calls out the per-cloud stickiness of identity management, security features, and even network functionality. And data gravity is still a thing, says James, that causes some to dismiss multicloud outright.

If you’re using multiple public clouds, you take advantage of the distinct value each offers, Armon says. Use native cloud services where possible so that you see the benefits from useful innovations, built-in resilience, and baked-in best practices. The value from that cloud-infused workload may outweigh the benefits of seamless portability.

Do: Recognize different stakeholder interests and needs

James smartly points out that many multicloud debates happen because people are arguing from different perspectives. Context matters. If you’re an infrastructure engineer who invests heavily in a given cloud’s identity and access management model, multicloud looks tricky. Or if you’re a data engineer with petabytes of data homed in a particular cloud, multicloud may look unrealistic. James highlights that many developers default to multicloud because their local tools—where all the work happens—are multicloud. A developer’s IDE and preferred code framework(s) aren’t tied to any given cloud. Be aware that groups within your organization will come at multicloud from distinct directions. And this may impact your approach!

Don’t: Go it alone

Corey talks about the importance of asking others what worked, and what didn’t. Tammy offers her best practices around sharing results from experiments. It’s about sharing knowledge and tapping into it for community benefit. Others have probably tried what you’re trying, and can help you avoid common pitfalls. If you’ve just made an architectural choice that didn’t work out, share it, and help others avoid the pain. 

Read research from analysts, go to conferences or watch videos to observe case studies, and join online communities that offer a safe place to share mistakes and learn from others.

Do: Experiment first using techniques like multi-region deployments

If you think you can operate systems across clouds, how about you first try doing it across regions in a specific cloud, suggests Corey. Getting a system to properly work across cloud regions isn’t trivial, he says, and that experience can help you uncover where you have architectural or operational constraints that will be even worse across cloud providers.

This is great guidance if your multicloud aspirations involve using multiple clouds to power one application—versus the more standard definition of multicloud where you use different clouds for different applications—but can also surface issues in your support process or toolchain that fail when faced with distributed systems. Start with muti-region deployments and chaos engineering experiments before aggressively jumping into multicloud architectures.

The Google Cloud take

Do the things above. It’s great advice. I’ll add three more things that we’ve learned from our customers.

1. Don’t fear multicloud. You’re already doing it. You don’t single-source everything. As Corey mentioned, you probably already have one cloud for productivity tools, another for source code, another for cloud infrastructure. You’ll use software and application services from a mix of providers for a single app. You have that experience in your team and have been doing that for decades. What people do rightly worry about is using more than one infrastructure service beneath an application, as that can introduce latency, security, and logistical hurdles. Make sure you know which model your team is considering.
2. Embrace the right foundational components, including Kubernetes. Will everything run on Kubernetes? Of course not. Don’t try to do that. But it also represents the closest thing we have to a multicloud API. Companies are using Kubernetes to stripe a consistent experience across clouds. And this isn’t just to orchestrate containers, but also to manage infrastructure and cloud-native services. Also, consider where you need other fundamental consistency across clouds, including areas like provisioning and identity federation.
3. Use Google Cloud as your anchor. Here’s a fundamental question you have to decide for yourself: Are you going to bring your on-premises technology and practices to the cloud, or bring cloud technology and practices on-prem? We sincerely believe in the latter. Anchor to where you’re trying to get to. We offer Anthos as a way to build and run distributed Kubernetes fleets in Google Cloud and across clouds. By using a cloud-based backplane instead of an on-prem one, you’re offloading toil, leveraging managed services for scale and security, and introducing modern practices to the rest of your team.

We learned a lot about multicloud through these discussions, and it seems like others did too. That’s why we’re going to do a second round of interviews with a new crop of experts so that we can keep digging deeper into this topic. Stay tuned!