AWS Security Profiles: Avni Rambhia, Senior Product Manager, CloudHSM
In the weeks leading up to re:Invent 2019, we’ll share conversations we’e had with people at AWS who will be presenting at the event so you can learn more about them and some of the interesting work that they’re doing.
How long have you been at AWS, and what do you do enjoy most in your current role?
It’s been two and a half years already! Time has flown. I’m the product manager for AWS CloudHSM. As with most product managers at AWS, I’m the CEO of my product. I spend a lot of my time talking to customers who are looking to use CloudHSM, to understand the problems they are looking to solve. My goal is to make sure they are looking at their problems correctly. Often, my role as a product manager is to coach. I ask a lot of why’s. I learned this approach after I came to AWS—before that I had the more traditional product management approach of listening to customers to take requirements, prioritize them, do the marketing, all of that. This notion of deeply understanding what customers are trying to do and then helping them find the right path forward—which might not be what they were thinking of originally—is something I’ve found unique to AWS. And I really enjoy that piece of my work.
What are you currently working on that you’re excited about?
CloudHSM is a hardware security module (HSM) that lets you generate and use your own encryption keys on AWS. However, CloudHSM is weird in that, by design, you’re explicitly outside the security boundary of AWS managed services when you use it: You don’t use AWS IAM roles, and HSM transactions aren’t captured in AWS CloudTrail. You transact with your HSM over an end-to-end encrypted channel between your application and your HSM. It’s more similar to having to operate a 3rd party application in Amazon Elastic Compute Cloud (EC2) than it is to using an AWS managed service. My job, without breaking the security and control the service offers, is to continue to make customers’ lives better through more elastic, user-friendly, and reliable HSM experiences.
We’re currently working on simplifying cross-region synchronization of CloudHSM clusters. We’re also working on simplifying management operations, like adjusting key attributes or rotating user passwords.
Another really exciting thing that we’re working on is auto-scaling for HSM clusters based on load metrics, to make CloudHSM even more elastic. CloudHSM already broke the mold of traditional HSMs with zero-config cluster scaling. Now, we’re looking to expand how customers can leverage this capability to control costs without sacrificing availability.
What’s the most challenging part of your job?
For one, time management. AWS is so big, and our influence is so vast, that there’s no end to how much you can do. As Amazonians, we want to take ownership of our work, and we want bias for action to accomplish everything quickly. Still, you have to live to fight another day, so prioritizing and saying no is necessary. It’s hard!
I also challenge myself to continue to cultivate the patience and collaboration that gets a customer on a good security path. It’s very easy to say, This is what they’re asking for, so let’s build it—it’s easy, it’s fast, let’s do it. But that’s not the customer obsessed solution. It’s important to push for the correct, long-term outcome for our customers, and that often means training, and bringing in Solutions Architects and Support. It means being willing to schedule the meetings and take the calls and go out to the conferences. It’s hard, but it’s the right thing to do.
What’s your favorite part of your job?
Shipping products. It’s fun to announce something new, and then watch people jump on it and get really excited.
I still really enjoy demonstrating the elastic nature of CloudHSM. It sounds silly, but you can delete a CloudHSM instance and then create a new HSM with a simple API call or console button click. We save your state, so it picks up right where you left off. When you demo that to customers who are used to the traditional way of using on-premises HSMs, their eyes will light up—it’s like being a kid in the candy store. They see a meaningful improvement to the experience of managing HSM they never thought was possible. It’s so much fun to see their reaction.
What does cloud security mean to you, personally?
At the risk of hubris, I believe that to some extent, cloud security is about the survival of the human race. 15-20 years ago, we didn’t have smart phones, and the internet was barely alive. What happened on one side of the planet didn’t immediately and irrevocably affect what happened on the opposite side of the planet. Now, in this connected world, my children’s classrooms are online, my assets, our family videos, our security system—they are all online. With all the flexibility of digital systems comes an enormous amount of responsibility on the service and solution providers. Entire governments, populations, and countries depend on cloud-based systems. It’s vital that we stay ten steps ahead of any potential risk. I think cloud security functions similar to the way that antibiotics and vaccinations function—it allows us to prevent, detect and treat issues before they become serious threats. I am very, very proud to be part of a team that is constantly looking ahead and raising the bar in this area.
What’s the most common misperception you encounter with customers about cloud security?
That you have to directly configure and use your HSMs to be secure in the cloud. In other words, I’m constantly telling people they do not need to use my product.
To some extent, when customers adopt CloudHSM, it means that we at AWS have not succeeded at giving them an easier to use, lower cost, fully managed option. CloudHSM is expensive. As easy as we’ve made it to use, customers still have to manage their own availability, their own throttling, their own users, their own IT monitoring.
We want customers to be able to use fully managed security services like AWS KMS, ACM Private CA, AWS Code Signing, AWS Secrets Manager and similar services instead of rolling their own solution using CloudHSM. We’re constantly working to pull common CloudHSM use cases into other managed services. In fact, the main talk that I’m doing at re:Invent will put all of our security services into this context. I’m trying to make the point that traditional wisdom says that you have to use a dedicated cryptographic module via CloudHSM to be secure. However, practical wisdom, with all of the advances that we’ve made in all of the other services, almost always indicates that KMS or one of the other managed services is the better option.
In your opinion, what’s the biggest challenge facing cloud security right now?
From my vantage point, I think the challenge is the disconnect between compliance and security officers and DevOps teams.
DevOps people want to know things like, Can you rotate your keys? Can you detect breaches? Can you be agile with your encryption? But I think that security and compliance folks still tend to gravitate toward a focus on creating and tracking keys and cryptographic material. When you try to adapt those older, more established methodologies, I think you give away a lot of the power and flexibility that would give you better resilience.
Five or more years from now, what changes do you think we’ll see across the security landscape?
I think what’s coming is a fundamental shift in the roots of trust. Right now, the prevailing notion is that the roots of trust are physically, logically, and administratively separate from your day to day compute. With Nitro and Firecracker and more modern, scalable ways of local roots of trust, I look forward to a day, maybe ten years from now, when HSMs are obsolete altogether, and customers can take their key security wherever they go.
I also think there is a lot of work being done, and to be done, in encrypted search. If at the end of the day you can’t search data, it’s hard to get the full value out of it. At the same time, you can’t have it in clear text. Searchable encryption currently has and will likely always have limitations, but we’re optimistic that encrypted search for meaningful use cases can be delivered at scale.
You’re involved with two sessions at re:Invent. One is Achieving security goals with AWS CloudHSM. How did you choose this particular topic?
I talk to customers at networking conferences run by AWS—and also recently at Grace Hopper—about what content they’d like from us. A recurring request is guidance on navigating the many options for security and cryptography on AWS. They’re not sure where to start, what they should use, or the right way to think about all these security services.
So the genesis of this talk was basically, Hey, let’s provide some kind of decision tree to give customers context for the different use cases they’re trying to solve and the services that AWS provides for those use cases! For each use case, we’ll show the recommended managed service, the alternative service, and the pros and cons of both. We want the customer’s decision process to go beyond just considerations of cost and day one complexity.
What are you hoping that your audience will do differently as a result of attending this session?
I’d like DevOps attendees to be able to articulate their operational needs to their security planning teams more succinctly and with greater precision. I’d like auditors and security planners to have a wider, more realistic view of AWS services and capabilities. I’d like customers as a whole to make the right choice for their business and their own customers. It’s really important for teams as a whole to understand the problem they’re trying to solve. If they can go into their planning and Ops meetings armed with a clear, comprehensive view of the capabilities that AWS offers, and if they can make their decisions from the position of rational information, not preconceived notions, then I think I’ll have achieved the goals of this session.
You’re also co-presenting a deep-dive session along with Rohit Mathur on CloudHSM. What can you tell us about the session that’s not described in the re:Invent catalog?
So, what the session actually should be called is: If you must use CloudHSM, here’s how you don’t shoot your foot.
In the first half of the deep dive, we explain how CloudHSM is different than traditional HSMs. When we made it agile, elastic, and durable, we changed a lot of the traditional paradigms of how HSMs are set up and operated. So we’ll spend a good bit of time explaining how things are different. While there are many things you don’t have to worry about, there are some things that you really have to get right in order for your CloudHSM cluster to work for you as you expect it to.
We’ll talk about how to get maximum power, flexibility, and economy out of the CloudHSM clusters that you’re setting up. It’s somewhat different from a traditional model, where the HSM is just one appliance owned by one customer, and the hardware, software, and support all came from a single vendor. CloudHSM is AWS native, so you still have the single tenant third party FIPS 140-2 validated hardware, but your software and support are coming from AWS. A lot of the integrations and operational aspect of it are very “cloudy” in nature now. Getting customers comfortable with how to program, monitor, and scale is a lot of what we’ll talk about in this session.
We’ll also cover some other big topics. I’m very excited that we’ll talk about trusted key wrapping. It’s a new feature that allows you to mark certain keys as trusted and then control the attributes of keys that are wrapped and unwrapped with those trusted keys. It’s going to open up a lot of flexibility for customers as they implement their workloads. We’ll include cross-region disaster recovery, which tends to be one of the more gnarly problems that customers are trying to solve. You have several different options to solve it depending on your workloads, so we’ll walk you through those options. Finally, we’ll definitely go through performance because that’s where we see a lot of customer concerns, and we really want our users to get the maximum throughput for their HSM investments.
Any advice for first-time attendees coming to re:Invent?
Wear comfortable shoes … and bring Chapstick. If you’ve never been to re:Invent before, prepare to be overwhelmed!
Also, come prepared with your hard questions and seek out AWS experts to answer them. You’ll find resources at the Security booth, you can DM us on Twitter, catch us before or after talks, or just reach out to your account manager to set up a meeting. We want to meet customers while we’re there, and solve problems for you, so seek us out!
You like philosophy. Who’s your favorite philosopher and why?
Rabindranath Tagore. He’s an Indian poet who writes with deep insight about homeland, faith, change, and humanity. I spent my early childhood in the US, then grew up in Bombay and have lived across the Pacific Northwest, the East Coast, the Midwest, and down south in Louisiana in equal measure. When someone asks me where I’m from, I have a hard time answering honestly because I’m never really sure. I like Tagore’s poems because he frames that ambiguity in a way that makes sense. If you abstract the notion of home to the notion of what makes you feel at home, then answers are easier to find!
Want more AWS Security news? Follow us on Twitter.
The AWS Security team is hiring! Want to find out more? Check out our career page.