Who Wants to Save a Billion Dollars?

Transcript

Rahul Subramaniam: Hi, Hilary.

Hilary Doyle: Hi.

Rahul Subramaniam: I have a question that I’ve got to start with.

Hilary Doyle: Sure.

Rahul Subramaniam: What would you do with a billion dollars?

Hilary Doyle: A billion dollar? Are you offering?

Rahul Subramaniam: Not quite, but at CloudFix, I’m really happy to inform you that we have now analyzed over a billion dollars of spend on AWS.

Hilary Doyle: I mean, this is great news for our listeners and for CloudFix. Congratulations. I feel like it was a mildly cruel setup for me. It took about 0.8 seconds for me to spend that billion on my business in my mind, which is why this episode is so perfectly timed because I’m hoping you will help ensure I do not spend a billion dollars on our business, but rather I save a billion on our AWS bill.

Rahul Subramaniam: I would love to do that.

Hilary Doyle: Great.

Rahul Subramaniam: So, today, the plan is this. I’m going to take you through the key lessons that we’ve learned at CloudFix over managing that billion dollars spend on AWS, and we are going to apply those lessons to our resident small business owner.

Hilary Doyle: Aka me. And not small for long. Aha. Bring it on, CloudFix.

This is AWS Insiders, an original podcast from CloudFix, bringing you what you need to know about AWS through the people and the companies that know it best. CloudFix is the nonstop automated way to find and fix AWS recommended savings opportunities. I’m Hilary Doyle. I’m the co-founder of Wealthie Works Daily.

Rahul Subramaniam: And I’m Rahul Subramaniam. I’m the founder and CEO of CloudFix.

Hilary Doyle: Woot.

Rahul Subramaniam: So, like I said before, Hilary, at CloudFix, we’ve analyzed the spending of a lot of money on AWS.

Hilary Doyle: A lot of money being, keyword, $1 billion.

Rahul Subramaniam: I didn’t want to say it again, but yeah. Thank you, Hilary. Yes.

Hilary Doyle: Avec B, aha.

Rahul Subramaniam: Yeah. So we’ve been incredibly lucky to be in a position where we have been able to look at AWS spend not only across thousands of customers but tens of thousands of AWS accounts. And while the big numbers are exciting, and in fact, I think you’ll find some of what we have to share today truly shocking…

Hilary Doyle: Whoa.

Rahul Subramaniam: I just wanted to give our listeners an example to hang their hats on. So I invited you and your startup, Wealthie Works Daily, to be the example for our audience of how to use these learnings to save. I’ve taken a look at your AWS bill, and I think I’ll be able to show you some more efficient ways to spend.

Hilary Doyle: That sounds wonderful. And we are very happy to be your guinea pig.

Rahul Subramaniam: So why don’t you tell the listeners a bit about Wealthie Works Daily so we know what your priorities are?

Hilary Doyle: I’ll give you a quick elevator pitch on Wealthie Works Daily.

Rahul Subramaniam: Great.

Hilary Doyle: Because it’s a fantastic business. So we are an investment and financial literacy platform for kids and families. We build proprietary gift cards that open and then can fund investment accounts for kids to help make use of those 20 years of time in compounding that often get forgotten about.

We’re really a fintech and a media company rolled into one. So we talk about ourselves as either a pension fund for babies or a kind of Sesame Street meets Bloomberg. And at the same time, we are educating kids and families about financial literacy, their own financial scripts, money, and value more broadly.

So yeah, we’ve needed an infrastructure in place that can reasonably carry us through our beta testing where things are smaller and more focused and then really be ready to scale with us the second we unleash our marketing campaign, which starts in the late fall.

But I should let you know that our entire marketing strategy for our India expansion is just your family since I’ve never met anyone who has a larger extended family. So I think we’ve got India on lock.

Rahul Subramaniam: Perfect.

Hilary Doyle: So not only are you going to help me with our AWS spend but also with one of our major launches. So thank you for that. I mean, just the first of many thanks in this episode, no doubt.

Rahul Subramaniam: Knowing all that, I’m excited to apply some lessons from CloudFix to the Wealthie Works Daily AWS bill and save you some money.

Hilary Doyle: Thank you very much. We are into saving money. That’s the whole point of the business. But first, as always, let’s get to our latest news from AWS.

We’ll start with some hot news about cold storage. Amazon’s S3 Glacier, now 11 years old… Can you believe it? Time flies. It’s always been known for providing long-term secure low-cost cold storage for data. Now, however, using standard retrieval tier and S3 batch operations, you can improve Glacier’s flexible restore time by up to 85%.

Hey, the ice caps may be melting, but Glacier sounds like it’s better than ever. Rahul, what does this upgrade really mean?

Rahul Subramaniam: Okay. Now, you’ve heard me scream about why everyone should use S3 Intelligent-Tiering, right?

Hilary Doyle: Right.

Rahul Subramaniam: Well, now, the Glacier tiers’ restore times are down by 85%. It makes even more sense to switch to S3 Intelligent-Tiering and save up to 80% on S3 storage. I’m going to repeat a fact yet again. 90% of S3 objects are accessed just once. So why is everyone still paying full price for sub-millisecond access to all the storage that they’re never using again?

Hilary Doyle: Save your money and save Rahul’s vocal cords.

Next up, AWS has just launched EC2 M7i-flex and M7i instances powered by custom 4th Generation Intel Xeon Scalable processors. If that sounds like gibberish, no problem. What is important is that they are now up to 15% faster than the Intel processors used by those other deadbeat cloud providers. Rahul, why does AWS get all this special treatment?

Rahul Subramaniam: Well, the competition across the latest generation of EC2 instances is getting really hot. With a huge variety of all of these variants, basically, each one is trying to carve their little niche and show themselves to be the most performant processor in the market. Now, the M7i and the M7i-flex are trying to compete with AMD processors in a very, very niche segment.

The issue, though, for AWS customers is that choosing the right instance type is just getting to be a lot harder. On my part, I am basically betting on serverless to avoid having to make any of these complex decisions.

Hilary Doyle: Finally, a bit of a flashback throw-forward. A few episodes ago, we took you behind the scenes for Amazon Prime Day with AWS chief evangelist Jeff Barr. If you haven’t listened, question your life choices and clear the afternoon.

Jeff just posted a blog covering this year’s numbers, and oh my goodness, the single largest sales day ever for Amazon and its independent sellers, more than 375 million items purchased, a peak of 126 million API requests per second to DynamoDB. Rahul, the only thing I need to know from you is, what were your purchases?

Rahul Subramaniam: Well, I can’t stop thinking about those figures. They were absolutely mind-boggling.

Hilary Doyle: They are.

Rahul Subramaniam: And I’m so glad that we had the opportunity to talk to Jeff just before the Prime Day actually happened. And about my purchases, well, I bought a ton of PLA filament for my 3D printer, and I’ve been enjoying printing tons of stuff with it.

Hilary Doyle: Hmm. That answer disappoints me. I was expecting more LEGO. Nevertheless, here’s to next year, and that’s it for your AWS headlines. Let’s get back to saving a billion dollars.

Rahul Subramaniam: All right, Hilary, let’s start with this. Why did you pick the cloud and particularly AWS as you went on this journey of building out your company, Wealthie Works Daily?

Hilary Doyle: I mean, I really appreciate that you’re asking me this, but I’ve learned from the best, Rahul. No, to be honest, we have a brilliant technologist who is extremely well-versed in all things AWS and, frankly, in all of the platforms.

And he made this determination early on that we were obviously a cloud-based business. We launch in Canada late fall, and we’ll expand into the rest of the world in 2024, 2025. And by that, I mean US, UK, parts of West Africa. We are built to scale, and we are built to move, and so it makes sense to be cloud-native.

Rahul Subramaniam: A hundred marks for that particular answer, Hilary. I mean-

Hilary Doyle: Oh, thank you. I was waiting for the billion.

Rahul Subramaniam: It’s coming there. We’re getting there slowly. But yeah, I mean, making the decision to move to the cloud for the reasons that you just stated, which is scale, you want to grow quickly, you want to be agile as you do this, that sounds perfect for an AWS-like setup.

So when I heard your sales pitch, the first thoughts that came to mind were, “Security is going to be key.”

Hilary Doyle: Yeah.

Rahul Subramaniam: Second one is you need compute that can scale up as you distribute all this content and all of these gift cards all over the world. And then, thirdly, you need to be very efficient as you scale up. So I think that’s a perfect segue to talk about the four biggest areas of spend that we have seen so far across all AWS customers.

Hilary Doyle: Okay, four areas.

Rahul Subramaniam: That’s right. The first one literally being compute. Now, compute accounts for about 40% of AWS spend, and you want to make sure that that is being used as efficiently as possible. The second one is storage.

Hilary Doyle: Right.

Rahul Subramaniam: And here, because you’re a financial services firm, you’re going to have tons of audit trails, logs, things like that where you can trace back to any transaction that has happened on your system. And those costs can start piling up very, very quickly.

The third one is around networking. And this is basically… We don’t pay much attention to networking-

Hilary Doyle: Good.

Rahul Subramaniam: … but those numbers can be significant if you don’t do things the right way. And the fourth one is RDS because that’s where you’re going to store all of your data, and you want to make sure that you have all the transactional integrity, you have all the backups in place so that if something were to go wrong, all this financial information is retained. And that can be a big expense as well.

Hilary Doyle: Okay. So the four areas are compute, storage, network, and RDS?

Rahul Subramaniam: Exactly. So let’s go through these sections one by one and see if we can shrink your AWS bill in each of these big four sections.

Hilary Doyle: That sounds great.

Rahul Subramaniam: Okay, so we are going to start with compute. And that, basically, is your EC2 bill.

Hilary Doyle: Love a good EC2 bill.

Rahul Subramaniam: You want a small one, right?

Hilary Doyle: Mm-hmm.

Rahul Subramaniam: Okay. So let’s start with a trivia question for you, Hilary.

Hilary Doyle: Oh, God.

Rahul Subramaniam: What do you think is the resource utilization for compute across all AWS customers? Which means that if they basically sign up and pay for a hundred units of compute, how much exactly are they using?

Hilary Doyle: I bet they’re using, like, 42% and paying for a hundred.

Rahul Subramaniam: So you think that they are somewhere near half as efficient as they could be in the ideal world?

Hilary Doyle: Yes.

Rahul Subramaniam: Okay, here’s a surprise and a shocker. It’s only 6%.

Hilary Doyle: You are kidding me.

Rahul Subramaniam: 94% of compute is just wasted away.

Hilary Doyle: Ugh.

Rahul Subramaniam: And here’s the additional piece of insight that we found with analyzing enterprise software companies. That number drops to about 2%.

Hilary Doyle: Oh, my goodness. I mean, this doesn’t surprise me since we only use, what, like, 9% of our brains? So we’ve just transferred that right over to compute God, humans, do better, do better. Wow.

Rahul Subramaniam: There are some very, very simple tips that AWS customers can adopt pretty much right away to start improving their resource utilization on compute. If you are starting from scratch, as you are, I would say start with Graviton. Don’t try to start with the Intel machines and then migrate over later.

But for folks who are already on AWS and have compute that’s running on Intel-based machines, the first and the simplest things that I would do is switch over to the latest generation of AMD processors because they provide 20 to 30% price performance improvement over the Intel ones, and they are cheaper, right?

So you get better performance with the latest machines, and they’re a lot more cost-effective. And the best part about that migration is that it doesn’t require any open heart surgery on your applications. You’re not changing the architecture of your applications as you’re doing that.

Hilary Doyle: Great. Sounds very wise.

Rahul Subramaniam: If I were to split out cost savings on EC2, I would split it out into two separate objectives. The first one is you always want to modernize, okay?

Hilary Doyle: Mm-hmm.

Rahul Subramaniam: Which means that when you’re in older generations of these compute instances, the thing that your team should always be focused on is, “How do I get to the latest instance as quickly as possible?”

Hilary Doyle: Right.

Rahul Subramaniam: Because one of the things that’s unique about AWS and completely different from your on-prem setup is that the latest generation of processors will always have a better price performance characteristic.

Hilary Doyle: Got it.

Rahul Subramaniam: The second category is what I call right-sizing. And this is where we just spoke about resource utilization. Right-sizing is basically reducing the size of your instances to a reasonable amount where you know what you’re utilizing, and you leave a little headroom on the top, and then you say, “This is the right size of the instance that I want to have for processing whatever compute workloads we have.”

Now, the on-premise mindset was, “Hey, I need to buy a machine now, and I’m going to keep it for the next five years.” So you start thinking about, “What kind of machines do I need five years down the line?” And then you basically over-provision like crazy, and you spend an insane amount of money upfront.

Hilary Doyle: Right.

Rahul Subramaniam: What we’re doing in the cloud is exactly the opposite, where you say, “This is what I need right now,” and you provision only that amount. And then, if you need more, you can always scale up. And that’s the beauty of the cloud versus the on-prem setup.

Hilary Doyle: What is the most common reason you see for companies over-provisioning?

Rahul Subramaniam: It’s on-premise mindset where you think that you are buying a server in AWS instead of thinking of it as a service that is dynamic and that can scale with whatever workload or throughput that you need to service.

Hilary Doyle: So maybe you’ve brought your original on-prem team up with you into the cloud as your business has migrated, and they are holding onto some old ideologies.

Rahul Subramaniam: That’s exactly right. Now let’s move on to the next big category.

Hilary Doyle: Okay. We’ve shored up compute, so now we’re onto storage.

Rahul Subramaniam: Right. So looking at Wealthie Works Daily, a vast majority of your spend happens to be in storage, specifically in S3. Now, that makes sense because you’re a fintech firm, in a sense, and you’re going to have lots of content, logs, and audit trails all stored on S3.

So storage is a very, very common area for over-provisioning and inefficiency that we have seen across all of our analysis. So let me just dive right into S3.

Hilary Doyle: I feel like storage is the bane of so many people’s existences. I mean, whether it’s downsizing your home or right-sizing your digital files, I think that this is just a universal challenge. It’s really hard to understand how much space you need. And people… You know, we all hang on to too much.

So we’re thinking about this kind of in real-time right now, and S3 is the storage solution that we have at the moment, but yeah, please keep our business from becoming a creepy hoarder. So let’s start there, yeah.

Rahul Subramaniam: Yeah. You’d be surprised. So here’s an interesting stat. AWS, not too long ago, published a stat that said that over 90% of objects that are stored in S3 are accessed exactly once.

Hilary Doyle: Oh, my goodness.

Rahul Subramaniam: And then they just stay there forever. And here we are talking about many tens of trillions of objects that are currently stored in S3.

Hilary Doyle: This is reminding me of my collection of old Canadian glass that came from my grandparents in Winnipeg, accessed exactly once.

Rahul Subramaniam: They’re on the top shelf collecting dust forever, and then you’re basically paying full price on it all the time.

Hilary Doyle: Yeah.

Rahul Subramaniam: So one of the things that AWS did very early on, especially with S3, knowing the fact that most of the data is never ever going to get accessed again, was create something called S3 Intelligent-Tiering.

Hilary Doyle: Right.

Rahul Subramaniam: S3 Intelligent-Tiering is a really, really awesome feature where AWS does all the analysis. They figure out how often you access objects or don’t access them, and then they can basically move those objects into other storage tiers that can be as much as 90% cheaper.

Hilary Doyle: Wow.

Rahul Subramaniam: In fact, if you go all the way up to Deep Glacier, I think it’s about 95% cheaper than your standard storage cost.

Hilary Doyle: I’m sorry, wait, Deep Glacier-

Rahul Subramaniam: It’s called Deep Glacier.

Hilary Doyle: That’s where it goes when you know it’s never going to be accessed again?

Rahul Subramaniam: Exactly.

Hilary Doyle: Although global warming, man. I don’t know. Those files are going to pop back up more often than people think.

Rahul Subramaniam: Correct.

Hilary Doyle: Okay. Pardon the interruption. Yeah.

Rahul Subramaniam: But the nice thing is that once you turn on S3 Intelligent-Tiering, it can move objects back into the more frequently accessed tiers. And AWS does all the hard work for you. The only thing that they don’t do is turn it on for you.

And it becomes a problem with AWS customers because they have to go bucket by bucket and turn on S3 Intelligent-Tiering to get this capability. And that’s kind of what CloudFix does for a lot of customers. We just turn that on automatically.

Hilary Doyle: I need S3 Intelligent-Tiering for my basement. But can we just talk about the environmental load for a moment of all of this storage? Because I think, as a society, we’ve now become quite accustomed to saving absolutely everything and knowing that, yeah, we’re not going to access it, but we like having it.

Rahul Subramaniam: Correct.

Hilary Doyle: Do you have a sense of what the environmental burn is on all of these files that we just keep out of habit?

Rahul Subramaniam: For me, the biggest proxy for the environmental load is the price.

Hilary Doyle: Is that true with AWS?

Rahul Subramaniam: Yeah, because energy is expensive, right? Everything is driven by energy, compute or storage, or any of these things. The lower the cost implies that you’re using lower energy. So as you move to Deep Glacier and some of these other tiers, they are stored away literally without being powered on.

So, for example, in Deep Glacier, it can take you up to four days to recover that data and bring that back into active service. However, when you’re on On-Demand, that data is actually replicated six times across six different availability zones-

Hilary Doyle: Wow.

Rahul Subramaniam: … so that it is made available to you in a few milliseconds of latency, right? So the energy consumption is many orders of magnitude more when you are in standard tiers or in On-Demand tiers.

Hilary Doyle: So think deeply about what you want your five nines to represent.

Rahul Subramaniam: You could either do that, or you could literally just turn on S3 Intelligent-Tiering and let the algorithms make that decision for you. It’s really that simple.

Hilary Doyle: So S3 storage is quite common. What about volume storage? How should we be managing volume storage?

Rahul Subramaniam: Okay. So that still forms a very, very large chunk of spend because a lot of the workloads come from the traditional background of an on-premise data center. People have migrated that stuff over. They still don’t treat all their file rights like objects. It’s still treated like a file system. So you basically attach these big disks to these big compute instances, and that’s a volume storage.

Now, about three years ago, AWS came up with a completely new volume type called gp3, which is the latest generation of their volumes, and it is 20% cheaper. And it is more performant in literally every scenario.

Hilary Doyle: Wow. 

Rahul Subramaniam: Now, the on-premise mindset would say that, “Hey, I’ve got something that’s next generation. It’s faster. It would obviously be more expensive. Or if it is cheaper, than it’ll be less performant.” But there is literally no downside to switching over from gp2 to gp3 volume types.

And yet, even today, only 15% of EBS volumes are on gp3. 85% are still sitting on gp2 volumes, which are 20% more expensive. They are also less performant, which is just such a shame.

Hilary Doyle: Such a shame.

Rahul Subramaniam: So that’s your primer on storage savings.

Okay. So now we are ready to move on to the third big area of spend.

Hilary Doyle: Networking.

Rahul Subramaniam: Exactly, networking. Okay. So we all know that AWS infrastructure is split up into 20-plus AWS regions, right, all around the world. Each region is then further broken up into anywhere from three to six or seven availability zones. And now, each availability zone has at least three data centers that constitute that availability zone.

Hilary Doyle: Hang on. I’m taking notes as you speak.

Rahul Subramaniam: Think of it as a big massive tree.

Hilary Doyle: Yeah, I’ve got a huge decision tree here that’s giving me a small hive, but continue.

Rahul Subramaniam: Now, when you’re deploying your applications, they fall somewhere within this tree. They’re in some region, some availability zone in some particular data center. And there are costs associated with when data moves from one place to the other, okay?

Hilary Doyle: Even in the cloud?

Rahul Subramaniam: Even in the cloud. You’d be surprised. So-

Hilary Doyle: Oh.

Rahul Subramaniam: … let’s start from the top.

Hilary Doyle: Okay.

Rahul Subramaniam: When you have any data coming into AWS, AWS says, “Hey, that’s awesome for us. It’s completely free.” But when data leaves AWS, there’s a hefty penalty you ought to pay.

Hilary Doyle: Ugh.

Rahul Subramaniam: So egress charges are pretty expensive if you don’t manage them the right way. Then there are other complexities. For example, within a single virtual private network, you basically have zero data transfer charges. But if your VPCs span multiple different availability zones and you have data transfer happening across availability zones, then there are certain availability zone transfer charges that apply.

Hilary Doyle: How do I create a VPC endpoint?

Rahul Subramaniam: Funnily, it is actually really, really simple. It’s one of those things that takes 10 seconds to execute. But sometimes, when you have accounts at scale and you have services that are being launched by teams all over your organizations, you just lose track of all of these ways in which people can possibly access your services. So one of the things that we do in CloudFix is we have a little automation that goes in and makes sure that those VPC endpoints just get created for you automatically.

Hilary Doyle: You’ve got a solution for everything. I’m starting to sense a pattern.

Rahul Subramaniam: There’s just one other thing that I’d like you to be aware of when you’re planning out your scaling and your high availability infrastructure in AWS.

Hilary Doyle: I’m ready.

Rahul Subramaniam: So, for the most part, if you don’t really need that kind of reliability, you don’t need to set up multi-region or multi-AZ redundancy. You should ask yourself the question, “What’s the worst that is going to happen to my business if one availability zone goes down for 10 minutes?” The probability of an entire availability zone going down for 10 minutes is remote and rare.

Hilary Doyle: Right.

Rahul Subramaniam: We have never in the history of AWS had an entire region go down for a while. So ask yourself, “Do I absolutely need it?”

Hilary Doyle: This is a very valuable tip because I think with our backs against the wall… At least if my co-founders and I were making this decision, absent our brilliant technologists, I’m sure we would vote in favor of having availability with multi-region redundancy just as a rule.

Rahul Subramaniam: Yeah. I mean, at the end of the day, it’s a risk-reward trade-off that you’re making. And there is a cost associated with mitigating all kinds of risks. For most businesses that I’ve ever looked at, I think you would be better served by just sticking with one region.

Hilary Doyle: Okay. All right. I mean, that is wise advice, and you start to see how easy it is for people who may not be technical founders to believe that they really need everything.

Rahul Subramaniam: Now let’s move on to our fourth and final big area of AWS spend, which is RDS.

Hilary Doyle: I’ve been looking forward to this one because I think that, as a company, we’re actually doing this pretty perfectly, but I’ll let you go on.

Rahul Subramaniam: Okay. So the way I think about RDS is you’ve got a database which comes from the old world. Databases, especially relational databases, which is the R in RDS, have been around for about 45 to 50 years.

Hilary Doyle: Right.

Rahul Subramaniam: Okay? And very little has changed in that world. The kind of consistency or asset properties that we associate with relational databases don’t traditionally sit in the cloud world.

However, one of the amazing things that AWS decided to do was to rethink what relational databases would be in a cloud world. And they rearchitected an entire database for a distributed computing mechanism under the hood. And that is what we call AWS Aurora.

It’s a database that was built from scratch, which has 10X the performance of any other relational database out there at one 10th the cost of a commercial database. And it has all the consistencies and redundancies and a whole lot of other amazing stuff baked into it and offered now as a serverless offering.

Hilary Doyle: What was AWS trying to fix or solve for with a relational database in the cloud?

Rahul Subramaniam: So this was a problem that Amazon had for themselves. They were massive Oracle customers.

Hilary Doyle: Oh, stop right there. Do you want to take a moment?

Rahul Subramaniam: The trigger word. But yeah. I mean, Amazon was spending billions with Oracle at one point of time, and they were constantly also struggling with performance of those Oracle rack servers. And they wanted an alternative.

Now, there was literally no other commercial alternative that was available that could scale to the number of transactions per second. I mean, take a look at any of the stats around Prime Day. We are talking about millions of transactions per second.

And so, the AWS side of the Amazon corporation decided that they were going to rethink what relational databases could do to scale up performance. And the architecture that they came up with was just so beautifully simple that it just made sense. And slowly, they started moving off Oracle completely and switched everything to Aurora databases.

Hilary Doyle: So just to be clear, we actually owe RDS Aurora in the cloud to Oracle.

Rahul Subramaniam: Yes. You could credit Oracle with the birth and creation of Aurora, and I think I can give them that credit.

Hilary Doyle: This has taken a real turn.

Rahul Subramaniam: But what we, as end users, gain from all of this is the fact that you now have a database that is way cheaper and way more scalable than anything else out there. But right now, over 80% of RDS instances are actually running non-Aurora databases.

Hilary Doyle: Oh, wow.

Rahul Subramaniam: And that’s just such a shame because you could be so much more cost-efficient if you were on Aurora. Not just cost-efficient, you just get so much more performance by just using Aurora.

Hilary Doyle: Why haven’t people moved over to Aurora?

Rahul Subramaniam: So I think it’s partially inertia and partially an unfair price comparison. I see a lot of folks look at the list price of Aurora and think that it is 15 or 20% more expensive than managing all that infrastructure themselves.

Hilary Doyle: Right.

Rahul Subramaniam: What they don’t account for is all the resources that are involved in keeping this infrastructure up, including patching and other maintenance activities. And, of course, the last thing that they don’t account for is all the new features and capabilities that get added to Aurora for free, which you could just never replicate on your own.

Take, for example, the fact that Aurora has a new Graviton-based instance that has 40% better price performance than the Intel-based ones. You just couldn’t replicate that with anything you managed yourself.

Hilary Doyle: You’ve given me so much to take back to our team. Thank you for all of these tips. And to the rest of you, I hope you also found this conversation as helpful. Please let us know if you have any follow-up questions for us or, rather, for Rahul. I’ll be here too as moral support. You can reach out to both of us at podcast@cloudfix.com.

Rahul Subramaniam: As always, please leave us a review and don’t forget to follow the show to get the new episodes as soon as they’re released.

Hilary Doyle: AWS Insiders is brought to you by CloudFix. They are an AWS cost-optimization tool. You can learn more about them, and you should, at cloudfix.com.

Rahul Subramaniam: Thanks for listening. Bye-bye.

Hilary Doyle: Bye-bye.