Max: 00:08 Hi. I’m Max Clark. I work with a lot of cloud native companies, companies spending 1,000,000 of dollars a month on AWS. And something that comes up a lot that end up talking about a lot is cost management and how you cost reduction within AWS and what’s the strategy that makes sense. A lot of what I’m gonna talk about comes out of conversations and dealing with companies that are spending between $1,300,000 on AWS.
Max: 00:30 When I say 1 to $3,000,000, I mean a month. 1 to $3,000,000 a month. So call it that, like, 10 to $30,000,000 a year range. If you’ve already crossed a $100,000 a month, so 1,000,000 ish a year on AWS and you’re scaling up, this will apply to you. You might be a little bit ahead of the curve, which is good for you.
Max: 00:45 If you are looking at EDP, this probably applies to you as well. If you’re in a position where you actually associating your cloud spend as a percentage of your revenue, this absolutely applies to you. So, you know, if you know, like, we’re spending 12% of our revenue on our AWS spend, you’re probably gonna pay attention to this. I’m gonna probably end up ranting around in a circle. Bear with me.
Max: 01:07 A lot to try to cover and talk about. The first thing is how people try to approach their AWS spend and spend reduction is this is usually wrong. If you do not already have your AWS environment tagged, all of your resources tagged, don’t spend any energy really trying to figure this out yet. You need to immediately try to figure out how to get your entire AWS environment tagged. If you can’t correlate your cost on your platform and your infrastructure to revenue it is generating, you’re not in a position to make decisions around what you’re spending and how much you’re spending and if it’s effective for you to spend it or not.
Max: 01:40 It’s a foundational issue. This isn’t a cost. We’re we’re not even talking about techniques to reduce your cost or or tackle your cost. We’re just talking about, do you know what you’re spending money on and why? You can attach lots of tags.
Max: 01:52 You can have tech things tagged automatically. If you’re in any sort of auto scaling groups, that auto scaling group can have it, you know, tags on it. If you’re using infrastructure as code like Terraform, you can attach tags to it. You can attach tags to resource accounts and fire stuff up to people that are I mean, you can create multiple dimensions of different tags. The important thing is, is at the end of the day, being able to come back and say this line of business, this product, this service, whatever it actually is for us generates this much revenue.
Max: 02:18 And this is what it costs us to run because you want to be able to have this conversation at the business level. 1st and foremost, we are spending X to get Y. Does it make sense for us to do that? And I have had lots of clients when they actually go through this exercise and they get into a conversation where the business is now evaluating what they’re spending versus what they’re making, I’ve seen massive realignments in how things are done. Like, just we’re just gonna wholesale eradicate an entire line of service or line of business because it just doesn’t make sense for us.
Max: 02:47 That’s fine thing. I mean, like, it sounds crazy maybe to you if you’re not doing it, but at the same time, like, that’s the kind of decision you wanna make. You wanna find out that, hey, you know, we’ve got a 70% margin on this thing and a 10% margin on this other thing. But that 10% margin is strategic for us. So we’re gonna continue to invest in it and grow it, but we understand what the margin is on it.
Max: 03:09 Or you could say, hey, you know, we’ve got this thing that’s generating 10% of margin for us and we don’t care about it. It’s just there because it’s legacy. Let’s kill that thing off. And I’m gonna tell you, you know, if you’ve got an engineering team that’s maintaining some sort of legacy anything and you tell them we’re gonna kill off this legacy application or service or whatever because it’s just not doesn’t make sense to the business, they’re they’re not gonna cry about it. You know, chances are people are gonna be really excited when you tell them this.
Max: 03:30 So number 1, if you’re not Chances are people are gonna be really excited when you tell them this. So number 1, if you’re not tagging, tag. If you can’t do cost reporting based on tags, tag. Always start with tagging. Now I get asked a lot by providers in our portfolio about multi cloud.
Max: 03:43 You know, how many of our clients do we see with multi cloud? And my answer to that is almost everybody’s maybe multi cloud curious. But when it really comes down to it at this size, usually not to the scale that anybody really thinks or expects. And there’s reasons for that. And I’ll circle back around to this in a little bit.
Max: 04:02 The big thing is when you start talking about an application, let’s, you know, call it, you know, $2,000,000 a month and spend, you know, or maybe you’re smaller. Maybe you’re you’re only a quarter $1,000,000 a month and spend. There just isn’t an opportunity to take an entire, you know, workload and and shift it to another cloud effectively. GCP did a really good job attacking this, you know, years ago where they introduced a data pipeline with, BigQuery and Bigtable. And the big thing that Google did that was really smart at that time was selling a fixed capacity or fixed slot processing so you could have a fixed expense for your data pipeline.
Max: 04:35 And this was great because you actually had controlled cost and you didn’t have to worry about what was going on with there and you could go crazy with it. It would just rate limit you and how much data and how much work it would do for you. But that was okay because it didn’t really matter if it finished in 10 seconds versus 90 seconds versus, you know, like, whatever. You know? The predictability of cost was more important.
Max: 04:55 You know? And this, of course, was back, you know, before Snowflake was what it is today, and there’s different techniques that you can you can do with Redshift. Now what happened if is, you know, GCP’s leadership changed and they shot themselves in the foot, and they changed the pricing for BigQuery and Bigtable, and that in conjunction with AWS shenanigans, let’s call them, you know, drove a lot of people off this platform. So where it was common for me to see people in that, like, $20,000,000 range have maybe, you know, 10, 15, 20% of their spend on GCP, I’ve seen all of that workload leave Google and go back to AWS and go back to single AWS. I see a lot of, vendors coming into market to do AWS, you know, cloud spend management that immediately try to attack this issue with instant sizing.
Max: 05:39 You know, we’re gonna shut down instances at nighttime for your development team. Look. Your development team is probably not large enough that turning off all the development instances is gonna make a meaningful impact to you and your spend. I I mean, you know, you’re it’s gonna get something. Don’t get me wrong, but it’s not gonna be like this massive win that’s gonna drastically change the, you know, arc of your business.
Max: 06:01 Unattached EBS volumes, I mean, yeah, of course, you’ve got them. You should find them. You should kill them off. It’s costing you money. It’s you know?
Max: 06:07 But you figure out a way to do that. You can use a platform to do it, but it’s not like finding lottery money by any means. Let’s see here. What else do we see? A spot.
Max: 06:14 We see spot a lot. By the way, we have vendors in our portfolio that manage spot and spot fleets. Big issue with spot is you do get a big reduction against your on demand cost because you’re getting resources that haven’t been allocated by AWS or something else. And the exchange of that is they can take those resources back from you at any time. So as long as your instance is relatively ephemeral, you can use spot effectively.
Max: 06:37 Places where I’ve seen that this becomes a really big issue is, of course, where instances aren’t really ephemeral or they have a long startup time or, you know, load penalty where they have to suck in a lot of data and it’ll do a lot of things before that instance is actually ready to serve traffic. If you’ve got that kind of application, spot is going to create other issues for you where now you have to maintain and figure out how to deal with state and, you know, seeding data in that instance in order to take advantage of the Spot fleet. And that maybe is engineering time you should be spending on other things. We’re gonna come back to EDPs. I’m gonna I’m gonna make an EDP note now, but we’re gonna come back to EDPs.
Max: 07:11 The conversation that I end up in with any provider that’s offering a bare metal service or a, you know, platform as a service, infrastructure as a service, any sort of alternative, you know, computing platform to the public cloud. Right? So for some reason, we call them hyperscalers, but I’ll just say public cloud, the AWS, GCP, whatever of the world is trying to attack this as a cost per compute or cost per byte of storage. And, you know, just to put it out there, the cost of compute for an EC 2 instance and the cost of byte of storage in s 3, yes, you can save money moving to different platforms. And, yes, you can save a good amount of money.
Max: 07:51 But this isn’t really the problem spends are. You know, cutting x percent, maybe 10, 20% of your EC 2 cost or, you know, 10, 20, 30% of your s 3 cost, it’s meaningful, but it’s not meaningful enough to actually drive, you know, that level of change where an organization is going to switch out of AWS, move out of the their cloud platform to go into a different system. It’s just that cost reduction isn’t worth the pain. Right? The pain that they’re dealing with doesn’t you know, going to that organization and saying, hey.
Max: 08:19 You’re spending 12% of your revenue on e c 2 and, you know, on on AWS, and the bulk of that’s on e c 2. It’s, like, 60 percent of your spend is on e c 2, and we’re gonna cut 10% of that out. And you’re gonna go from 12% of spend of, you know, revenue on spend, and you’re gonna go to, you know, 10.5 percent of revenue on spend. The engineering time and the resources and the risk and the question and the fear that you generate with that, it just you’re gonna spend a lot of cycles. And I will say this in 2 ways.
Max: 08:42 1 is 2 vendors trying to sell this stuff. If you’re wondering why your lead cycles are insanely long and don’t convert, you know, that’s why. And if you’re on the procurement sourcing, whatever side of this thing, if you’re an IT team managing DevOps, SecOps, whatever you wanna call yourselves nowadays, I don’t care how your org is titling you. Also, you’re gonna be barking up the wrong tree with one. We should also talk about KPIs.
Max: 09:01 I’m gonna come back to that too as well. The killer for you in your cloud platform is not what you think it is. It’s actually moving data around. It’s not storing data. It’s not it’s not process you know, it’s processing to some degree, and it’s not uploading data because they make sure you know, if you look at the at the building structure, oh, ingress is free.
Max: 09:17 It’s only the egress that’s the issue. But moving data around and understanding how data is flowing through your environment is the place where you really should be looking at and understanding what your spend is. And, boy, they make this really hard for you to understand if you’re looking at your invoicing. You know, there’s there’s I don’t want to say there’s a reason why the invoicing is so insane and hard to understand. And what you’re missing within that is what you’re paying really to move data around.
Max: 09:41 Right? And look, there’s some mistakes that you can do in terms of like a intra AZ, you know, that’s gonna cost you a fortune in data. There’s, you know, if you’re talking between, you know, you a fortune in data. There’s, you know, if you’re talking between, you know, what used to be like VPCs or, you know, intra AZ intra region, you know, kind of stuff, you’re probably getting fleeced without realizing it. If you’re moving data between regions, you know there’s managed NAT gateways there’s a ton of money there and in that example let’s say a $2,000,000 a month account maybe that account is spending $6 to $800,000 a month in EC2 but they’re also going to be spending you know, 3 or $400,000 in data egress and, you know, maybe 2, $300,000 a month on Kinesis.
Max: 10:19 This is a really common you know, and then you’ve got your RDS instances and then your s three buckets and then, like, you know, redshift and some kind of, you know, like odds and ends that flow around in that stuff. Now what’s interesting to me about that is when we talk about data egress, data egress is where, you know, these public clouds are just just absolutely crushing you in terms of what their cost is and margin on this 5¢ per gig egress. It’s a joke. It’s an absolute joke. I mean, it’s a 3000% margin or, you know, I’ve done the math a bunch of times in other videos.
Max: 10:46 If you’re really curious, we can maybe link it below here. But the big savings I see when people talk about and I hate this term, repatriation. Because I mean, so many companies are cloud native. Right? You can’t really say we’re coming back into a data center because you weren’t in the data center in the first place.
Max: 11:00 But selectively moving workloads out of the public cloud platform into a bare metal platform. You know, there’s lots of cases where maybe that is driven by a resource constraint. You know, you can’t get enough NVMe drive. You can’t have enough, you know, fast enough disk. You need a faster network interconnect.
Max: 11:13 You wanna have 100 gig or x by 100 gig interconnects between your servers. You wanna do real. You know, you’re doing AIML in a real serious way, and you wanna have InfiniBand between your servers. You know, you need more storage that you can get that’s actually attached to the node. Like, you wanna have a specific CPU combination at a higher quantity.
Max: 11:30 You wanna have you know, like, these sorts of things drive bare metal. They just do because you can’t get it out of the cloud platform. And what’s also funny about this is it’s not necessarily a I mean, price drives this, but it’s not price driving that decision as much as people wanna believe it is. It’s usually that we actually can’t do our job if we’re here in AWS. We need to do something else because we need to do X, Y and Z.
Max: 11:51 And by the way, that comes with a big cost savings. And then, you know, by the way, it’s a big cost savings, right? You’re making a trade. You know, there’s there’s different impacts, right? You lose the ephemerality.
Max: 12:00 And I would argue that you don’t really have the ephemerality that you think you do because you’re buying our eyes and savings plans and doing other stuff that eradicate the ephemerality. So you have the, like, the sense of ephemerality. Right? It’s the it’s the it’s the pretend lie that you have ephemerality. So you just have to think about your environment a little different.
Max: 12:14 Right, and how you manage it. Maybe you’re on a 12 month deal with a bare metal provider, which we do all the time with our clients. So it’s really the data. Right? So we talk a little bit about egress.
Max: 12:22 And by the way, there’s different techniques here that you can do, and we can talk about if it makes sense to you. But there’s different ways of managing your egress. You can change your NAT gateways. By the way, these are hugely big wins for companies. Little simple things.
Max: 12:33 I’ve seen, you know, old technique I’ve been doing for 20 years, like, literally, like, sit down with your team and draw out your pipeline. This whole, like, trying to do it in Visio or or Lucidchart or whatever you’re using today, Miro, like, you’ll never finish these things. By the time you actually document the diagram, you know, your environment’s already changed, like like, literally sit down with people and and draw it on a whiteboard and talk through the different steps. And as you ask questions about how things are being processed and actually work through, you’re gonna find some fun stuff like, oh, we have, an ingress pipeline that takes data from the Internet, goes to an ELB, and then hits an auto scaling group. And, you know, we’ve been pushed to go to to serverless, so we’re using Lambda then after you know, we’re taking to an auto scaling group, which then puts into an s three bucket, which then fires off, you know, this thing in Lambda, which then puts it into Kinesis, which then runs another Lambda thing, which then puts it back into Kinesis, which puts it into a it goes through an ELB to another auto scaler group, which goes back to Kinesis, which then writes to to DynamoDB.
Max: 13:28 It sounds funny. And, like, that’s not ever possible as a real world example from a client. And that one that client, by the way, also had another issue, which is when they created their DynamoDB tables. They never set TTLs on them, and so they just had this, you know, growing expense in Dynamo. And, when it became a problem and they recognized what was there, ended up having a decision paralysis issue because they couldn’t, you know, going through and retroactively assigning TTLs to Rose, the time and the cost of it was just so high, like like, so absurdly high to actually put TTLs, you know, inside of their I mean, it’s just so funny to think about at this point.
Max: 14:02 They actually ended up deciding to start writing to a second DynamoDB database and with a TTL on that one and ran them in parallel for 13 months. And then when they hit the 13 month mark, they just dropped the 1st database and eradicated it. And that was modeled out to be cheaper than trying to go back and retroactively assign TTLs and deal with the actual cost within AWS. Managed services inside of AWS is a big topic for me a lot. And there’s, like, points where it doesn’t make sense for you to figure out how to do another thing.
Max: 14:29 You know, like, under a certain dollar amount, it’s worth it to run EKS versus running Kubernetes natively inside of your infrastructure, you know, or running in containers in a different way. I’ve had clients with an auto scaling group. When the e c two instance starts, it just downloads and loads the container and fires a container up. There is no Kubernetes involved. It’s just container management for software delivery with no complexity of Kubernetes, which is actually crazy, but at the same time really smart in ways, you know, just because it was a very simplistic environment for them to manage.
Max: 14:58 And there’s different container platforms that give you different you know, like, you don’t have to be on Kubernetes. There’s different ways of doing this. So anyways, EKS versus Kubernetes, Kafka versus Kinesis, Cassandra, Sickla versus DynamoDB. And these things all run on top of an EC 2 instance inside of your AWS environment. It’s just are you paying do you need to pay the premium for the managed branded service from AWS or not?
Max: 15:20 And again, there’s points where it doesn’t make sense, and then there’s points where all of a sudden it makes a lot of financial sense to change that. And also, there are service providers in the market that we work with for and this is all they do. Right? You don’t have to run Kinesis and pay the premium for AWS branded Kafka. We can just install and run Kafka and manage it for you on top of VC 2, and it’s gonna give you the exact same thing.
Max: 15:41 And it’s gonna cost you a whole heck of a lot less. So these become like and and again, we’re not talking about leaving AWS here. We’re just saying how are you deploying resources and what atomic unit inside of AWS are you actually consuming? Because the atomic units inside of AWS compute and storage, you know, these are really good. Like, they’re really good.
Max: 16:00 And, yes, you’re gonna save money if you don’t use them and you go somewhere else. But if you’re already there, the switching cost, you you know, you have to model that and you have to figure it out. And as much as, you know, some of our providers and portfolio wanna pretend like this happens, I mean, the reality is is it just doesn’t to the scale that anybody really thinks it does. I wanna talk about KPIs. Let’s talk about KPIs for a moment.
Max: 16:18 And if you want to attack this, you have to make this an issue across the organization. Right? You can’t just go to your finance department and say, hey. Finance or procurement or sourcing, I need you to figure out how to save us 10, 15% on AWS this year. We gotta lower our cost 20%.
Max: 16:35 They have no control over any levers to actually do that. The only levers they have control over is they can buy a savings plan, they can buy our eyes, and they can sign an EDP. Those three things maybe maybe not what you want. Okay? So you’re putting them in a position where they they don’t actually have levers or control over what you’re trying to drive as an outcome.
Max: 16:53 I’ve seen this as well with, you know, what you would call, like, a DevOps team or the DevOps team is in charge of the cloud environment and cost management and cost control becomes part of their responsibility, but they’re not actually building the platform and making architecture choices for how it’s used. And that engineering team, and this is a verbatim quote, came from a client, you know, DevOps team, IT team was trying to make changes and get changes made, and they wanted to do things a little differently, not radically differently, just a little differently, and got into a conversation with, an engineering lead about spend. And let’s do something a little different that’s gonna drastically change how we’re spending money in AWS. And the engineering lead looked at the DevOps lead and so funny to think about. Literally said to him, that’s not my KPI.
Max: 17:44 I don’t have to worry about that. I mean, show me the incentive. I’ll show you the outcome. Right? So it’s just so funny to think about still to this day.
Max: 17:53 But it’s really true. You know, that engineering lead had a completely separate and disconnected set of priorities that they were being judged on and had to achieve. Right. You know, what was their performance? The application speed of delivery product cycle, like all these other things become big influences for your engineering team.
Max: 18:11 What is your product roadmap? What kind of are you on agile? You’re trying to do sprints. What’s your, you know, like what? How do you fit all this stuff into there with them?
Max: 18:17 And by the way, this is why I go back to tagging. You can introduce tagging without spending developer time to get it done. So you can get a lot of information inside about your environment without blowing engineering time before you actually make decisions of, like, hey. You know, this is where we’re spending money and we’re gonna change these things because it’s just not worth it to us. So this has to be, like, executive owned, push down the ranks of we’re gonna deal with this.
Max: 18:41 Okay. Let’s get into the EDP and talk about EPs. EDP starts showing up once you cross about a half a $1,000,000 a year is what it feels like. You know? So if you’re spending around 50 k a month on AWS, you’re probably starting to get phone calls from your named account rep at Amazon talking to you about these different programs.
Max: 18:58 And, again, this is what they’re incentivized to do. So So that’s what they’re gonna do. You are making a trade. The EDP is going to give you a discount in exchange for future commitments of spend. Okay?
Max: 19:09 You’re just horse trading. It’s important to really understand and think about that. If you are getting a 10% discount across the board on your AWS bill, AWS is getting that money back from you in the future. You’re kicking the can down the road. And this gets realized a couple of different ways.
Max: 19:29 The most common thing is that you’re gonna have an annualized commitment of multiyear typically. Right now, look, if you’re doing your first EDP, you’re gonna have a lot more leverage. The big secret here is it’s it’s heroin. Right? Like like the the first one’s free, and then it gets more expensive every time you come back.
Max: 19:49 Right? Because now you’re used to the discount. You know, you’re budging around it. Like, you you’re just you know, your your optionality and your leverage decreases with each renewal. Let’s just say you’re at $1,000,000 a year in spend and, you know, this EDP thing sounds great to you.
Max: 20:03 Right? And let’s say you got an offer for a 10% reduction. Right? So, you know, you’re gonna go for a $1,000,000 to $900,000. Right?
Max: 20:10 Like, it’s it’s meaningful. Like, you’re, you know, you’re talking about, you know, staffing a role usually at that point. Right? Like, you’ve got space and budget to do something with it now that you were already spending that just materializes out of thin air. That’s that’s usually a good trade.
Max: 20:25 Problem is is now you’ve just committed yourself to spending $1,100,000 at a 10% discount. And then year 3, you’ve also got another increase. And then when you come back year 4 to get an EDP renewal, you’re going to get your baseline reestablished at where you’re at. It is now again, if you’re in early cycles with EDP, you know, like you’re doing your first one, maybe you can negotiate away from being required to purchase enterprise support. At some point it’s gonna be forced on you.
Max: 20:56 This is the point that really makes just like I I mean, I’ve seen this so many times. It’s just it’s incredible. It’s like, oh, you know, we’re signing this deal with AWS. Oh, you get a 10% discount, but then at the same time, you’ve gotta spend 10%, you know, 8 to 12% on enterprise support. So it’s like, great.
Max: 21:11 You’ve saved money on this bucket, but then immediately spend it on this bucket. No. By the way, each year, your buckets are increasing. The secret with the EDP is negotiating PPAs. And if you’re not negotiating PPAs across each of your services, I mean, you’re really that’s where you’re leaving the money on the table.
Max: 21:28 And this is where you attack your egress costs and you, you know, data exchange and, like, basically anything that AWA look at the right commitment, everything is negotiable. Right? You know, you just have to understand that what you’re trading, like, this isn’t free money for you. You are making a trade. Buying an RI to get the maximum discount, you’re paying you’re you’re giving AWS in advance and you’re you’re buying that you know, you’re you’re prepaying for your usage of that thing at a lower margin.
Max: 22:00 And, you know, you’re switching columns in Excel on how you’re, you know, how they’re realizing that and it’s going from, like, oh, this is an on demand resource with no, you know, recurring revenue at x margin and in in exchange where you’re gonna prepay us for 3 years or maybe it’s not a prepaid. Maybe you’re just signing a 3 year deal for the RI, and we’re switching you from nonrecurring to recurring revenue. And we and we see I mean, it’s it’s better. It’s it’s it’s more valuable for us. Right?
Max: 22:23 So it’s worth less margin. The evils with EDPs that I’ve seen, of course, become these escalators, you know, and the renewals. Escalators and renewals are really emotionally complicated topics within teams. And this is also why at this size, multi cloud doesn’t become the reality of thing. People think that is.
Max: 22:42 And I’ve seen lots of clients get bound up where, you know, we talk about that, oh, we had that 10% of our spend was on GCP because we’re doing our data pipeline there. But now we have a shortfall in AWS that we’re contractually committed. We’re writing that check. Right? Like, we’re not spending the money there.
Max: 23:01 We’re not using the resource, but we’re still writing the check for it. And what happens? It’s like, oh, we started looking for things to backfill that gap, right, that shortfall. And I’ve seen almost every single environment, you know, that GCP, you know, data pipeline, boom, immediately comes back to AWS. When I say immediately, I mean, it takes a lot of time in reengineering resources and bunch of stuff.
Max: 23:20 I mean, it’s painful, but it comes back. I’ve seen CDN relationships end because they needed the revenue to go into Crowdfront in order to come up with with shortfalls. You know? Like, we’re talking about, like, core process and core infrastructure for this company and how they were actually leveraging their CDN and losing functionalities and features and things that they actually needed and needed to rearchitect application as a result of it, but they had to come up with that shortfall. Or it’s like, hey.
Max: 23:45 How do we get things that we’re spending that we have direct relationships with into the AWS marketplace in order to get spend draw downs? And, you know, when you when that happens, like, maybe the 1st year you’re getting a one for 1, but then you’re getting lower and lower ratios, you know, out of that spend commitment. So maybe, you know, you move New Relic or Datadog back into AWS into the marketplace and buy it there in order to deal with your your spend. But, oh, guess what? You know, you may you were getting 1 for 1, but now you’re getting 50¢ on a dollar, and then you’re getting 20¢ on a dollar, and then you’re gonna get 10 percent 10¢ on a dollar.
Max: 24:15 And, you know, that just changes. Like, you have no control over that. Like, all of a sudden, like, you know, sans sands are shifting at your feet. This is much longer than what I usually do for these. There’s a lot more to this topic.
Max: 24:25 We can get into this, you know, like like, the this it’s complicated. And a lot of this just depends. I’m trying to give really generic information that applies as a whole. I’ll circle back to the beginning of this, which is if you’re not already doing it, you have to tag all of your resources, and you have to be able to understand and correlate what you’re spending money on with what revenues coming in the door attached to it. And this is not development cycles.
Max: 24:48 We’re not talking about trying to get code coverage and unit tests across your entire environment where your developers have to write them. You just have to start enabling these things, and it it just needs to be perfect enough. Like, don’t don’t worry about the strategy of how you’re tagging these things. Just tag everything. And if you have to go retag everything later, then go retag everything later.
Max: 25:04 Like, who cares? You know? And if you got any sort of auto scaling anything, if you add the tags into it as the instance creates, you know, your auto scaling group is gonna end up retagging itself by itself with enough time, you know, so don’t just do it, you know, and just arm yourself with enough information that you can actually start making, like, informed decisions of, like, is this good or not? Like, do we spend money here or not? You know, are we spending too much money here based on what we’re making?
Max: 25:31 Are we losing money here? Is this strategic? Is this not? Is this essential for this other service? Is it not?
Max: 25:37 Is this a line of business that we don’t want anymore? Is it not like like, I mean, it’s it’s not sexy, but boy tagging and being able to do, you know, proper analysis of spend is is where you have to start. And then from there, we can talk about how you’re spending money and and how do you approach maybe spending less money, provided, of course, you don’t have an EDP that’s forcing to spend the money. But then we can have a conversation and we can talk about whether or not bare metal makes sense for you. We can talk about switching off of a managed service makes sense for you.
Max: 26:10 You know, And maybe it does. Maybe it doesn’t. But you you got to have the information in order to know, and you have to understand the trade offs. Right? And then at least you’re informed and you can understand the trade offs.
Max: 26:20 Yeah. It’s complicated. Simple, but it’s complicated. Anyways, I’m Max Clark. Rant on, you know, really AWS spending and cost management.
Max: 26:29 I hope this helps. If you have any questions, comment below. Reach out. Happy to chat. Happy to help.
Max: 26:35 I have been seen some really crazy stuff. Don’t worry. It’s more normal. Whatever you got going on is more normal than you think. You’re not alone.
Max: 26:43 I hope this helps you. Have a great day
