Cloud Computing Environments and Application Workloads – Rahul Subramaniam, CloudFix

This is Textron TV. Hey guys. Thanks for the throw.

We're here with Raul's subramanian who is the chief evangelist for cloudfix? And they were at the recent AWS reinvent conference and they actually surveyed a few of the attendees to get their Vibe and what they're thinking about and they're pressing concerns which of course Brings us to 2023. Well, welcome the shower.

Thanks for having me Mike. This is a pleasure. So walk us through the survey results a little bit.

What did you guys find and most importantly what if anything surprised you? Okay. So I think coming into reinvent is very clear that almost everyone had one thought on their mind which is cloud costs are rising.

And how do we better manage it and what everyone was looking for was some guidance from AWS on how to do this really? Well and I think what really caught me by surprise was the fact that They didn't really address it head on they could have you know. Just talked around the topic a little bit but didn't really address it at all.

And I think that was a missed opportunity for them. So when your survey results what exactly where people looking for or what kind of leaps at it you as the top priority for the coming year. Great.

So when you look at all the customer base of AWS over the last 15 years AWS has had over 200 different service launches and it's now at about a hundred thousand plus API are way more than that. So. The challenge that most customers have is how to use these apis and these Services the right way and AWS culturally as much as I love them have been extremely non-openinated about how customers use their services and they want customers to do whatever suits them the best and given that scenario what customers were really struggling with is navigating this massive space of services and API and trying to figure out what the right way to do.

I mean, this is the classic challenge of too much choice and too many options. Like there are 25 different ways of launching containers in AWS and while all of them have a place and there's a reason for it. Customers who are starting off with AWS and have truly bet on AWS are not asking the question.

How do we do it? Right because when they get started there's this initial promise of hey, the cloud is actually way more efficient, but when they get started they find that they costs are going out of control. And so their biggest concern is how do you do it?

Right so that you're leveraging Cloud for what? It is really good at? And that is the thing that I think AWS kind of missed addressing for that particular customer base.

Did you get to two of three other areas, which is to say that? given the current economic scenario You should really look to evolve and innovate. Otherwise, you will perish and die and this I think was very very beautifully pushed by one of Google's in his Keynote.

But I think that was one of the few very succinct and clear messages to the customer base to say do this just evolved don't don't go into a cost-cutting more given the current economic scenario try to innovate and evolve through this that's your best bet and then Swami came in with his keynote and said, Data is basically going to be the biggest driver because as the entire customer is has seen moving all your existing workloads to the cloud is hard. It takes time. It is painful and in a lot of cases it may not even make sense.

So the bottom line is that the biggest driver of innovation is going to be all the data that you already have get into the cloud do some amazing stuff with it. And there are amazing tools that are now available just to actually leverage the data to do something interesting. And then the last one is that AWS actually went ahead and spent most of the last year actually working on a lot of governance and making sure that you have all the tools for governance to make sure that your data is secure you don't have any of those concerns.

but the bottom line for all the customers really came down to Amazing tools amazing Services. We love AWS. We want to that's where we're making the big bets.

How do we be more efficient? It seems like to me at least there's a lot of blame to go around here because it's not just AWS. It's the problem a lot of organizations allow their developers just to consume resources and there's not a lot of review or governance over what those guys are doing.

Exactly and what costs what so And the customers themselves part of the problem. I do believe that traditional organizations aren't really geared to handling the new Cloud environment and the mechanisms. I mean historically you had finance that basically used, you know, capex as a mechanism of managing control over what gets spent and that was their mechanism of making sure that engineering could only spend x amount then you had procurement that it all the due to due diligence and made sure that when you were buying Hardware or your buying stuff for your on-premise environments that you had the best deal possible and then Engineers were basically doing whatever they wanted to do, but they were always capped by the resources that were made available to them using the capex model.

Now come Cloud you suddenly have this ability to launch instances to do amazing stuff and really accelerate the Innovation, which is again the core promise of the cloud. But Finance basically when moving to Opex has kind of lost control over costs. It's always retrospective.

You see a bill at the end of the month and you scream and you're going to agitated would say what the hell happened and that's kind of the current board and they don't know how to deal with it procurement finds that it's role is not really, you know defined because here everything is dynamic. Everything is on demand and their role where they were doing stuff up front doesn't really exist. I mean imagine if you are starting to launch a cloud resource and you had to go through, you know, seven layers of approval that just doesn't make sense at all in the cloud environment and Engineering has no capability to or they've never been trained to have Financial discipline because that's not their core competence either.

So I think Finos plays a really important role in trying to bring in some of these organizational structures to help organizational navigate the stuff better, but that's said guidance. the cloud service providers is really also key at the same time to making sure that the organization that are consuming their service have a roadmap have a guide on how they can leverage these things and do things efficiently. one of the issues that also comes up is the relative guidance you get from AWS about how to consume these services and it seems like they're pushing people towards graviton and they're saying that if you are concerned about costs, you should essentially refactor workloads to run on this.

New class of processors that they've built on top of an arm architecture. Is that feasible for organizations? I mean, what impact do you think that will have?

I think there are three tiers of getting cost efficient and cost savings. It's not just the graviton stuff. So Tier 1 is really the super simple stuff that AWS have been recommending forever.

Like if you look at their blogs they put out about 3,000 different recommendations every year Um, and that's a whole lot to go through but there are tons of very insightful simple recommendations that you literally just can turn on and get a bunch of savings and this is probably 10 to 20% of savings on your bill. They've been recommending it in a very subtle way not really being explicit about it, but those recommendations are there and that so we build a solution called Cloud fix which basically just tackles that yes, super simple AWS recommended fixes that you know, you just turn on like a Roomba and let it go and you know get you the savings. So that's tier one and it's super simple.

It actually fixes your problems. Not just you know, find them. The second tier is where?

You actually start using AWS as managed services. And you start, you know, instead of managing these Services yourself and having custom, you know builds for these use a managed service the economies of scale and the efficiency that you get from not having to manage stuff. Is phenomenal and you get to save another 10 to 20% by using those kind of services?

The third set is literally architecture changes these some of these changes require major application surgery some of them required to change the way you architect your product. These recommendations include things like moving into a completely vendorable architecture or moving to graviton as an example. These are all very higher order changes that you could technically make to your product.

But the benefits of there is you can actually end up saving up to you know, 70 80% on your existing costs if you do everything, right? so there are different levels and maturity levels that organizations are usually at and I believe that they have to take us very stepwise approach to achieving these Cloud cost savings. What is the challenge in getting people to have that higher level of maturity because you we hear about Cloud Ops and devops forever in a day and doesn't seem like a lot of organizations are applying that with a high level of discipline to their cloud computing environments so fundamentally, what's the issue?

I think the issue is the distributed nature of responsibility. This is a combination of engineering and finance. Where the execution is the responsibility of engineering but the controls the requirements and the constraints are provided by Finance.

And finance doesn't know the trade-offs that engineerings has to make and Engineering doesn't know or doesn't usually relate to the constraints that Finance puts on them and they invariably tend to be on in loggerheads with each other across organizations. And I think that's where the finops organization and phenobs as a practice. is evolving to help organizations solve that that fundamental problem in their organizational structures you need a team that really understands the execution as well as the financial aspects of it.

To be able to do this really well and unfortunately traditional organizational structures don't allow for that. There are very clear boundaries about what Finance does and what engineering does but this one is starting to merge the two for this to be effective. Do you think in some point we'll be using artificial intelligence that figure out how to optimize these environments.

It seems like perhaps we need a machine to save ourselves from the machines. That's a good point. I think there are certain tools that use machine learning.

I wouldn't say artificial intelligence that's specifically talk about some machine learning algorithms which apply which absolutely are very useful. So I normally detection for example is a great way to make sure that you know, when you have a certain cost Trend and you suddenly see a big spike, you can identify those using machine learning techniques and take action. Unfortunately at the end of the day.

What you deployed and the intent of what you deployed and how you deployed it and where you've deployed it is hard for a machine to discover today. There are a lot of best practices there are a lot of you know. A lot of simple things that you can do to be efficient, but I think those are more rule-based at this point.

Right, like there is a new service that AWS creates which is cheaper Next Generation faster and more performant. And you know, the price is great. The rule of thumb should be get on the latest generation of the instance or get on the latest generation of the EBS volume so that you save money.

Um, that's not something that you need AI for I don't think that we are at that maturity level where AI drives a lot of that cost savings and stuff like that. I think we need to start with some of the basics. We are not yet there at that level of maturity where the basics are all done.

And then we need these higher order, you know machine learning and artificial intelligence driven cost savings. I think we are way off from that level. We are still scratching the surface on getting started with cost savings.

Speaking of Basics. What's your assessment of cloud security these days. I mean in theory the cloud platform is probably more secure than the on-premise platform, but it sure seems like the processes that we're using leave a lot to be desired.

Absolutely, I think without a doubt if used right. What you can do in the cloud far exceeds anything that you could dream of. on your in your on-premise environment, like it just doesn't make sense to make anywhere close to the Kind of Investments that the cloud providers are making on security but at the end of the day, it's really important also to realize that this is a shared security model.

There is a stack below your application that the cloud security provides to the cloud service providers give you Which is incredibly secure if you use it the right way. Unfortunately a lot of organizations. Still haven't caught up on learning on about what it means to use the cloud in the right way.

I have I still see a bunch of you know, security and compliance requests that come to me that still say. Hey, we want to make sure that you have a fire extinguisher next to your door in your data center. And please check the box and a lot of the security these days ends up becoming this checkbook exercise without truly understanding what's happening under the hood and I think that needs to change and I think the cloud providers like AWS make it really easy to leverage a lot of the very heavy lifting that needs to be done.

They've already done all that hard work for you. So if you use something like the well architect framework to kind of Define all the other security parameters around your application, I think invariably you end up in a place that's 10x better than anything you could have imagined on premise. but at the same time if you don't do it, right you can also screw up a lot of things so it comes with There's a lot of power in there.

But there's also a lot of responsibility that that comes with it speaking of the opportunity to screw things up. We hear a lot about Cloud native and kubernetes in the cloud these days but there's a lot of complexity involved. So how do I kind of balance the benefits of kubernetes which allows me to scale up and down more easily and the complexity of a platform that has more bells and whistles on it than anything I can remember.

so I'm probably going to take a very contrarian view to the views of the industry here. I actually do not think that kubernetes is cloud native. It's it's a misnomer.

kubernetes is the least common denominator product that allows you to span your workloads across the cloud and I don't even understand why people want to do that because they they had this false notion of some kind of vendor lock-in that doesn't make any sense and kubernetes while started off being aware of utilizing your you know, spare or underutilized servers on premise. and just making the most of it it turned from there into the goto tool for everything and I think that was the big mistake kubernetes doesn't really get you any of the real benefits it in fact complicates life for you know, and that's it Cloud native for me is using cloud service providers managed services that do all the heavy lifting for you with with and just give you a simple API and a simple interface to deal with it when I talk about this. It's like AWS forecast AWS pinpoint.

These are all kinds of services. Which are higher order services that they manage they run they and they have a very crisp and clear interface about what the product does. And how you should access it right kubernet is giving control to your developers to do, you know a million things with it.

Without the Deep expertise and how to manage it just doesn't make any sense at all. So, you know no offense to the cloud native foundation and the kubernetes team. I think they built a great product but the use case for that product was I've got a ton of servers that are lying around being wasted.

Let me figure out a better way to use it. Yes. It is.

Awesome for that case not multi-cloud not I want to build the next Cloud native application and let's me. Let me build in kubernetes that those are the wrong decisions to make there. No, ultimately is complexity kind of our biggest enemy here because if I look inside of any organization, they're consuming a ton of cloud resources, but it's across a very narrow band of surfaces because the other ones are just too complex to get it.

Yes, I would say complexity is the enemy but the bottom line for most organization just becomes how efficiently are they operating and running to solve The Core Business problem that they've set out to to solve right complexity is one of the factors like at the end of the day it all boils down to cost. That's that's the way every organization sees things. What is the cost for me to solve the problem that I have at hand?

The cloud solves the problem efficiently for certain classes of problems. And organizations need to learn how to leverage that. If you have tons of data, if you've got certain, you know, if there are insights you want to get out of the data that you have or there are very standard workloads that you already run today get them to the cloud get them to be managed simplify them.

That's what brings the cost down. Right and use them the right way. So there is a ton of hygiene and stuff that needs to be done.

And yes with the cloud providers now with 100,200,000 apis that are available to you. There is complexity in that alone. Just knowing how to do things, right?

Is is hard to keep up with and that's kind of why I keep saying efficiency and Cloud cost savings are non-stop. You have to keep up with it all the time that are new Services coming all the time. There are new best practices that are being launched, you know, all the time and customers requirements about what they want to do in the cloud is also changing all the time.

So at the end of the day it all boils down to what are the super simple things that you can do. to be most efficient in executing and delivering the the value that you're looking to deliver. with the cloud All right.

Well, it sounds like those kiss principles still apply in the ear in the cloud real. Thanks for being on the show. Thanks for having me.

All right and back to you guys in the studio.