Cost management on Microsoft Azure Red Hat OpenShift

Meet the speakers

Roy de Milde (00:00):
Welcome everyone to the Microsoft Azure Red Hat® OpenShift® video series. I'm Roy, I work for Microsoft and I'm part of the Global Black Belt team as an application innovation specialist. And today I'm joined by...

Yuri Titov (00:11):
Yuri Titov. I'm a Senior EMEA Black Belt for Red Hat Managed Services. So today we would be talking about cost management for ARO. A reason for it is quite straightforward, right? Because finance is better. Essentially it should be noted that this topic is not so complicated as it seems quite often, right? It's rather complex. So therefore to highlight, I mean different aspects of it, I would probably, Roy ask you: what tools do we have in Azure for cost management?

Cost management in Azure Red Hat OpenShift

Roy de Milde (00:41):
Sure. So let me start off by saying we have the Azure Price Calculator. And with this Azure Price calculator, you can actually define every single service that you want to use in Azure. You can configure it and, based on your requirements and your configuration, a monthly price comes out of it, right? The cool thing with this Azure Price Calculator is that you can actually log in with your company credentials as well. If you have any discounts or special things in your contract, it will reflect in that Azure Price Calculator. So this is an easy way to get an estimate. Let's take an Azure Red Hat OpenShift cluster. I select the service, I select which work nodes I want to use, which OS disks I want to attach, which master nodes, and this will give me an estimate of what it's gonna cost me on a monthly basis.

(01:27)
Now because it's a hyperscale platform, Azure, a public cloud, what we can actually do is we can do different financial modeling. So one of the things that we do often is the pay-as-you-go model and you basically pay for what you use, right? If you use it for an hour, you pay for an hour. If you use it for eight hours, guess what? You're paying for eight hours. Very flexible, but also the most expensive way of doing this. So we have another financial model called Reserved Instances. You're basically gonna reserve specific capacity. What you're gonna do is you're gonna say, “Look, I will have these masters, I will have these workers, and I'm gonna run this 24/7 for the next year. I'm not gonna scale this down or delete it or whatever.” And if you tell us that you're committed to do that, what we will do is give you an additional discount, right?

(02:16)
It removes the flexibility as pay-as-you-go, but it reduces the cost of the service because you're committed to doing that with us. And you can do this for one year or for three years. Yeah. So what we see a lot of customers do is mix and match those financial models. If you run an Azure Red Hat OpenShift cluster, let's take the bare minimum and you're gonna do that under Reserved Instances to have a little bit of discount, a cheaper price. But now if you need to scale up, because Red Hat OpenShift actually allows you to quickly scale up and have "burstable" scenarios and with compute resource of Azure. Indeed, you can actually add that pay-as-you-go layer and scale it up for a couple of hours, scale it down again, and you will pay that specific part under the pay-as-you-go model.

Simple and flexible payment with Azure Red Hat OpenShift

Yuri Titov (03:01):
Flexible!

Roy de Milde (03:02):
Super flexible! The cool thing is as well, because it's a first-party service, you pay for Azure Red Hat OpenShift directly to Microsoft. It's not that they need to pay you separately and they need to pay us for the resources. No, it's one Azure bill, and Azure Red Hat OpenShift is on top of that as well.

Yuri Titov (03:19):
Single bill.

Roy de Milde (03:21):
Single bill for the customer. Make it as easy as possible. The last thing that we have which I would like to mention is the cost management tooling of Azure. It's built in, in Azure, and this shows some trends or analyzes. You can see the burn rate and actually a really cool tool to look at your Azure services. What are you paying, what are you gonna pay, do some forecasting and etc. Now, I know this is all Azure and specifically on the whole ARO cluster, but there's obviously a little bit more to financial modeling and capacity in that sense. So if you look at the OpenShift part, could you tell me a little bit about that?

Cost management and OpenShift workloads

Yuri Titov (03:58):
Exactly. So after we set capacity and we have our cluster running, right? What we put on top of it is basically model of a particular customer workloads, which could be really many for many tenants, etc. That should show how in timeline cost management actually evolves, right? For different product services, etc., which reflect institutional organization or could be teams, products, and so on and so forth. Let me probably show it on the diagram what tools we have for cost management and ARO.

Roy de Milde (04:30):
Yeah, perfect.

Yuri Titov (04:31):
Okay. So we set it on Azure level for cost management, and now we need to specify also cost model for our workloads, right? Let us take a look what it consists from. Well, when you install cost management operator in an ARO cluster, basically what it starts doing, it starts checking your inventory and corresponding capacity for lots of things, where you basically specify a price. Additionally, it collects also metrics. Where you also specify a price for things like CPU, memory, etc. Additionally, it collects so-called "Azure raw costs". Well, they're not raw, but they’re called raw, so they also go as an input. To answer the question: how this data is collected. Let us take a look at this part. So as I mentioned, we installed cost management operator that, after activation, starts generating CSV files. Yes, comma separated value files, good old, and copy them over encrypted connection over TLS transport into encrypted database. So everything is secure and preserved. Within corresponding levels of security and access to cloud, where the data is aggregated for derived costs and being prepared. But you actually can open those files and take a look into them.

(06:05)
So the most interesting part in your cost model is derived costs. So, as we mentioned, basic costs serve as input into derived costs. Derived costs are essential parts of your cost management, and they could be applied to a project space within an ARO cluster. It could be applied to whole cluster. Maybe you have several clusters and that's why you're interested in this development. Test, production clusters, several production clusters with different workloads, etc. You could also attach and specify derived costs for particular nodes if you have different instances, and also for additional Azure services or subs. So that's why you get a distribution here.

(07:04)
Okay. And the reason you are doing this basically is to get the full picture about cost management that actually creates a transparency, converts your resources consumption on workload level project service level for technical colleagues and for business people, right? Because business matters. And actually there would be some parts that have shared semantics. The tool in cost management of ARO that you could do for that is basically tags. Tags, tags are everything. And this is so simple and flexible too, that you could mark with tags. For example, different tenants here for technical people or a marketing campaign for a particular service. It could be, for example, a particular business unit or region geography. So it could be for example, particular orchestra that is being provisioned. So provisioning and some additional things like groups or users with security. All right? Additionally, last but not least, for example, something like financial statements for particular products or services. And as I said at the end, the key core semantics is to get the shared information for products, services and particular workloads you have on your cost management model.

Red Hat cost management tool

(09:05)
Important to mention is also that as long as you specify this, you start collecting data. You have also your cost model with direct costs. The Red Hat cost management tool also creates a forecast using machine learning mechanisms. 

(09:28)
And that helps, helps you to support, for example, with things like allocating budget. If you need more budget, if consumption goes up. If you need for example, to reduce amount of notes, etc. If consumption goes down and actually you need less budget, which is also an important point for the planning. And especially if you have a multi-tenant cluster, additional important thing is that in your cost model, you could specify different roles for different users, right? So they could see only particular groups for particular projects, services, etc. So this is the role-based access control. And that's essentially how you could form a cost model for your workloads.

Roy de Milde (10:16):
That was super insightful and to be very honest, nice drawing. 

Yuri Titov (10:21):
Oh, thank you! But well, now we said Azure level focused cost management, and also we said cost management for workloads inside of ARO, right? With Red Hat tools. What should we do with the drawing?

Roy de Milde (10:38):
Use it.

Yuri Titov (10:39):
Ah! Okay.

Roy de Milde (10:41):
To be fully honest around that. So I think the first thing is enable this, right? Right. And it's out of the box. It's easy to enable. And I think if you combine both systems, so the tools and capability for measure, and then the tools and capabilities Red Hat and what Red Hat created in Openshift, you get a full idea of the whole cost around the solution.

Yuri Titov (11:02):
Complemented.

Roy de Milde (11:04):
One hundred percent. The cool thing is that if you enable this, you combine the different data points that you have, you can actually be super flexible and adopt it to your financial strategy. Maybe you need to create or talk to billing systems, or maybe chargeback mechanisms. Maybe you wanna do some forecasting on your burn rate for your IT systems and etc. So those capabilities all are at your fingertips when you start to enable it and use it. And do you want to learn more about Azure Red Hat OpenShift? Go check out the other amazing videos that we have created. Or what you can do, is reach out to Red Hat or Microsoft, and we're more than happy to help support you in this journey.