Google Compute - Optimization Costs and Performance in Google Cloud Platform

6. Step 06 - Understanding Custom Machine Types

Welcome back. In a step, let's look at another important feature of your compute engine's custom machine types. What do you do when predefined Motions are not appropriate for your workload? You can create a machine type customized to your needs. That's basically a custom machine type. With a custom machine type, you get to choose the number of vCPUs you want and the amount of memory you want. Do you want to add a GPU to your specific machine or not? There are a few restrictions on using custom machine types. It's applicable only to certain types. So E two N, two N, and one are example machine types for which custom machine types are supported. It also supports a wide variety of operating systems, from OSDE and Red Hat, to Windows, etc. Whenever you're using a custom machine type, you have built per vCPU and memory provision. So you're billed based on whatever resources you are provisioning.

An example hourly price is zero zero $3 per V CPU and zero per GB of memory that you provision. Now, where can you configure a custom machine type? Let's go back to the screen where you create a new VM instance, and this is where you can go in and configure your custom machine type. Over here, we first use the Machine family. That's the series. And then we can choose the machine type. In the machine type, you can go in and say "Customer." So this is where you can actually create a custom machine type, as well as configure how many courses you want. You can see that I can say I want two courses, three courses, ten courses, or whatever number of courses that I would want. So you can say how many CPUs you'd want and how many virtual CPUs you'd want. You can also configure the amount of memory you want. You can also increase and decrease the amount of memory you would want with that specific machine type. In this specific type, we talk about custom machine types when existing predefined VM machine types are not efficient for your workloads. In those kinds of situations, you can go online and customise the machine type you would want. You can adjust the number of vCPUs and the GPUs you want, and you can create your own custom machine type. I'll see you in the next step.

7. Step 07 - Exploring GPUs in Google Compute Engine – GCE

Back in a step Let's look at a very important compute engine feature as far as performance is concerned, which is called the GPU, or a graphic processing unit. How do you accelerate math-intensive and graphic-intensive workloads? If you have graphic intensive or math intensive workloads that you are running for AI or ML in those kinds of scenarios, you can add a GPU to your virtual machine. This provides excellent performance for formath and graphics-intensive workloads. However, there is a higher cost attached to it.

And very important to remember is that when you add a GPU,make sure that you have the software that is needed to make use of the GPU installed on your virtual machine. So make sure that you use the GPU libraries that are preinstalled. If you don't do that, your GPU will not be efficiently used. There are a few restrictions when it comes to GPU as well. It is not supported on all machine types. For example, it does not support shared core or memory optimised machine types. Another important restriction that we already talked about is that you cannot do live migration. For instance, with a GPU attached, your onhost maintenance value should be terminated. VM instances only. Typically, the recommended availability policy for GPU is to have automatic restart set to "on". This would ensure that even if there is hardware maintenance done,the instance would be automatically restarted. Now, where can you configure the GPU? Now we are on the screen where we can actually create a VM instance. And over here, there are two ways you can actually create GPU machines. One is that you can choose a GPU machine family,so you can go to the machine family,you can choose GPU, and there are specific machine families that are GPU specific machine families. So you can go and choose them.

The other option is that you can actually choose any other machine family and you can add GPUs to them. GPUs are not available with all machine families. So if you actually select E two, which is the default which is suggested for me, and you expand CPU and GPU,you'd see that the add GPU button is disabled. However, if I go and choose N one, for example, and scrollfurther down, you'd see that I can add GPUs to it. So you need to choose the right machine family, and for those you can explicitly add GPUs as well. In this step, we talked about a very important compute engine feature called GPUs. GPUs are very very important. If you want to speed up workloads that require math and graphics, There are two ways you can use GPUs with your virtual machines. One is to opt for a GPU family, the other is to opt for a normal family and add a GPU. One important thing to ensure is that you are using the right image. You need to use an image with the GPU libraries installed. Otherwise, your GPU will not be efficiently used. I'm sure you're having an interesting time, and I'll see you in the next step.

8. Step 08 - Quick Review - Virtual Machines in Google Cloud Platform

Welcome back. In this step, let's look at some of the important things that you need to read. You need to know about virtual machines. A virtual machine is associated with a project until we have created a number of virtual machines, and by default we are creating them in a project called my first project. It is on Google Cloud. You can use projects to organize your resources. You can create multiple projects, and you can create different kinds of resources in that specific project. So if I'm creating a virtual machine or a database, I would create it within a specific project. Until now we have been creating all our virtual machines in the same project, which is my first project, which is provided by default for us. If you click that, you'd also see that you can create new projects and do things around them. We'll see how to create new projects a little later in the course. For now, the important thing for you to remember is that a virtual machine is associated with a specific project. The other thing you need to remember is that machine type availability can vary from region to region.

Google Cloud offers us a number of regions, and in each of these regions, the available machine types can change. The next important thing to remember is that you can change the machine type of a stopped instance only. So, once you've created a VM, you won't be able to change the number of vCPUs or the memory without stopping it. So I have a running instance right now and if I go in and try to edit this, you'll see that you'll not be able to change the machine type of it. So you can't change the number of vCPUs or memory without stopping the VM instance. Let's go ahead and stop this. Let's go ahead and stop this instance. ead and stIt took a little while, and the instance has now stopped. I can go in now and if I say edit, I can go ahead and change the machine type of that specific instance. As a result, you cannot change the machine type of an existing VM instance. Let's go back and let's start this. Good. I would want to start this up. That's cool. Important thing for you to remember is that if youwant to adjust the number of vCPUs or if youwant to increase the amount of memory which is associatedwith running instance, you need to stop it first andthen you'll be able to make the change. The other interesting thing is that if you have a lot of VMs, you can also filter them using various properties.

Over here we just have one VM, but let's say I had ten or twenty,then filtering would be really, really interesting. I can go into the filter and add a filter based on name, or I can go by zone and say I would only want to look at, let's say, the instances in US Central One A. Or you can add other things, like you can go by machine type, network, network tabs, or you can even go by labels. As a result, attaching labels to your VMs is considered best practice. So if you have a business unit and the label for that business unit is associated with a virtual machine,then you can filter by that specific label as well. You can see that you can also filter by status. So you can, let's say, filter by stopped or only want to see running instances and things like that. If you have a large number of instances, then filtering might be very interesting for you. So you can filter by name, zone, machinetype, internal or external IPS network, and labels. Another important thing to remember is that whenever we create an instance, we are creating the instance in a specific zone. So whenever we create an instance,we choose the region and the zone.

However, whenever we talk about images, images are global. Whenever I create an image, I can use the same image to create an instance in any other region if I want. I can even provide access to other projects to use these images as well. Instance templates are also global by default. So, whenever we create an instance template,we can use it to create resources in multiple regions and multiple zones. If you're making use of a zone-specific resource in your instance template, then that instance template becomes zonal. But by default, instance templates are global. You can use them to create resources in any region. The last important thing to remember is that for a virtual machine,automatic basic monitoring is automatically enabled. I have this specific instance which has been running for a little while now. I've been keeping it running for almost 15 minutes now and I'm going to click here and say "view Monitoring." Let's see if any monitoring details come up. It's saying loading graph data, and you can see a little bit of graph information on the CPU utilization. You can see that I have had this machine up and running for the last 10 minutes or so. If you don't see any metrics here, then make sure that you wait for ten or fifteen minutes before you'll be able to see the metrics here. And over here you can look at the CPU utilization. You can see that the current CPU utilisation is around 12%.

And if you actually scroll down, you can see that you can also see metrics around network bytes: how many bytes are coming in, how many bytes are going out. By default, you will not be able to see the memory utilization. If you want to see memory utilization, you need to actually install something called Cloud Monitoring Agents. All the metrics are provided through a service called Cloud Monitoring. We'll talk about Cloud Monitoring a little later, and then we'll talk a lot more about Cloud Monitoring Agents. You can see that if I wanted to look at disk-based utilization, I would need a Cloud Monitoring Agent as well. If I scroll further down, I can see metrics for network packets, mirror bytes, disclosure, and disc IOPS. So the important thing that you need to remember is that, by default, basic monitoring is automatically enabled. You can see metrics like sleep utilization, network bytes, disc throughput, and IOPS by default. However, if you want memory utilisation and dissipation metrics, then you need to install the Cloud Monitoring Agent. In the following step We will look at a few important things that you need to remember about a virtual machine. I'll see you in the next step.

9. Step 09 - Best Practices - Virtual Machines in Google Cloud Platform

Welcome back. Let's look at some of the important things that you need to read. You need to know about virtual machines. A virtual machine is associated with a project until we have created a number of virtual machines, and by default we are creating them in a project called "my first project. It is on Google Cloud. You can use projects to organize your resources. You can create multiple projects, and you can create different kinds of resources for that specific project. So if I'm creating a virtual machine or a database, I would create it within a specific project. Until now, we have been creating all our virtual machines in the same project, which is my first project, which is provided by default for us. If you click that, you'd also see that you can create new projects and do things around them. We'll see how to create new projects a little later in the course.

For now, the important thing for you to remember is that a virtual machine is associated with a specific project. The other thing you need to remember is that machine type availability can vary from region to region. Google Cloud offers us a number of regions, and in each of these regions, the available machine types can change. The next important thing to remember is that you can change the machine type of a stopped instance only. So, once you've created a VM, you won't be able to change the number of vCPUs or the memory without stopping it. So I have a running instance right now, and if I go in and try to edit this, you'll see that you'll not be able to change the machine type of it. So you can't change the number of vCPUs or memory without stopping the VM instance. Let's go ahead and stop this. Let's go ahead and stop this instance. St and ead It took a little while, and the instance has now stopped. I can go in now and if I say edit, I can go ahead and change the machine type of that specific instance. As a result, you cannot change the machine type of an existing VM instance. Let's go back and let's start this. Good. I would want to start this up. That's cool. The important thing for you to remember is that if you want to adjust the number of vCPUs or if you want to increase the amount of memory that is associated with a running instance, you need to stop it first. You'll be able to make the change. The other interesting thing is that if you have a lot of VMs, you can also filter them using various properties.

Over here we just have one VM, but let's say I had ten or twenty,then filtering would be really, really interesting. I can go into the filter and add a filter based on name, or I can go by zone and say I would only want to look at, let's say, the instances in US Central One A. Or you can add other things, like you can go by machine type, network, network tabs, or you can even go by labels. As a result, attaching labels to your VMs is considered best practice. So if you have a business unit and the label for that business unit is associated with a virtual machine,then you can filter by that specific label as well. You can see that you can also filter by status. So you can, let's say, filter by stopped or only want to see running instances and things like that. If you have a large number of instances, then filtering might be very interesting for you. So you can filter by name, zone, machinetype, internal or external IPS network, and labels. Another important thing to remember is that whenever we create an instance, we are creating the instance in a specific zone. So whenever we create an instance,we choose the region and the zone. However, whenever we talk about images, images are global. Whenever I create an image, I can use the same image to create an instance in any other region if I want. I can even provide access to other projects to use these images as well. Instance templates are also global by default. So, whenever we create an instance template,we can use it to create resources in multiple regions and multiple zones.

If you're making use of a zone-specific resource in your instance template, then that instance template becomes zonal. But by default, instance templates are global. You can use them to create resources in any region. The last important thing to remember is that for a virtual machine,automatic basic monitoring is automatically enabled. I have this specific instance which has been running for a little while now. I've been keeping it running for almost 15 minutes now, and I'm going to click here and say "view Monitoring." Let's see if any monitoring details come up. It's saying loading graph data, and you can see a little bit of graph information on the CPU utilization. You can see that I have had this machine up and running for the last 10 minutes or so. If you don't see any metrics here, then make sure that you wait for ten or fifteen minutes before you'll be able to see them here. And over here you can look at the CPU utilization. You can see that the current CPU utilisation is around 12%. And if you actually scroll down, you can see that you can also see metrics around network bytes: how many bytes are coming in, how many bytes are going out.

By default, you will not be able to see the memory utilization. If you want to see memory utilization, you need to actually install something called Cloud Monitoring Agents. All the metrics are provided through a service called Cloud Monitoring. We'll talk about cloud monitoring a little later, and then we'll talk a lot more about cloud monitoring agents. You can see that if I wanted to look at disk-based utilization, I would need a Cloud Monitoring Agent as well. If I scroll further down, I can see metrics for network packets, mirror bytes, disclosure, and disc IOPS. So the important thing that you need to remember is that, by default, basic monitoring is automatically enabled. You can see metrics like sleep utilization, network bytes, disc throughput, and IOPS by default. However, if you want memory utilisation and dissipation metrics, then you need to install the Cloud Monitoring Agent. In the following step, we will look at a few important things that you need to remember about virtual machines. I'll see you in the next step.

10. Step 10 - Scenarios - Virtual Machines in Google Cloud Platform

Back on the step Let's talk about best practices when it comes to virtual machines. One of the important decisions that you need to make when you create a virtual machine is to choose the right zone and the right region. You need to choose the zone and the region based on the cost, regulations, availability needs, latency, and specific hardware needs. The cost of your virtual machine will slightly vary depending on the zone and the region as well. Also, different machine types might or might not be available in specific zones. You need to also take your users into consideration. Where are your users situated? You want to create your virtual machines as close to your users as possible. If you need high availability, then you'd want to distribute your virtual machines across multiple zones and multiple regions.

As a result, it is best to distribute instances across multiple zones and regions. For high availability, the next important thing is to choose the right machine type for your needs. If you have computer intensive workloads, then go for CPU machine types. If your workloads need a high amount of RAM, then go for memory-optimized instances. So, depending on your workload, choose the right machine type. Sometimes, identifying the right machine type can be difficult. In those kinds of situations, you can try running your workloads on different types of machines to see which one gives you optimal performance. If you have math and graphic-intensive applications, make sure that you are making use of GPUs. The other important thing is to take care of the cost. So, if you have constant workloads that are always running, such as a web application that is always running. In those kinds of situations, you can go for committed use discounts. If you have fault-tolerant workloads that are not critical, which are not time-critical In those kinds of situations, you can even go for preemptible instances. This can reduce your costs by up to 80%.

The other best practise is to use labels on your virtual machine. Whenever you create a virtual machine,add labels indicating which environment, which team, and which business unit. You'll be able to easily search for them using labels, and labels will also be useful in generating reports around your virtual machines. In this step, we talked about some of the best practises around virtual machines. In future sections, we'll talk about creating groups of virtual machines using instancegroups and also load balancing between them. Using load balancers in those sections, we will get back to the topics of availability, regions, and zones. Again, I'm sure you're having a wonderful time, and I'll see you on the next step.

11. Step 11 - Quick Review - Google Compute Engine

Back to this Let's look at the few scenarios related to the Compute Engine. Let's get started with the first one. What are the prerequisites to being able to create a VM instance? Until in this course, we are making use of the default project which is present here. We made use of the first project. This is created by default first. A little later, we'll also see how you can actually create new projects. The second important thing is a billing account. Whenever you're provisioning resources in the cloud,you need to pay for them. And in GCP, you pay for your resources by using a billing account. So you have a project and the project should be associated with a billing account. Earlier, when we were creating a Google Cloud platform account, we provided our credit card details, and those details were used to create a billing account for us. If you type in "billing account" in here, you can go over to the billing account which is present already.

So you'd see that there is a default billing account which has already been created for us. It's called the "My Billing Account." And you can see the overview of the billing account and also the payment overview here. If you are on a free trial, your balance will most likely be zero. And you can also see the card you have configured here. The third important thing is that Compute EngineAPIs should be enabled whenever you're making use of a Google Cloud service. The APIs for them should be enabled before we create resources using that specific service. If I'm creating a Compute Engine virtual machine, the Compute Engine APIs should be enabled. If I type in Compute Engine APIs, you can see this is the Compute Engine API which comes up if I go in there. We already enabled this when we were creating the first virtual machine. However, if you go to a new project, if you create a new project and go to it, you'd see that the API would not be enabled by default. Before we create the virtual machine, you need to go and enable it for this specific project. We have already enabled it. So you can go in and say "Manage."

And over here you can see the details of the specific API. You can see that this is the Compute Engine API and there is a specific service name that is present for this. This is Compute Google Api.com and it is used to create and run virtual machines on the Google Cloud platform. That's what we have been doing. And you can see the usage of that specific API here. You can see that I started recording this course a week back, and you can see all the information about that here. So the three prerequisites to being able to create any resource are: you need a project; you need a billing account so that you can build for your resources. Number three is APIs. For that specific resource, I'm going to create a VM instance here, sothe Compute Engine API should be enabled. Next, you want dedicated hardware for your compliance, licensing, and management needs. Typically, whenever we create a virtual machine, it is created on shared hardware. So the host on which your virtual machine is created might be shared by multiple customers. In certain scenarios, you might dedicate hardware to that specific hardware. You only want virtual machines for your enterprise to be present. How do you ask for it? The way you can ask for it is by going in as a sole tenant. So if you just search for "solve lops," it is not bringing up anything.

Let's go to the compute engine. So let's go back to the compute engine. And over here you can see Sol tenant nodes. Sole tenant nodes This is where you can come in and you can actually create a sole tenant node. The first thing that you need to do is to create a node group. So you can create a sole tenant node group. And over here, you can give this node group a name. You can specify. I'll choose one of the regions that is present here. We are not going to create this notebook anyway, so it doesn't really matter. You can choose any one of these regions which are present there. The next thing that you need to configure is the note template. We don't really have a note template yet, so you can go in and say "Create No Template" and create a note template here. In the Note template is where you specify what kind of node you want. So I can choose. Is it n one or n two? For example, I can choose if I want a local SSD. I can choose if I want a GPU accelerator. The important part here is the affinity labels.

Whenever you create a new virtual machine instance, you would want it to be deployed to this specific node group, so you would need to assign it a write then label. So you can say Add finite label and say this is group eight, and you can say Create. When I try to create a nodetemplate, I get an error saying the theodefenk labels group violates the format constraints. It should only have lower case characters. So let's go ahead and make this group. The same is the case with values. So I'll go ahead and change it to A as well. So group A and say, "Create." Now this would create our note template, and we can use that note template in our node group to create a note group. Once you create a note template, you can say "Continue." And if you want, you can add in auto scaling and specify the number of nodes that you would want. For now, let's say off. Let's just say one note. We are not going to do this because these are really, really expensive. Once you configure auto scaling, you can configure maintenance settings. I'm not going to create this alternate node. What I would do is cancel. Now, how can you create virtual machines for that specific notebook? The way you can do that is by going to VM instances This is exactly what we usually do to create virtual machine instances. However, the only difference over here would be to go to Management Security, this networking and sold tenancy, and go over to sole tenancy.

And over here you can specify the affinity label. And over here, in the known affinity label, is the place where you can specify the group A. So over here you can see the key operator value. So we can say group in a similar way to this. It says workload in the front end, so workload in the front end. Or you can specify a group and say the value of the group is A. So that's how you can specify the nodefinity label. What would happen now is if you have the node with this specific FNC label created already, then the VMinstance will be created on that specific node. So, if you want dedicated hardware for your compliance, licensing, and management needs, you can go for sole tenant nodes. Next scenario I have thousands of VMs. I want to automate OS patch management,OS inventory management, and OS configuration management. For example, I would want to manage software which is installed on thousands of VMs. How can I do that? If I go back to Compute Engine and minimise virtual machine storage instance groups, this is where you can see VM Manager. This is where you can actually do patch management. So, as you can see here, welcome to VM Manager. VM Manager is a suite of tools that can be used to manage operating systems for large VM fleets. So, if you want to manage thousands of VMs and automate patch management, the option to use is VM Manager. Let's say you want to log into your specificVM instance and you want to install some software, for example, Python or Tomcat web server or Apache web server. How do you do that?

The way you can do that is to SSH into it. We did this earlier in the course. You don't want to expose a VM to the Internet. How can we do that? A little later, when we talk about VPCs, we'll talk about firewall rules and different kinds of things. But for now, the simplest option is not to assign an external IP address. Once a VM instance does not have an external IP address, it will not be accessible from the Internet. You don't have HTTP traffic to your VM. How can you do that? You already looked at it. You can configure firewall rules. Where can we configure the firewall rules? You can go to virtual machine VM instances, and when we create the virtual machine, we can actually say of HTTP traffic or Elevate press traffic. So if your VM instance is a Web Server and you don't have LTP or HTTPS traffic, this is where you can actually configure the firewall rules. In this app, we look at some of the important scenarios regarding the Compute Engine. I'm sure you're having a wonderful time, and I'll see you later.

Prepared by Top Experts, the top IT Trainers ensure that when it comes to your IT exam prep and you can count on ExamSnap Associate Cloud Engineer certification video training course that goes in line with the corresponding Google Associate Cloud Engineer exam dumps, study guide, and practice test questions & answers.