7 Cloud Compute Infrastructure

Search in book...
Toggle Font Controls
Create new playlist

Name your new playlist

Playlist description (optional)
Sign In

Email address

Password

Forgot Password?

or

Continue with Facebook

Continue with Google
Sign Up

Full Name

Email address

Confirm Email Address

Password

or

Continue with Facebook

Continue with Google

Chapter 7 Cloud Compute Infrastructure

Images

CERTIFICATION OBJECTIVES

7.01 Virtualization in the Cloud

7.02 Cloud Virtual Machine Components

7.03 High-Performance Computing

Two-Minute Drill

Q&A Self Test

Virtual machines are the workhorses of cloud computing. In this chapter you will learn about cloud-based compute resources such as virtual machines. Clusters (collections of virtual machine nodes) also fall under the umbrella of cloud compute services.

We will start with a discussion of how cloud virtualization is possible through hypervisors in cloud service provider data centers, followed by a brief overview of how to plan virtual machine sizing to accommodate VM workloads.

Next, we’ll identify specific virtual machine components, such as operating system images, virtual network interfaces, autoscaling, and custom routes. We will also cover remote management of both Linux and Windows cloud VMs in the context of IP addressing and jump boxes.

Finally, we will discuss when and how clusters of parallel processing nodes should be used.

CERTIFICATION OBJECTIVE 7.01

Virtualization in the Cloud

As you will recall from Chapter 1, hypervisors are the physical servers that host guests. Cloud service provider data centers contain vast numbers of Type 1 hypervisors such as VMware ESXi, Microsoft Hyper-V, or the Linux-based Kernel Virtual Machine (KVM). Type 1 hypervisors are also called bare metal hypervisors because they run directly on the physical hardware and do not depend on an existing operating system. Type 2 hypervisors run as an application within an existing operating system.

Cloud compute resources—such as virtual machines—run on these hypervisors. Planning cloud virtual machine deployment and usage depends on the workload that will be running in virtual machines.

With both on-premises and cloud computing environments, you will read literature and hear technicians use the term hypervisor to refer to the physical server. Technically, the hypervisor is the operating system software; the physical server could be used for non-hypervisor purposes.

Some cloud service providers offer other types of virtualization such as virtual desktop infrastructure (VDI) solutions, which serve up user desktops from a centralized virtualization server; users can connect remotely to access their cloud-stored desktop and applications.

Virtual Machine Sizing

Running a small website with a small number of users on a virtual machine won’t require nearly as much horsepower as running complex data analyses on large datasets. The virtual machine size, sometimes called the instance type, determines virtual machine horsepower. Common VM sizing constituents include

Number of vCPUs

Amount of RAM

Number of supported data disks

Disk IOPS throughput measurement

Number of supported network interfaces

Network performance

You might choose a specialized virtual machine sizing that addresses specific needs such as a graphics-intensive application. Just bear in mind that more VM compute power and speed mean higher cost. Another consideration is that some VM sizes might be available only in some geographical regions, depending on the CSP.

Virtual machine sizing not only should be addressed during planning and deploying but should also be monitored over time to ensure the best fit (one size does not fit all!). You can change virtual machine sizing, as shown in Figure 7-1, even after the VM already exists, though the VM might require a restart to pick up the new setting.

FIGURE 7-1 Selecting the Amazon Web Services virtual machine instance type

Images

While some exam questions might refer to virtual machine sizing, you could see questions refer to instance sizing. Both in the exam and the real world with cloud computing, virtual machine and instance are synonymous.

You should configure alerts so that you are notified when virtual machines are overly busy or idle to optimize the sizing and, by extension, related costs.

CERTIFICATION OBJECTIVE 7.02

Cloud Virtual Machine Components

Much like a physical computer consists of many components working together, virtual machines in the cloud also consist of many virtual components. The workload that you plan to run in the virtual machine dictates which virtual components you should use and how you should configure them, as described in this section.

Operating System Selection

Cloud virtual machines use an operating system image. In addition to containing the operating system (OS), the image might have specific OS settings, as well as additional software such as a MySQL database server and support for the Python programming language.

As your cloud use increases, you will most likely find yourself creating custom images that will be used with future cloud VM deployments. Figure 7-2 shows how you can select an OS image during deployment in Microsoft Azure.

FIGURE 7-2 Selecting the Microsoft Azure virtual machine image

Images

Connecting to cloud-based Windows virtual machines for management purposes is done differently than for cloud-based Linux virtual machines. Windows and Linux authentication options also differ, but some cloud-based Windows and Linux configurations, such as VM autoscaling, are not specific to the selected OS.

Remote Management

Whether you deploy a Windows VM or a Linux VM, at some point you may need to manage the OS remotely. If you plan on using an on-premises device to remotely manage cloud VMs, you’ll need a way to contact those VMs either directly through a VM public IP address assigned to each VM (not recommended for security exposure reasons) or through a single VM configured with a public IP address (a jump box), as depicted in Figure 7-3. After you establish a connection to the jump box, you can connect to the internal private IP addresses of cloud VMs.

FIGURE 7-3 Cloud jump box configuration

Images

Windows remote management using Remote Desktop Protocol (RDP) occurs over TCP port 3389. Linux remote management using Secure Shell (SSH) occurs over TCP port 22. Your on-premises firewall needs to allow this type of outbound traffic, and cloud firewalls also need to allow this traffic to the VMs that will be remotely managed.

Linux VM Authentication

Where cloud-based Windows virtual machine authentication is most commonly configured to use usernames and passwords, Linux virtual machines can use one of two authentication methods:

Username and password

SSH public key authentication

Authenticating to Linux through a username and password is single-factor authentication (something you know) and not considered as secure as SSH public key authentication. Public key authentication uses a public and private key pair. Both keys are mathematically related. The public key is stored with the Linux host in the cloud, and only the authorized user (or device) has access to the related private key, which should be backed up to a safe location. Depending on the configuration, authentication might only require possession of the private key, or it might also require a password and a username. Figure 7-4 shows the SSH public key authentication settings when deploying a Linux VM in Azure.

FIGURE 7-4 Linux SSH public key configuration in Microsoft Azure

Images

Virtual Machine Autoscaling

You can deploy sets of VMs together for horizontal scaling. Horizontal scaling comes in two forms:

Scaling out Adding virtual machine nodes

Scaling in Removing virtual machine nodes

When an application is in high demand, virtual machines can be added to handle the workload. When things quiet down, VMs can be removed since they are no longer needed—this also reduces unnecessary cloud costs. Figure 7-5 shows a sample autoscaling configuration.

FIGURE 7-5 Enabling autoscaling for a Microsoft Azure virtual machine scale set

Images

Virtual Network Interface Cards

To communicate on a cloud virtual network, a virtual machine must be associated with a network interface, which is its own cloud resource, as shown in Figure 7-6. This means you can associate and dissociate network interfaces to and from different VMs over time, as needed.

FIGURE 7-6 Microsoft Azure network interfaces

Images

Each network interface has settings to configure details such as the following, as shown in Figure 7-7:

FIGURE 7-7 Microsoft Azure virtual network interface settings

Images

Dynamic or static (unchanging) IP address

DNS name resolution

Firewall rule settings to control network traffic into and out of the network interface

IP forwarding (routing)

Images

Think beyond the straight facts presented in this book. You might get exam questions related to multiple network interfaces associated with a single VM—this is a valid configuration and really no different than using multiple NICs in a physical machine, such as when it is configured as a routing or firewall appliance. In the same way, a single network interface can have multiple IP configurations.

Routing Table Entries

If you need to control network traffic flow, you can configure custom routing table entries. Why would you do this? Normally, you would do so to ensure network traffic is forwarded to a firewall or antimalware appliance of some kind for inspection.

When configuring custom routes, you specify the target network range that must be matched in a network transmission, such as 10.244.0.0/24 in the example shown in Figure 7-8. Any traffic that needs to get to the 10.244.0.0 network will be sent to the next-hop virtual appliance, 10.240.0.4. The emerging pattern is that network settings that were normally configured directly within the OS are no longer configured there; instead, the settings reside in the cloud, and the virtual machine OS consults the cloud network settings.

FIGURE 7-8 Microsoft Azure custom route

Images

CERTIFICATION OBJECTIVE 7.03

High-Performance Computing

In a scenario where individual VMs don’t have the power or ability to run complex tasks that need to be accomplished, such as in medical research or financial modeling, consider grouping high-performance VMs together. The purpose of high-performance computing (HPC) is to speed up complex data analysis of large datasets through a cluster of VM nodes. This is also referred to as big compute or parallel computing—running tasks in parallel simultaneously across multiple VMs.

Each HPC VM uses the fastest vCPU configuration, SSD-based storage (as opposed to the older and slower HDD-based storage), large amounts of RAM, and the fastest network throughput. The HPC cluster needs a way to coordinate tasks among the worker nodes. You can get a sense of the HPC layout in Figure 7-9.

FIGURE 7-9 High-performance computing

Images

HPC clusters can be configured on premises or in the cloud. In the cloud, HPC is a managed service. This means that the complexity of installing and configuring operating system and clustering software to work together is hidden from the cloud customer. Monitoring the performance of cloud-based HPC ensures that complex tasks are processed efficiently with as little cost as possible. An example of monitoring in an AWS cloud environment would be to configure alert notifications through the AWS CloudWatch service. Figure 7-10 shows how an HPC head node can be deployed in Azure by selecting the appropriate OS image.

FIGURE 7-10 Deploying a Microsoft Azure HPC head node

Images

EXERCISE 7-1

Deploy a Windows-Based Virtual Machine in the Cloud

In this exercise, you will create a Windows virtual machine in the Microsoft Azure cloud. This exercise depends on having completed Exercise 1-1.

1. Use your web browser to sign in to the Microsoft Azure portal at https://portal.azure.com.

2. At the top of the navigation pane on the left, click Create a Resource.

3. Click the Compute category and select Virtual Machine. This will launch the Create a virtual machine wizard and place you on the Basics screen.

4. Select an existing resource group or create a new one.

5. Use the following values to deploy the Windows virtual machine:

Name: WinSrv1

Image: Windows Server 2019 Datacenter

Username: admineast1

Password: Enter a password that satisfies requirements as per the web page Note that you will need this password in Exercise 7-2.

Public inbound ports: Allow selected ports, RDP (3389)

Opening RDP ports directly to the Internet poses a security risk for important VMs, but it serves the purpose of demonstration in this exercise. In the workplace, use a hardened jump box to allow remote access to cloud VMs that are configured with only private IP addresses.

6. Click the Next button at the bottom of the screen repeatedly until you arrive at the Management screen of the wizard. Ensure the Boot Diagnostics option is set to Off.

7. Click Review + Create, then click Create.

EXERCISE 7-2

Remotely Manage a Cloud Windows Virtual Machine

In this exercise, you will connect to an existing Windows virtual machine in the Microsoft Azure cloud. This exercise depends on having completed Exercise 7-1.

1. If you are not already in the Microsoft Azure portal, use your web browser to sign in to it at https://portal.azure.com.

2. In the left-hand navigator, click Virtual Machines.

3. From the list, click WinSrv1.

4. On the Overview page, look to the right side of the screen and copy the IP address of the virtual machine displayed in the Public IP Address field, as shown in Figure 7-11.

FIGURE 7-11 Microsoft Azure virtual machine public IP address

Images

5. On your local Windows computer, open the Start menu and launch the Remote Desktop Connection app.

6. Paste the previously copied public IP address into the Computer field.

7. Click Show Options in the lower-left corner and enter admineast1 into the User Name field, then click Connect.

8. When prompted, enter the password configured for admineast1. (This password was established in Exercise 7-1.)

9. If prompted, choose to trust the connection.

10. You are now remotely connected to a Windows VM running in the Microsoft Azure cloud.

11. Move the upper-right corner of the screen and click the X to close the remote desktop connection.

12. In the Azure portal, select WinSrv1. Click the Stop button to prevent unnecessary cloud charges.

INSIDE THE EXAM

Templates and Automation

The CompTIA Cloud Essentials+ CLO-002 exam could present you with questions regarding automation and repetition of cloud administrative tasks. As an example, when manually deploying virtual machines in Microsoft Azure, the last page of the wizard provides an option to save the settings as a template for future deployments. Templates can also be created manually or acquired through a variety of websites, such as https://github.com.

CERTIFICATION SUMMARY

This chapter discussed compute infrastructure in a cloud environment. You learned how hypervisors make operating system virtualization possible and how you should select the appropriate virtual machine sizing based on VM workloads.

You have been exposed to how both Windows and Linux virtual machine images can be deployed and how to manage them over RDP and SSH, respectively, through public IP addresses or a jump box. Cloud-based virtual machines are related to other cloud resources such as virtual network interfaces and custom route tables.

You also learned how autoscaling addresses application performance by scaling out (adding VMs) when the application is busy and scaling in (removing VMs) when demand for the application quiets down.

Finally, you learned how high-performance computing (HPC) can be used to perform large-scale, complex computing tasks across a cluster of VM nodes working together in parallel.