In this module, you will learn about persistent disks and network interface controllers. Persistent disks resembles iSCSI disks in on-premise. Persistent disks are networked attached block storage. As the name implies, you're data is persistent, meaning that your data is durable and outlives the machines life-cycle. Persistent disks are a zonal resource, and you can even have a dual zone disk for redundancy. The first persistent disks that a virtual machine attaches to is the boot disk, which is where the operating system exists. You can have more than one persistent disk, and there is a direct correlation between disk size and its performance. That means that if you need more speed from your persistent disk, you can scale it's size to match precisely the speed you need. Persistent disks comes in two flavors, HDD and SSD. The choice comes down to cost and performance. HDD is great for long-tail files or general bulk data that does not need fast performance, and it is more economic option per gigabyte. SSD is designed for random reads and writes, and provides better performance for databases. Another feature of persistent disks is that you can dynamically resize them, even while they're running and attached to the VM, and therefore, also benefit from the increase in performance as we saw in the last slide. You can also attach a disk in read-only mode to multiple virtual machines. This allows you to share static data between multiple instances, which is cheaper than replicating your data to unique disks for individual instances. By default, Compute Engine encrypts all data at rest. GCP handles and manages disk encryption for you without any additional actions on your part. However, if you want to control and manage disk encryption yourself, you can either use Cloud key management service to create and manage key encryption keys, or create and manage your own encryption keys as customer supplies encryption. Compute Engine also provides physically attached SSDs, called local SSDs. Because they're locally attached, these disks are considered ephemeral, but provides very high IOPS. Data on this disk will survive a reset, but not a virtual machines stop or terminate because these disks cannot be reattached to a different virtual machine. Currently, you can attach up to eight local SSD disks with 375 gigabytes each, resulting in a total of three terabytes. The persistent disks offers data redundancy because the data on each persistent disk is distributed across several physical disks. We recommend using a persistent HDD disk when you need an economic storage solution and performance requirements are relatively low. If you have a high performance requirements or your workloads rely more heavily on random reads and writes like databases, we recommend SSD options. For non-persistent storage, local SSD provides the highest throughput and lowest latency because they're physically attached to your virtual machine. There are many differences between a physical hard disk from your on-premises environment and a Compute Engine persistent disk, which is essentially a virtual network attached device. First of all, if you remember with normal computer hardware disks, you have to partition them. Essentially, you have a drive and you're carving up a section for the operating system to get its own capacity. If you want to grow it, you have to repartition and if you want to make changes, you might even have to reformat. If you want redundancy, you might create a redundant disk array. If you want encryption, you need to encrypt the files before writing them to the disk. With Cloud persistent disks, things are very different because all that management is handled for you on the back-end. You can simply grow disks and resize the file system because disks are virtual network devices. Redundancy and snapshots services are built in and disks are automatically encrypted. Each compute engine virtual machine comes with a virtual network interface controller or VNIC. The overall network throughput of a virtual machine scales at two gigabits per second per vCPU, up to 32 gigabits per second with 16 vCPU cores. Because persistent disks are accessed over the network instead of physically attached to the virtual machine, they're are also using the allocated network bandwidth a machine has. You can have up to eight network interface controllers, each attached to a different VPC network. For example, if you want to have a network appliance that has one network interface controller in a DMZ VPC and one in your internal VPC. One important aspect to remember is that once a virtual machine is created, you cannot make any modifications to the network interfaces. That means that if you want to change the number of NICs connecting to different networks or Ed NIC, you'll have to recreate the virtual machine.
