Now we're moving on from object storage, S3 and Glacier, to block storage, the diagram in the middle. Notice it has a grid made up of numbered blocks from 0-19. This is the structure of physical hard disks. Block storage splits up larger things such as operating systems, applications, and files into very small blocks. If you make a small change to an application or a file, block storage just needs to change the few affected blocks, not the entire object. Amazon EBS stands for Elastic Block Storage. It gives you block storage volumes for your EC2 instances, which provide persistent storage for applications running on EC2. The latest generation of EC2 instances allows you to attach up to 28 EBS volumes to a single EC2 instance. However, you can only attach an EBS volume to a single EC2 instance. You can't attach it to multiple EC2 instances at once. Why does AWS use EBS to provide storage for EC2? Well, if you're running an operating system and you'd like to apply an update, you don't need to install the entire operating system all over again. You can just change the few blocks that need to be changed for that update. That's why EC2 uses EBS for storage and why people call EBS the virtual hard disk in the Cloud. Some EC2 instances also come with non-EBS storage because they come with an instance store. Instance store is a type of ephemeral storage. However, you want to attach an EBS volume because instance stores being ephemeral will lose their data when the EC2 volumes stops or terminates. One Amazon EBS customer is Equifax. Equifax is a data aggregator perhaps best known for giving credit scores, which are ratings of creditworthiness for individuals. Equifax does this by running more than 850 virtual CPUs in EC2, which read from and write to over 1000 terabytes of data across 350 EBS volumes. The last type of storage we have to cover is File Storage. File storage protocols turn Block Storage into a hierarchy of folders, sub-folders, and sub-sub-folders that you're probably familiar with using computers. AWS has two different file storage services, EFS and FSx. Amazon EFS, which stands for Elastic File System, gives you Linux file storage that's a great choice for most purposes. EFS is also a network file system, meaning it allows multiple EC2 instances to connect to it and collaborate on the same set of files. One EFS customer is HERE Technologies, which quote, "Powers four out of five in-car navigation systems in North America and Europe." HERE uses Amazon EFS to serve 750,000 downloads and 50,000 uploads each day. Sometimes, however, you don't necessarily want to use a Linux file system, you want to use your specific file system. In which case you should see if that file system is supported by Amazon FSx. At time of recording, they support two file systems, Amazon FSx for Windows File Server, which is a very popular file system, and Amazon FSx for Lustre. Lustre is a high-performance computing file system used as part of machine learning, video rendering, and financial simulations. If you're interested in any of those, check out Amazon FSx for Lustre. Now, let's move from Cloud storage to hybrid storage. Hybrid storage means connecting your on-premises infrastructure and applications to unlimited Cloud storage. Two AWS services that allow you to do this are AWS Storage Gateway and AWS Outpost. AWS Storage Gateway offers file, volume, and tape gateways. You can run AWS Storage Gateway either virtually or using a hardware appliance, which looks like this. Let's briefly look at each type of the three gateways. File gateway saves your on-prem files to S3 or FSx. Volume gateway saves your on-prem server hard disks to EBS snapshots on S3. Tape gateway saves on-premises tape backups to S3. Many organizations still use physical tape to store information. Customers can migrate that physical tape to a virtual tape backup library on S3. A customer for Storage Gateway Tape Gateway is Formula 1, the racing sport that owns an increasing amount of raw video footage, car telemetry, and other data. Before Storage gateway, they kept their physical tapes in an off-site warehouse where retrieval took 5-7 days. Using Storage gateway, they virtualized their tapes and stored them in Amazon S3 Glacier, where they cut retrieval times to 3-5 hours. Meaning they use standard retrieval from Amazon Glacier, not expedited or bulk retrieved. They also eliminated the use of warehouses and office space that was previously dedicated to managing their physical tapes. AWS Outpost is another form of hybrid storage. Outposts are physical servers from AWS that run EC2 and S3 on your on-premises data center, instead of accessing those services virtually using the Internet. By running AWS compute and storage capabilities on Outposts, instead of on the AWS Cloud, you can receive even faster responses and keep your sensitive data on-premises. One possible use case might be real-time medical imaging. Our storage types so far are meant for primary storage, to serve real customers in production environments. Our final storage type is secondary storage, inactive data that backs up primary storage. This is where I work and it's your first line of defense against a ransomware attack. Recently, both Colonial Pipeline and JBS recovered from their ransomware attacks by restoring their backups. The AWS-wide solution for secondary storage is AWS Backup, which you can configure to automatically take snapshots of your resources across AWS, including many of the services we have studied and will study. A snapshot, like a photograph, is just a copy of what your resource looked like at that point in time. You can restore snapshots at anytime. My co-instructor, Nancy Wang, is the General Manager of AWS Backup. If you're interested in working with this service, I included a link to an AWS 10-minute tutorial on performing Amazon EBS backup and restore using AWS Backup. It's linked at the end of this module.
