From the course: Oracle Cloud Data Management Foundations Workshop

Exadata deployment options

(note plays) - Welcome friends to the Oracle University Module Exadata Cloud Service. I'm your host, Alex. Let's get started. Today we're going to take a look at Exadata Cloud Service in more specifically Exadata Cloud service X9M which is the latest release of Oracle's premier database cloud service. Over the next few minutes, we will take a look at this service and highlight some of the new features in this latest release that further cement Exadata Cloud Service as the best way to run your databases in the cloud. Before we get started, let's take a look at the Exadata platform. Exadata is the best platform for running the Oracle database. It provides superior performance, scalability, security, and availability at the lowest cost. Deep engineering integrates the hardware and software to create this optimized database platform. We use the best hardware available on the market, but what makes Exadata special is a smart database aware system software. This software implements specialized algorithms to improve all types of workloads, including OLTP analytics, batch processing, and of course consolidation. We then combine that with smart management tools to make Exadata easy to manage and to ensure that it is always deployed and configured consistently with best practices. Lastly, note that Exadata is available both on premise and in the cloud. It's Exadata deployment in Oracle Cloud known as Exadata Cloud service that we will focus on today. So what is Exadata Cloud service? Well, it's exactly what it sounds like. It's the Oracle database running as a service on an Exadata in our Oracle Cloud infrastructure. It includes all the advanced database features and options available for the Oracle database. You can license everything as a service or bring your on-premise licenses to the cloud. Since it runs the same Oracle database you run on premise it is a hundred percent compatible with any existing applications. For all practical purposes, it looks exactly like an Oracle database running on an Exadata on premise. The big difference is the public cloud features and benefits, specifically the simplicity and elasticity. Elastic scaling means you can grow and shrink your service to match your workload requirements. This means you only pay for what you use and this paper use capability is how you can really lower your total cost of ownership. In addition, Oracle will manage all the infrastructure for you so you can focus on your business and don't have to worry about the infrastructure and for those operations you do continue to manage, Exadata Cloud Service provides sophisticated automation, making most operations as easy as pushing a button. This is just a quick summary of some of the capabilities you get with Exadata Cloud Service. On the left are all the features available in the Oracle database. On the right, all the Exadata innovations available with the platform. Everything you expect from Exadata is available with the Exadata Cloud service. Currently, Exadata cloud service is available in multiple shapes. Customers can choose an entry-level based system which is not expandable or subscribed to the Exadata cloud service X8M. The X8M shapes start with a quarter RAC and allow elastic expansion by adding additional database and storage servers to enable higher compute and storage capacity. As you can see from the table, the base system also has only a fraction of the OCPU and memory resources as well as storage capacity that is available in Exadata Cloud service X8M. An Exadata Cloud service X8M benefits from the performance improvements associated with Persistent Memory, PMEM or RoCE. More on that in a minute. So what is new with Exadata Cloud Service X8M? First of all, we've adopted the hardware innovations of Exadata X8M, specifically the persistent memory or PMEM and a new secure network fabric based on RDMA or remote direct memory access over converged ethernet known as RoCE. As we will see in a minute, this gives Exadata cloud service some great performance benefits but perhaps even more interesting, this new Exadata cloud service supports elastic configurations just like it's on on-premise cousin. This allows you to build arbitrary configurations, independently adding database servers and storage servers. You configure only what you need which means you pay for only what you need. We also switched from XCN virtualization to KVM virtualization, which means Exadata Cloud Service is now using the same virtualization technology as the rest of Oracle Cloud. One of the advanced features supported on Exadata Cloud service is Oracle Real Application Clusters also known as Oracle RAC. RAC allows you to scale database instances across multiple database servers for greater performance and higher availability. With Exadata Cloud service, you can provision Oracle RAC with a single click using Oracle Cloud infrastructure automation. Second, we use remote direct memory access or RDMA to accelerate OLTP messaging along with SQL reads and writes, more on that in a minute. Third, one of the really great features of Exadata in general is what we call Smart Scan, which offloads basic operations like SQL queries, analytics processing, and machine learning algorithms from database server cores to the CPU cores and intelligent storage servers. This has three major effects. It accelerates runtimes, it frees up database server cores to work on other processing, and it reduces costs particularly on Exadata cloud service because the CPU cores and intelligence storage servers do not require Oracle database licenses so you get more work done per database or CPU you enable. And finally, Exadata Cloud service offers fully active remote database copies that are typically created for disaster recovery purposes but can also be used for secondary processing like reporting, QA, or development and testing. I mentioned RDMA earlier and promised that we would come back to it, so here it is. RDMA allows the database CPU cores to read data from and write data directly into the memory of intelligent storage servers without involving the operating system or networking stack overhead. This combines with the extremely low latency of persistent memory located in intelligent storage servers to lower both SQL read and write latencies, which are two of the most critical factors impacting OLTP performance on a database system. With the combination of RDMA, fast networking, and persistent memory, SQL right latencies are reduced by up to a factor of eight x, read latencies are 10 x better as though as 19 microseconds and overall transaction processing IOPS are increased by up to 2.5 x. Effectively, this makes Exadata Cloud service X8M the best place for OLTP in the cloud with the lowest latencies, highest throughput, and lowest costs. Because your workloads consume fewer resources which is what you're actually paying for. Exadata Cloud Service X8M allows fully elastic expansion. You start out with a small minimum sized configuration with two compute servers and three storage servers that provide full high availability against compute, storage and networking component failure, and allows you to use up to a hundred OCPUs and roughly 150 terabytes of storage. These systems are dedicated to you. They aren't shared with other users. Then you can add storage and compute servers as needed. Each compute server adds 50 usable OCPU cores and nearly 1.4 terabytes of DRAM for database operations. Each storage server adds just under 50 terabytes of database storage, 25 terabytes of flash, 1.5 terabyte of PMEM and 48 CPU cores for SQL queries, analytics, et cetera. And without requiring database licensing. Let's look at some scaling examples that can be achieved with Exadata Cloud X8M. You can start out small with a balance configuration for general purpose workloads. You can choose to add a lot of compute with a little storage to create a system that's highly tuned for OLTP or you can dramatically increase storage capacity without growing compute power for petabyte data warehouses. Or you can maintain a balance of compute and storage for consolidating a variety of Oracle database workloads. Ultimately, you could be running thousands of databases in a single service, creating what amounts to a private cloud using public cloud resources and economics. As we discussed a few minutes ago, Exadata is coengineered with the Oracle database, enabling Exadata cloud service to provide the best data management platform in the cloud. On top of that, Exadata cloud service delivers massive scalability built on a maximum availability architecture supporting up to 1,600 OCPUs and databases as large as 25 PD compressed, assuming a 10 to one ratio with HCC. Imagine the cost cutting and efficiencies you could achieve by consolidating databases onto one service. Simply stated, no database is too demanding or workload too large for Exadata Cloud service. Now, I want to quickly provide an overview of the service. In terms of service operation, Oracle manages Exadata infrastructure and as I just mentioned, customers manage everything running in the database VM. As you would expect, both the Exadata hardware and software releases are supported with Exadata cloud service and the service currently supports Oracle database versions 11.2.0.4 to 19c. Exadata Cloud service simplifies lifecycle tasks such as provisioning, scaling, patching, backup, and disaster recovery through cloud automation. And with Exadata cloud service, customers benefit from cloud economics and only pay for what they use. Customers pay for database and storage server infrastructure with a 48 hour minimum. Customers also pay for the OCPUs with per second billing and they have a choice of a license included or bring your own license pricing model. More on that in a minute. Exadata Cloud service provides a wide choice of management interfaces. The easiest to use is the web browser interface which allows you to use a browser to graphically configure and initiate operations. This is great for one-time actions but most customers prefer a more programmatic interface for things done repeatedly such as provisioning and patching databases. Anything you can do with a browser you can also do with a corresponding rest API. Similar to the browser, rest APIs transit the internet via HTTPS and require no special software installed on a local system. All interfaces are also exposed via a command line interface that can be used for scripting and for building custom tooling. There is a software development kit to integrate with common languages such as Java, Python, Ruby, and Go. If you prefer to manage your infrastructure as code, there's also a Terraform interface. Exadata Cloud Service follows a simple cloud management model. As I mentioned, Oracle owns and manages the infrastructure. This includes the database servers, VM hosts, storage servers, and the fabric network. Customers can schedule maintenance windows for Oracle to perform infrastructure maintenance. When customers subscribe to Exadata cloud service, they're responsible for managing everything running in the database VMs. They manage the VMs, grid infrastructure, and the databases using the cloud automation tools I previously mentioned, and of course, customers own everything inside the database. This includes data, schema, and encryption keys. Exadata Cloud service is available in Oracle Cloud infrastructure. Here's the latest map of Oracle Cloud infrastructure regions around the world where Exadata Cloud service is available. Let's take a minute to understand the importance of vertically scaling your virtual machines. Scaling your VMs as your workload scales up and down is important to reducing your costs and is one of the greatest advantages of cloud over on-premise deployments. When you deploy an on-premise system, you must size it to handle the peak workload as an on-premise deployment do not support scaling up and down. In fact, you probably oversized it a little bit as it is difficult to scale an on-premise system and you want to ensure capacity for workload spikes. However, during most normal operations, the system is underutilized and you're paying licensing fees on unused CPUs. Contrast that with a cloud deployment where you can scale the service up and down depending on demand. If you have a workload spike, simply scale up to accommodate it. If demand drops, scale the system down. Unlike most cloud database services, Exadata allows you to scale online with no disruption to the database service. As I mentioned, customers have a choice of license models in the cloud. You can subscribe to our enterprise edition Extreme Performance Service which includes a license to use all Oracle database features and options, as well as all database management packs. It also includes all Exadata software features. This is the best option if you're without existing licenses or have databases with variable workloads that can really benefit from pay for use fee pricing. If you already have all the licenses you need, you can also bring your own licenses to the cloud. Through an inexpensive BYOL platform service, we will provide you entitlements to use transparent data encryption which is mandatory in the cloud, the data masking and subsetting management pack, diagnostics pack, tuning pack, and real application testing. In addition, all Exadata software features are included. Finally, you can switch between the two models at any time as often as you like. This means, for example, that you could use BYOL most of the time and switch over to the licensed service to handle infrequent spikes on workload that require more capacity than you have licensed via BYOL. The last topic we're going to discuss today is security. Identity management network controls and encryption are built into Exadata Cloud service to ensure end-to-end data protection. It should also come as no surprise that Exadata Cloud service supports all Oracle database security options and can be used with Oracle data safe. Thanks for hanging with me for Exadata cloud service. I hope you learned something useful.

Contents