Oracle SuperCluster is an integrated server, storage, networking, and software platform that is typically used either for full stack application deployments or consolidation of applications or databases. Because it incorporates Oracle’s unique and innovative Exadata Storage, Oracle SuperCluster delivers unrivaled database performance. And the platform also hosts the huge range of Oracle and third-party applications supported on Oracle’s proven, robust, and secure Oracle Solaris operating environment.
Virtualization is a particular strength of Oracle SuperCluster, with Oracle VM Server for SPARC serving up high performance virtual machines with zero or near zero virtualization overhead. These virtual machines are known as I/O domains. Further, an additional layer of highly optimized nested virtualization is offered in the form of Oracle Solaris Zones. All of these virtualization capabilities come at no additional license cost. For more information about virtualization on Oracle SuperCluster, refer to the recent blog Is "Zero-Overhead Virtualization" Just Hype?
The platform also utilizes a built in high throughput, low latency InfiniBand fabric for extreme network efficiency within the rack. As a result, Oracle SuperCluster customers enjoy outstanding end-to-end database and application performance, along with the simplicity and supportability featured on all of Oracle’s engineered systems.
Can these benefits be realized in a cloud environment, though? Oracle SuperCluster is not available in Oracle’s Cloud Infrastructure, but private cloud deployments have been implemented by a number of Oracle SuperCluster customers, and Oracle Managed Cloud Services also hosts many Oracle SuperCluster racks in their data centers worldwide. In this blog we will consider the building blocks provided by Oracle to simplify deployments of this type on Oracle SuperCluster.
The virtual machine lifecycle on Oracle SuperCluster is orchestrated by the SuperCluster Virtual Assistant (SVA), a browser-based tool that supports the creation, modification, and deletion of domain-based virtual machines, known as I/O domains. Functionality has progressively been added to this tool over the years, and it has now become a single solution for dynamically deploying and managing virtual machines on SuperCluster, including both I/O domains and database-oriented Oracle Solaris Zones. SVA is a robust tool that is widely used by SuperCluster customers across a range of different environments.
The current SuperCluster Virtual Assistant v2.6 release offers a set of capabilities that deliver benefits and features consistent with those outlined above in the IaaS Introduction. As an alternative to SVA’s intuitive browser user interface, SVA’s IaaS functionality on Oracle SuperCluster can be managed from other orchestration software using the provided REST interfaces. SVA REST APIs are self-documenting and therefore easier to consume, thanks to the included Swagger UI.
The following screenshot shows an initial window from the tool listing I/O domains in a range of different states.
Both physical domains and I/O domains (virtual machines) are managed, along with their component resources. New I/O domains can be created, and existing I/O domains modified or deleted, with additional cores and memory able to be added dynamically to live I/O domains. Database Zones based on Oracle Solaris can also be managed from the tool, and a future SVA release will allow Oracle Solaris Zones of all types to be managed. I/O domains can be frozen at any time to release their resources, and thawed (reactivated) whenever required. As well as providing a cold migration capability, the freeze/thaw capability allows resources used by non-critical I/O domains to be temporarily freed during peak periods for use by other mission critical applications.
Resources are assigned automatically from component pools that manage CPU, memory, network interfaces, IP addresses, and storage resources. VLANs and other network properties can be pre-defined, allowing access to DNS, NTP, and other services. An integrated resource allocation engine ensures that cores, memory, and network interfaces are optimally assigned for performance and effectiveness. Compute resources are allocated to I/O domains at a granularity of one core and 16GB of memory, or using pre-defined recipes. Network recipes can also be set up to simplify the allocation of network resources, including simultaneous redundant connectivity to different physical networks thanks to quad-port 10GbE adapters. Recipes are illustrated in the screenshot below.
A number of SVA policies can be set according to customer requirements. One set of policies relates to users. User roles are supported, allowing both privileged and non-privileged users to be created. A single SVA user can consume all resources. Alternatively, multiple SVA users can be created, with resource usage tracked by user. Resources can be unconstrained, allowing a user to consume any available resource, or limits can be set, to ensure that no user consumes more than a pre-defined allowance.
The screenshot below illustrates an early step in the process of creating an I/O domain.
A comprehensive Health Monitor examines the state of SVA services to ensure that the tool and its resources remain in a consistent and healthy state.
SVA functionality continues to be extended, with a number of new features currently under development. Oracle SuperCluster M8 and Oracle SuperCluster M7 customers are typically able to leverage new features simply by installing the latest quarterly patch bundle, which also upgrades the SVA version.
Oracle SuperCluster customers can realize cloud benefits in their own data centers, taking advantage of improved time to value, greater simplicity, and better scalability, thanks to the Infrastructure-as-a-Service capabilities provided by the SuperCluster Virtual Assistant. Database-as-a-Service (DBaaS) capabilities can also be instantiated on Oracle SuperCluster using Oracle Enterprise Manager.
The end result is that Oracle SuperCluster combines the proven benefits of Oracle engineered systems with IaaS and DBaaS capabilities, allowing customers to reduce complexity and increase return on investment.
Allan Packer is a Senior Principal Software Engineer working for the Solaris Systems Engineering organization in the Operating Systems and Virtualization Engineering group at Oracle. He has worked on issues related to server systems performance, sizing, availability, and resource management, developed performance and regression testing tools, published several TPC industry-standard benchmarks as technical lead, and developed a systems/database training curriculum. He has published articles in industry magazines, presented at international industry conferences, and his book "Configuring and Tuning Databases on the Solaris Platform" was published by Sun Press in December 2001. Allan is currently the technical lead and architect for Oracle SuperCluster.