Oracle Grid-based Virtualization Reference Design

The article starts with an introduction to Oracle Grid-based Virtualization followed with a comparison of Oracle Grid-based Virtualization against traditional server virtualization solutions.

Grid computing’ roots can be traced back to the late 1980's (e.g. Condor¹, Load Sharing Facility²), used by academic and research institutions to process numerically intensive jobs. Before grid computing, the ability to process numerically intensive jobs was limited by the processing power of a single machine. By the mid 1990s, (e.g. Globus³) Grid computing allowed academic and research institutions to pool computing resources, which dramatically reduced the time required to process numerically intensive jobs compared to processing the same jobs on a single machine.Broad adoption of grid computing within the academic and research communities led to an early adoption in the mid 1990s of grid computing within the public sector. Both the public and private sectors benefit from grid computing’ ability to reduce the time it takes to process numerically intensive jobs.

Grid computing solutions are a mix of management, provisioning, clustering, storage, virtualization, scheduling and queuing technologies that queue, schedule and process numerically intensive jobs across a pool of computing resources. The business value of grid computing is the time saved to process numerically intensive jobs using pooled resources when compared to processing the same jobs on a single machine.For example, one of my customers in the chip design sector explained the difference between job execution times when using a single 16 processor Sun e25k box and when using their in-house LSF based grid, which pooled resources across 100 duel processor Sun v440s. A numerically intensive job that was executed on the Sun e25k required 94 hours (almost 4 days) to process, the same job processed on the grid in a little over one hour. My customer explained that grid computing provides their organization a considerable competitive advantage over their competitors. Grid computing enables my customer to process and consume data quicker than the competition, which shortened development cycles and shortened their time to market. 

The classic application of grid computing using pooled resources to process numerically intensive jobs has been transformed into the ⁴High Performance Computing (HPC) market. High Performance Computing solutions are a mix of hardware and software that enable the management of large computing environments that process numerically intensive jobs.

In contrast to High Performance Computing solutions, which focus on managing large computing environments that process numerically intensive jobs, Oracle has taken a different approach to grid computing by developing a flexible Grid-based Virtualization architecture to centrally manage and deliver business applications across large heterogeneous computing environments.

From a server perspective the Oracle Grid-based Virtualization architecture supports commodity x86/64 and proprietary hardware, commodity physical and virtual operating systems such as Oracle Unbreakable Linux, Red Hat Enterprise Linux, SUSE Linux Enterprise from Novell, Microsoft’s server and desktop platforms as well as Solaris, AIX and HPUX on their respected hardware platform with physical or virtual storage. From a data center monitoring and management perspective, Oracle Grid-based Virtualization provides comprehensive monitoring and life cycle management for server platforms, Windows desktops and Oracle applications, as well as third-party databases, third-party middleware, storage and network devices. Rather than limit customers to maintain a single vendor or configuration, Oracle Grid-based Virtualization works across various platforms and hardware configurations.

Figure 1.1 shows the scope of Oracle Grid-based Virtualization.

The broad platform support offered by Oracle Grid-based Virtualization allows organizations to pool their existing computing resources to deliver business applications and add capacity, one commodity x86/64 machine at a time. Oracle customers can leverage the flexibility of Oracle Grid-based Virtualization to seamlessly transition from expensive proprietary platforms, one commodity x86/64 server at a time.

Table 1.1 shows an abbreviated platform support matrix for Oracle Grid-based Virtualization.

Let’s review Figure 1.1 and identify the role of each Oracle grid technology within the storage, virtual and physical tier.

The bottom of Figure 1.1 shows us the storage tier. Within the storage tier, Oracle Grid-based Virtualization supports SAN, NAS and local storage. ⁵Oracle Automatic Storage Management (ASM) is used within the storage tier to virtualize storage objects and load balance I/O across disk groups. Oracle Automatic Storage Management builds network clustered storage pools from disparate disks, LUNs, volumes, partitions, and files, providing centralized management of storage objects.

Above the storage tier, Figure 1.1 is broken into two separate tiers. The left side of Figure 1.1 is the virtual tier, which runs virtualized mixed mid-tier clusters, ⁶Oracle Real Application Clusters (RAC) and ⁷virtual desktop infrastructure (VDI). 8Oracle VM is Oracle’s server virtualization technology that powers the virtual tier. Oracle VM supports Oracle ⁸Unbreakable Linux, Red Hat Enterprise Linux and Windows virtual machines that run mixed mid-tier clusters. Oracle VM also supports Oracle Real Application Cluster workloads. The mixed mid-tier cluster workloads are a mixture of ¹⁰Oracle’s Fusion Middleware Stack and Oracle Real Application Clusters. Oracle Real Application Clusters is used to distribute database workload among multiple physical or virtual instances. The right side of Figure 1.1 is the physical tier, which runs Oracle’s Fusion Middleware stack and Oracle Real Application Clusters on bare metal, without server virtualization technologies.

All three tiers, the storage, virtual and physical tiers shown in Figure 1.1 are managed from a single centralized console using ¹¹Oracle Enterprise Manager Grid Control. Oracle Enterprise Manager Grid Control provides comprehensive life cycle management of all objects within the Oracle Grid-based Virtualization environment from a centralized browser based console. 

Figure 1.2 shows the Oracle Enterprise Manager Grid Control user interface.

Let’s shift gears and contrast Oracle Grid-based Virtualization with traditional server virtualization solutions.

Early adopters of server virtualization technologies have validated that server virtualization helps to streamline IT and that server virtualization offers significant cost savings on hardware and facilities (power, real estate, etc). Early adopters of server virtualization have also identified a new challenge with the adoption of server virtualization. The new challenge is called ¹³virtual machine sprawl.

Server virtualization technologies provide the ability to clone and deploy virtual machines in minutes, which helps drive virtual machine sprawl. Virtual machine sprawl is the proliferation of virtual machines. Virtual machine sprawl challenges IT managers in regards to system management, security, resource optimization and IT control.

In terms of virtual machine management, server virtualization technologies offer virtual machine life cycle management capabilities. Virtual machine life cycle management focuses on creating, cloning, templating, starting, stopping, pausing, resuming and migrating virtual machines. The virtual machine life cycle management capabilities in today’s server virtualization technologies do not provide visibility into the virtual machine operating systems or into the applications that run in the virtual machine. The lack of visibility and management capabilities into virtual machine operating systems and applications exacerbate the effects of virtual machine sprawl.

Organizations respond to virtual machine sprawl with procedures and data center management technologies. From a procedural perspective, organizations develop well-defined operational procedures that outline the workflow and approval process to deploy a virtual machine into their environment. From a data center management perspective, organizations try to extend the scope of existing data center management technologies to include the server virtualization technologies, virtual machines and applications. If existing data center management technologies fall short, subsequently new data center management technologies are deployed to augment existing technologies.

Oracle Grid-based Virtualization is a software stack that offers comprehensive data center life cycle management for virtual and physical environments on mixed platforms. Oracle Grid-based Virtualization includes integrated server and storage virtualization technologies. Oracle Grid-based Virtualization does not limit customers to maintain a single vendor or configuration. ¹²Oracle product certification is extremely broad in terms of platform support, which allows customers to use or repurpose what they already have and to increase capacity one commodity x86/64 server at a time.

Footnotes:

¹http://www.cs.wisc.edu/condor/

²http://www.platform.com

³http://www.globus.org

⁴http://en.wikipedia.org/wiki/High-performance_computing

⁵http://www.oracle.com/technology/products/database/asm/index.html

⁶http://www.oracle.com/technology/products/database/clustering/index.html

⁷http://www.ericom.com/oraclevm.asp

⁸http://www.oracle.com/technology/tech/virtualization/index.html

⁹http://www.oracle.com/technology/products/oem/index.html

¹⁰http://www.oracle.com/products/middleware/bea.html

¹¹http://www.oracle.com/technology/products/oem/index.html

¹²http://www.oracle.com/technology/support/metalink/index.html

¹³http://www.linux-mag.com/id/5907