Oracle Exalogic Elastic Cloud 2.0 (“Exalogic”) adds features that will make the system better suited for its role as a Cloud deployment target, says Tony Baer, Prinicpal Analyst, Ovum Enterprise Solutions. It adds server virtualisation capabilities that accelerate provisioning and deliver the expected elasticity.
A new lightweight virtualisation engine optimised for Oracle workloads improves application performance. This comes atop incremental enhancements to existing features such as Oracle Exabus.
The importance of the 2.0 release is not necessarily in the individual improvements, but rather in how they help the system live up to its original “elastic Cloud ” branding.
So far, Oracle has successfully sold Exalogic to its installed base, primarily as the middleware complement to the Exadata database platform and as the infrastructure for its business applications.
Oracle has promoted the “soft” business benefits, such as improved responsiveness and efficiency that comes from its engineered systems, but it still needs to formally state the total cost of ownership case for Exalogic to show that the engineered system approach is better than commodity platforms for the customer’s bottom line.
Evolution of role for Oracle Exalogic
Cloud branding aside, until now Oracle Exalogic has mostly acted as a logical middle-tier add-on for Exadata customers seeking workload consolidation and business applications performance improvement. Ovum estimates that up to half of Oracle’s Exalogic business has gone to that existing base.
While Exalogic could also be used as a shared Cloud deployment resource, until now it has lacked the server virtualisation and management features that would support the rapid provisioning of workloads deemed critical for Cloud deployment.
The new release is an important first step toward addressing that gap.
A logical next step in Cloud management
Although Oracle has positioned Exalogic as a Cloud deployment platform, the initial release focused on scalability and performance optimisation for Oracle’s middleware.
However, the release of Oracle WebLogic Server 12c at the end of last year started putting some of the pieces in place. For instance, Oracle Traffic Director added a feature that automatically routes and load-balances WebLogic Server workloads onto Exalogic’s high-speed Infiniband backplane.
Oracle Exalogic 2.0 adds several features that are the logical next step. They start with a new high-level management layer that allows users to create virtual servers and allocate infrastructure resources such as memory, networking, and storage.
Beneath the hood, some of the new Cloud -friendly features of the version 2.0 release include:
- Oracle VM 3.0, which provides server virtualisation, a building block for provisioning and running of multiple workloads that is critical to providing the elasticity that is expected of Cloud operation.
- Oracle VM Templates, which is the format for ready-to-run VM images on Oracle Exalogic.
- Single Root I/O Virtualisation (SR-IOV), which optimises the ability of Oracle Fusion Middleware and Oracle Virtual Server (vServer) workloads to access the Infiniband backplane.
- Oracle Traffic Director, which provides software-based load-balancing.
- Oracle Tuxedo 12c Application Runtime (ART) for Oracle Exalogic used for re-hosting mainframe OLTP and batch applications.
- Oracle’s work is not yet finished. While it initially focused on reliability and performance for the new SR-IOV feature, Oracle still needs management tools to make migration from earlier Oracle VM versions seamless.
We expect that other features, such as Oracle Virtual Assemblies (composite artifacts containing multiple VM templates and configuration of tiers such as network and storage) to be slated for a subsequent dot release.
Buy your packaged data centre from Oracle or somebody else?
Oracle Exalogic is one of a growing set of offerings designed to address key pain points for enterprises including deployment complexity, volatility of demand, and cost. Customers have a growing range of options:
- Rent capacity from Cloud service providers, eliminating the capital costs, but not necessarily the complexity.
- Build and deploy commodity scale-out cluster farms (often with open source infrastructure management software) to reduce operating cost, but with the need of sophisticated internal expertise.
- Buy optimised systems that are general-purpose platforms, which automate or simplify management and deployment, but with higher up-front cost compared to acquiring commodity clusters.
- Buy optimised systems that are purpose-built for specific tasks, such as running databases (for example, Exadata), analytics (for example, Exalytics), middle or application tier (for example, Exalogic), or more specialised compute-intensive functions such as firewall or encryption.
- Comparing Oracle’s engineered systems, such as Exalogic, to rivals such as VCE, IBM PureSystems, or HP Cloud Systems is not an apples-to-apples comparison. The latter are general-purpose systems while Oracle’s is purpose-built.
There are further architectural differences, such as the use of a single virtualisation or storage engine or giving the customer a choice. The decision for enterprises involves factoring in the complexity and total cost of ownership (TCO).
With Exalogic 2.0, Oracle has taken important steps to reduce the complexity of managing Cloud deployment, and it now needs to make the cost case more explicit. Oracle claims that it has the TCO data versus IBM and HP, but has not yet made it public.
Oracle has pointed out the business benefits of an engineered system approach like Exalogic, and it now needs to complete the message by quantifying that engineered systems are cheaper in the long run than building your own scale-out clusters from commodity components.