Database Lifecycle Management for Cloud Service Providers
By Sudip Datta-Oracle on Dec 31, 2013
Adopting the Cloud Computing paradigm enables service providers to maximize
revenues while driving capital costs down through greater efficiencies of
working capital and OPEX changes. In case of enterprise private cloud, corporate IT, which
plays the role of the provider, may not be interested in revenues, but still
care about providing differentiated service at lower cost. The efficiency and
cost eventually makes the service profitable and sustainable. This basic tenet has to be satisfied irrespective of the type of service-infrastructure (IaaS), platform (PaaS) or software application (SaaS). In this blog, we specifically focus on the database layer and how its lifecycle gets managed by the Service Providers.
Any service provider needs to ensure that:
- Hardware and software population are in control. As new consumers come in
and some consumers retire, there is a constant flux of resources in the data center. The flux has to be managed and controlled
- The platform for providing the service is standardized, so that operations can be conducted predictable and at scale across a pool of resources
- Mundane and repeatable tasks like backup, patching, etc are automated
- Customer attrition does not happen owing to heightened compliance risk
While the Database Lifecycle Management features of Enterprise Manager have been widely adopted, I feel that the applicability of the features with respect to service providers is yet well understood and hence appreciated. In this blog, let me try addressing how the lifecycle management features can be effective in addressing each of the above requirements.1. Controlling hardware and software population:
Enterprise Manager 12c provides a near real-time view of the assets in a
data center. It comes with out-of-box inventory reports that show the current
population and the growth trend within the data center. The inventory can be
further sliced and diced based on cost center, owner, etc. In a cloud, whether
private or public, the target properties of each asset can be appropriately
populated, so that the provider can easily figure out the distribution of
assets. For example, how many databases are owned by Marketing LOB can be
easily answered. The flux within the data
center is usually higher when virtualization techniques such as server
virtualization and Oracle 12c multitenant option are used. These technologies
make the provisioning process extremely nimble, potentially leading to a higher
number of virtual machines (VMs) or pluggable databases (PDBs) within the data
center and hence accentuating the need for such ongoing reporting. The inventory reports can be also created using BI Publisher and delivered to non-EM users, such as a CIO.
Now, not all reports can always be readily available. There can be situations where a data center manager can seek adhoc information, such as, how many databases owned by a particular customer is running on Exadata. This involves an adhoc query based upon an association, viz. database running on Exadata and target properties, viz. owner being the customer. Enterprise Manager 12c provides a sophisticated Configuration Search feature that lets administrators define such adhoc queries and save them for reuse.
2. Standardization of platform:
The massive standardization of platform components is not merely a nice-to-have for a cloud service provider, it is rather a must-have. A provider may choose to offer various levels of services, tagged with levels such as gold, silver and bronze. However, for each such level, the platform components need to be standardized, not only for ease of manageability but also for ensuring consistency of QOS across all the tenants. So how can the platform be standardized? We can highlight two major Enterprise Manager 12c features here:
The ability to rollout gold images that can be version controlled within Enterprise Manager's Software Library. The inputs of the provisioning process can be "locked down" by the designer of the provisioning process, thereby ensuring that each deployment is a replica of the other.
The ability to compare the configuration of deployments (often referred to as the "Points of Delivery" of the services). This is a very powerful feature that supports 1-n comparisons across multiple tiers of the stack. For example, one can compare an entire database machine from storage cells, compute nodes to databases with one or more of those.
3. Automation of repeatable tasks:
A large portion of OPEX for a service provider is expended while executing mundane and repeatable tasks like backup, log file cleanup or patching. Enterprise Manager 12c comes with an automation framework comprising Jobs and Deployment Procedures that lets administrators define these repetitive actions and schedule them as needed. EMCC’s task automation framework is scalable, carries functions such as ability to schedule, resume, retry which are of paramount importance in conducting mass operations in an enterprise scale cloud. The task automation verbs are also exposed through the EMCLI interface. Oracle Cloud administrators make extensive use of EMCLI for large scale operations on thousands of tenant services.
One of the most popular features of Enterprise Manager 12c is the out-of-box procedures for patch automation. The patching procedures can patch the Linux operating system, clusterware and the database. For minimizing the downtime involved in the patching process Enterprise Manager 12c also supports out-of-place patching that can prepare the patched software ahead of time and migrate the instances one by one as needed. This technique is widely adopted by the service providers to make sure the tenants' downtime related SLAs are respected and adhered to. The co-ordination of such downtime can be instrumented by Enterprise Manager 12c's blackout functionality.
4. Managing Compliance risks:
In a service driven model, the provider is liable in case of security breaches. The consumer and in turn, the customer of the consumer's apps need to be assured that their data is not breached into owing to platform level vulnerabilities. The security breaches often happen owing to faulty configuration such as default passwords, relaxed file permissions, or an open network port. The hardening of the platform therefore, has to be done at all levels-OS, network, database, etc. The security breaches often happen owing to faulty configuration such as default passwords, relaxed file permissions, or an open port. . To manage compliance, administrators can create baselines referred to as Compliance Standard. Any deviations from the baselines triggers compliance violation notifications, alerting administrators to resolve the issue before it creates risk in the environment.
We can therefore see how four major asks from a service provider can be satisfied with the Lifecycle Management features of Enterprise Manager 12c. As substantiated through several third party studies and customer testimonials, these result in higher efficiency with lower OPEX.