Tuesday Dec 31, 2013

Database Lifecycle Management for Cloud Service Providers

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.

Stay Connected:

Twitter |  Face book |  You Tube |  Linked in |  Google+ Newsletter

Monday May 21, 2012

Benefits of using Ops Center to deploy and manage Solaris 11

One of the more significant new features in Oracle Enterprise Manager Ops Center 12c is the ability to install Ops Center on Oracle Solaris 11, and to deploy and manage systems running Solaris 11.  The Solaris 11 capabilities are in addition to the analogous features for Solaris 10 and Linux, which can all be handled from the same Ops Center infrastructure.

When the Ops Center Enterprise Controller (EC) is installed on a system running Solaris 11, the EC can create a Solaris 11 Software Update Library containing Oracle Solaris Image Packaging System (IPS) content that is synchronized with the main Oracle repository at pkg.oracle.com. The Ops Center managed Solaris 11 repository becomes the package (pkg) publisher for downstream Solaris 11 deployments and updates on all Solaris 11 systems being managed by Ops Center.

Ops Center provides the ability to define Solaris 11 OS and Software Profiles comprised of Oracle Solaris 11 packages, user-supplied custom packages, scripts, and other files.  Such Software Profiles profiles can then be used to install and update software on systems already running Solaris 11 in a structured and consistent way.  Ops Center not only caches the main Oracle Solaris IPS repository, but more importantly it gives admins the ability to define their own preferred collection of packages so that systems can easily be kept in sync with each other, running a well-defined, life-cycle-managed Standard Operating Environment (SOE), instead of just whatever the latest content is at pkg.oracle.com.

Ops Center 12c also adds Solaris 11 features for bare-metal OS Provisioning, based on the Solaris 11 Auto Install (AI) facility.  Ops Center configures the Solaris 11 AI in a way that shields admins from needing to write custom AI manifests or custom "first boot" packages.  Solaris 11 deployments using Ops Center follow similar profile-based patterns as for Solaris 10 or Linux, all of which can all be deployed from the same Ops Center infrastructure running on Solaris 11.  The gory details of all these different times of bare-metal OS Provisioning are handled automatically for the user so that he or she does not need to put time and resources into manually creating and maintaining infrastructure for deploying different OS's natively -- Solaris 11 with AI, Solaris 10 JumpStart with JET, or Linux with Kickstart or AutoYast.  All of those OS's are handled by Ops Center under the covers, based on whatever network boot capability the OS requires (PXE/DHCP, WANBOOT, or AI), and all from the same Ops Center infrastructure running on Solaris 11.

Specific to Solaris 11 OS Provisioning (via AI), Ops Center provides its own "first-boot" package+scripts to customize the Solaris 11 deployment, and in particular this approach automatically installs the Ops Center agent.  With the Ops Center Agent in place right from the start, it is easy to handle post-install steps using the Ops Center features for handling Solaris 11 OS and Software Profiles containing additional packages, scripts, and content from the Solaris 11 Software Library, described above.

Tying bare-metal Solaris 11 deployment and post-install customization together is a key way that Ops Center simplifies the overall life-cycle management for Solaris 11 (in addition to Solaris 10 and Linux).   For example, a top-level plan based on "Configure and Install Logical Domains" can create numerous Logical Domains into an Oracle VM Server for SPARC "Server Pool" and provision Solaris 11 into each LDom Guest based on a powerful multi-step "Install Server" plan.  Such a plan can cover the end-to-end steps for installing and updating the OS, running scripts and adjusting monitoring parameters, etc.

Here is an example of the kinds of activities that can be performed in conjunction an OS Deployment, or separately, depending on the need:

Oracle Enterprise Manager Ops Center 12c Deployment Plan

NOTE: In the above example, the last step "Create Guests" can be used to create one or more Solaris Containers within the LDom Guest, rounding out the end-to-end deployment all the way from LDom Guest to Solaris 11 Global Zone to multiple Solaris Containers, if so desired.

One of the nicest aspect of deploying and managing Solaris 11 using Ops Center Plans and Profiles is that the same content can be applied as updates to existing Solaris 11 systems, aligned to the same content as chained off a bare-metal OS Provisioning.  It is up to the user which steps they want to include in a deployment plan -- whether they are updating Software Profiles on systems deployed 6 months ago to match the latest standard, or they are deploying new systems based on that same standard, Ops Center provides the means to insure that the outcome is consistent.

Here is an example of the kinds of activities that can be performed on an existing Solaris 11 OS -- either ad hoc, or immediately after a Solaris 11 OS Provisioning step, so that whether the life-cycle started with a new system, or the intent is to update a system deployed six months ago, the outcome can be the same:

S11 Update Multi-Step Plan

In short, Ops Center running on Solaris 11 can manage Solaris 10 and Linux systems, all from a common infrastructure, and all based on a simplified, consistent, profile- and plan-based way to do the OS and Software deployments and updates.  The net effect is an easy to use way to managing the life-cycle of heterogeneous systems, in a very consistent way through automation and re-use.

Please let us know what you think?  Until next time...
\Leon
--

Leon Shaner | Senior IT/Product Architect
Systems Management | Ops Center Engineering @ Oracle

The views expressed on this [blog; Web site] are my own and do not necessarily reflect the views of Oracle.



For more information, please go to Oracle Enterprise Manager  web page or  follow us at : 

Twitter | Facebook | YouTube | Linkedin | Newsletter

About

Latest information and perspectives on Oracle Enterprise Manager.

Related Blogs




Search

Archives
« April 2014
SunMonTueWedThuFriSat
  
3
5
6
7
9
10
11
12
13
14
15
17
18
19
20
21
22
23
24
25
26
27
28
29
30
   
       
Today