Extreme Automation also the Key to Successful Private Clouds

A couple of recent articles in the past few weeks on TechTarget caught my eye.  One article, “SAP Cloud embrace helps feed the beast for AWS, Azure”, was focused on the news that both AWS and Azure had just certified SAP Hana on their clouds.  Both AWS and Azure also added new capabilities to their offerings for Hana at the same time that they announced their certifications.

Extreme Automation

The event is a big win for SAP which gives enterprises that would like to run SAP in the cloud additional options.  It also adds a new source of potential growth to both AWS and Azure.  And because applications like SAP can be extremely data intensive, the win for AWS and Azure includes not just compute services but also data services.

The other article, “Public IaaS market leaves some cloud vendors in the dust” is about the dominance of three Public Cloud vendors in the Infrastructure as a Service marketplace.  These three vendors, AWS, Azure and Google, have so far captured the lion share of both the market and mindshare for Public Cloud IaaS.  The article includes a quote by Melanie Posey of IDC where she says that these three vendors “dominate the market by encouraging and supporting self-service models, credit-card swipe, extreme automation, scalability, fault tolerance and a primarily net-new DevOps approach to applications and workloads”.

The Same Requirements across Public and Private Clouds

The continued and in fact accelerated adoption of Public Cloud services has some deep implications for organizations seeking to build a Private Cloud.  Other than the credit card swiping, the set of characteristics that Melanie Posey calls out in the article on the Public IaaS marketplace provides an excellent North Star not only for what Public Clouds need to deliver but also for what Private Clouds should aspire to.

As Melanie points out, Extreme Automation does indeed characterize Public Clouds.  Private Clouds need to emulate the same by collapsing the many disjointed steps involved in provisioning resources into a set of integrated actions that can be run in a fraction of the time it takes most internal IT organizations to provision data center resources.  Self-service in multiple forms including both catalog access and direct access to component resources via an API is also a fundamental requirement if the goal is to truly meet the expectations of end users.

Besides speeding up resource delivery and making self-service possible, Extreme  Automation also provides the foundation for making a cloud hyper-efficient and one capable of maximizing both human and infrastructure resources.  There is just no way that a data center that does not contain extremely high levels of automation can compete with the efficiencies achieved in the best Public Cloud environments.

A High Bar for Private Clouds

To better understand just how high a bar the Public Cloud has set for the Private Cloud when it comes to fully automating data center operations consider some research done by 451 Research back in October of 2015.   At that time, their 3 part Cloud Price Index report provided 5 benchmarks focused on Cloud Computing costs (see table below).  The on premise cloud costs of the first three benchmarks include data center costs along with the cost of software license and support but they do not include the labor involved in managing the data center environment.  Public and Hosted Cloud benchmarks, by their very nature, include all the costs necessary to deliver the service including the cost of the labor associated with managing the environment.

Cloud Computing Approach OpenStack DIY OpenStack Distro Proprietary Solution (VMW, MSFT) Public Cloud Hosted Private
Cost Per Hour of VM $          0.06  $          0.08  $          0.10  $          0.17  $          0.26

Given this apples to oranges problem of comparing Private Cloud to Public Cloud costs – to help with compares across on premise and Public Cloud scenarios 451 Research provided some “golden ratios” that could be used to facilitate the comparisons.  These golden ratios were essentially the breakeven points where a buyer of cloud services should be indifferent to one scenario over another from a cost perspective.

Compares Using “Golden Ratios”

At the time of the report, for proprietary solutions (10 cents per VM hour) versus Public Cloud (17 cents per VM hour) 451 Research provided a golden ratio of 250:1 for this compare.  This means that you have to be able to support at least 250 VMs for every infrastructure admin associated with those same VMs before a Private Cloud would make more sense than a Public Cloud if you are only considering costs.

You should keep in mind that this would include the costs not only VI admins but physical server admins, storage admins and likely some number of your network admins as well.  For this to be a true apples to apples compare you would also need to account for the management overhead associated with your data center since you can bet the cost you pay for VM at AWS includes the fully loaded cost of everything.

You should definitely check out the reports on 451 Research if you can but netting this all out it becomes clear that for a Private Cloud to be as cost effective as the Public Cloud you need an organization that is extremely efficient.  Efficient in terms of both compute utilization and also the application of man power used to manage the environment.  The only way to achieve these high levels of efficiency is to fully automate all of the core IT processes associated with provisioning infrastructure in a Private Cloud.

It is Possible to Meet the Bar

Companies like Tribune Media are good examples of the kinds of stats organizations will need to achieve in order to provide an on premise Private Cloud that is the equal to what their customers could experience in the Public Cloud.  Tribune Media was able to dramatically decrease the time it takes to provision production ready infrastructure by automating everything they could in their data center.  They were also able to dramatically decrease the cost to deliver infrastructure while at the same time providing a self-service environment that met the needs of their business.

After implementing a software defined data center based on VMware technologies Tribune Media was able to reduce the cost of running their data center by 80%.  You can learn more about what Tribune Media did by checking out Rich Bourdeau’s blog on Tribune Media creating a “Frictionless Enterprise” .

Today the world has moved past the initial cloud wars of “Public Cloud first” or “Private Cloud only” strategies to détente.  Numerous analyst studies now confirm that most companies are looking to adopt a Multi-Cloud strategy.  These companies expect to continue to deploy on premise workloads in a Private Cloud but also to run workloads on one or more Public Clouds as part of comprehensive IT strategy.

Given this new reality, for IT organizations to remain relevant they must:  (1) Build and operate a Private Cloud that is hyper efficient, delivers resources on a time scale appropriate for the business and provides a user experience that meets the expectations of line of business partners – AND – (2) Manage a Multi-Cloud environment capable of treating each participating cloud as unique in terms of the level of governance, automation and operations management needed to effectively leverage each cloud as part of a comprehensive cloud services strategy.

Accelerate Multi-Cloud Service Delivery with VMware – Learn More

VMware’s vRealize Suite is an Enterprise-ready, cloud management platform that delivers the industry’s most complete solution for managing a heterogeneous, Private, Public, Hybrid and Multi-cloud environments.  vRealize Suite supports companies addressing cloud management requirements across day one service delivery and day two operations for compute, storage, network and application level resources.  The easiest and quickest way to learn more about VMware Cloud Management Platform solutions is to visit us online at VMware vRealize Suite.


One comment has been added so far

Leave a Reply

Your email address will not be published. Required fields are marked *