Of the 350 companies surveyed, 74% had moved an application into the public cloud, and then for a variety of reasons and circumstances, decided to move it back into their on-premises or private cloud infrastructure.


More you can find on the following link:



The first interesting finding is the growing sense that the dynamic multi-cloud reflects reality.

Of the 350 companies surveyed, 74% had moved an application into the cloud and then moved it back into their own infrastructure.


According to survey, driving factors to move applications back to own infrastructure where performance and security (each with 52% share).

Forty percent of respondents noted that in some cases the cloud deployments they moved back into its own infrastructure were “planned temporary” deployments.

Temporary infrastructure set up during the inevitable IT transition associated with a merger or acquisition is a great example of planned temporary cloud deployments, but there are many more.


Regulatory issues came up for 21% percent of respondents; this is an example of how the “dynamic” part of the dynamic multi-cloud may not always be in your control.


Performance as argument to move heavy Cloud load back to on-premises


The shared nature of cloud infrastructure ensures that no one really gets a complete picture of what is happening on it.

Besides shared nature of Public Cloud, there are so many layers between the hardware and application that is running within your multi-tenant image and many constraints (IO, CPU, memory, network bandwidth imposed by SND – Software Defined Network) imposed by selected Cloud VM, that justifies performance arguments.

While Cloud is designed for economy, Grid architecture is designed for performance.

Additionally, due to the complex pricing, sometimes running some heavy load on-premises can be a better & cheaper option compared with a Cloud.

That explains why performance is one of the two top reasons why enterprises had moved back in to their own infrastructure.


Security & regulatory issues (21%) as the arguments to move back to on-premises


Cloud, especially Public Cloud, is an attractive option for economically running your apps in various part of the world without building Data center, establishing remote office and many other expenses.

Unfortunately it is the least secure option (remember a bug in OpenStack that allows data leakage among multi-tenants?).

For many large enterprises, shared Public Cloud model is not an option.

There are also many regulatory, legislation, security and data privacy restrictions in place, especially in banking and telco industry.

This is where private Cloud justify its existence.


Running a Cloud on-premises


Although a public Cloud is the most popular Cloud architecture, especially for developers, testers and small shops, due to its numerous limitations is not solution the best solution for all cases.

Big enterprises always put security as one of the most important features they want to have under the full control.

Here is the list what largest vendors have to offer where on-premises Cloud or hybrid Cloud is the only option.


Amazon AWS


AWS is the largest Cloud vendor (47.8 percent of the IaaS market in 2018 according to the Gartner).

Amazon called its on-premises Cloud AWS Outposts, which is a new service slated to launch in late 2019.

An Outpost works as an extension of the AWS Region into your own data center.

AWS Outposts bring native AWS services, infrastructure, and operating models to virtually any data center, co-location space, or on-premises facility.

AWS Outposts enable you to develop once and deploy in the AWS cloud or on-premises without having to rewrite your applications.

With Outposts, you have the same hardware and software infrastructure and a consistent set of services and tools across your AWS cloud and on-premises environments to build and run modern, cloud-native applications anywhere.

AWS Outposts allow you to choose the AWS native variant of AWS Outposts or the VMware Cloud on AWS Outposts variant as your management plane for your APIs, management console, automation, governance policies, and security controls for all your applications across the AWS cloud and on-premises locations.

AWS Outposts are fully managed and supported by AWS, and AWS automatically manages and updates AWS Outposts as part of its ongoing operations in the public AWS region. You do not have to worry about updating or patching infrastructure.

AWS Outposts come in two variants:

  • VMware Cloud on AWS Outposts allows you to use the same VMware control plane and APIs you use to run your infrastructure
  • AWS native variant of AWS Outposts allows you to use the same exact APIs and control plane you use to run in the AWS cloud, but on premises


You can use AWS Outposts to launch a range of Amazon EC2 instances and Amazon EBS volumes locally.

In addition to EC2 and EBS you can also run a wide range of AWS services locally on Outposts such as:

  • Amazon ECS and Amazon EKS clusters for container-based applications
  • Amazon EMR clusters for data analytics (BigData)
  • Amazon RDS instances for relational database services
  • Amazon SageMaker and Amazon MSK are coming soon after launch
  • VPC

More on the following links:







With Outpost you can run AWS Cloud within your DC.

Amazon will take care of bringing a hardware, installation and setup, and after going live, maintenance (upgrading & patching) of AWS Outpost.

AWS is currently the largest Cloud provider, with revenue larger than the following three competitors combined.

AWS is also the most mature Cloud provider, with the largest number of availability zones and regions.

Additionally, AWS is the most reliable cloud provider with availability approaching 100%.



Major issue with Outpost is in it’s appearance on the market a bit tool late.

Reason for that is mainly due to Amazon corporate culture which aims at end-customer, not large enterprises.

AWS Cloud is propriety which means you still don’t have a full control of what is going on behind the scene (sometimes it is required due to security & legislative restrictions).

Amazon will maintain you AWS on-premises Cloud, thus you can’t take over and manage AWS instead of Amazon.

In case that is required, you should look at other vendors in this list.



Microsoft Azure


Azure is the second largest Cloud vendor (15.5 percent of the IaaS market in 2018 according to the Gartner).

While AWS targets mainly Public Cloud, Azure aims at enterprises.

This can be seen by looking at the date of public release of on-premises Cloud: while AWS Outpost will enter production by the end of 2019, Azure Stack shipped 2017.

Besides running cloud apps on-premises, Azure offers Azure Stack HCI solutions which allows you to run virtualized on-premises apps on hyper converged infrastructure with easy integration with classical Cloud services.

As hybrid-cloud architecture style dominant among large enterprises, with two years ahead of its main competitor (AWS) in on-premises Cloud, partnership with Oracle and other major on-premises player, Microsoft won the hearts of enterprises.

More you can find on the following link:



There are two ways to buy Azure Stack:

  • a system you manage
  • a managed service

More on the following link:



There are two possible connection models:

  • connected
  • disconnected

With connected mode (default mode) you can choose between Azure Active Directory (Azure AD) and Active Directory Federated Services (AD FS).

Only the later is possible with disconnected mode.


Another important difference is a billing model.

In connected mode you can choose either pay-as-you-use or capacity-based mode.

In disconnected mode you can choose only the later one.

Azure Stack disconnected mode is the best option when you need to fulfil regulatory and security restrictions.




Azure is the second largest Cloud provider, backed by Microsoft who really understand enterprises.

Large number of regions and availability zones with enterprise flexibility where customers can choose among many options available from the Public Cloud, dedicated host, hybrid and private cloud options, for majority of customers Azure is a serious option to choose from.

Vast majority of enterprise are already using some from large portfolio of Microsoft products, especially Active Directory, mail servers and Office.

Although previously Azure targets mostly Windows customers, that has changed significantly.

Today the most of services and Cloud Virtual Machines are coming from Open Source.

Azure is also very simple to use and very polish platform.

With connected and disconnected mode, there is not much left that can be done to secure your Cloud deployment.


More on:

https://docs.microsoft.com/en-us/azure-stack/operator/azure-stack-overview?view=azs-1908 https://docs.microsoft.com/en-us/azure-stack/operator/azure-stack-connected-deployment?view=azs-1908





Azure has slightly less reliability comparing it with AWS.

Number of availability zones and regions are also a bit smaller, but Azure is aggressively expending.

One of Microsoft partners (HPE, Dell, Cisco…) will bring the hardware and perform installation and setup, which implies you’ll have at least two points of contact:

  • Microsoft
  • Microsoft Azure Stack partner

In case of managed service option, you have to add your service provider as a third contact.


In case of Amazon, you have one point of contact (Amazon instead of partner network) who will choose the most appropriate hardware, and who will perform install and setup of your on-premise Cloud.




Oracle Cloud Appliance also allows you to run Cloud within your Data Center.

Oracle will bring its hardware and install it.

You have two options:

  • you can leave to Oracle to manage your private Oracle Cloud
  • you can manage Oracle Cloud (which is based on OpenNebula) by yourself

The later option is the key difference as it allows to keep everything under your control.

That will allow you to to pass all architectural and security requirements, as well as all regulatory compliance tests.

It is an interesting option especially for organizations that process, transmit or store payment card data, as it fulfill PCI DSS (Payment Card Industry Data Security Standard).

As Oracle is a small player in a Cloud industry, on the OpenWorld it announce always Cloud free tier with Always Free Services and Always Free Oracle Autonomous Database:





Like Microsoft, Oracle even better understands the largest customers (Microsoft is more in SMB segment of the market).

Oracle is the only Cloud provider that can offer its own hardware (ex Sun Microsystem hardware) and software in on-premise Cloud setup.


Besides Oracle, from all Cloud providers only IBM can offers its own hardware + Cloud Software, but at the moment IBM doesn’t have on-premises Cloud in its offering.

Oracle Cloud is based on Open Nebula Open Source, which gives you an additional level of control.

Like Microsoft, Oracle also let you to choose between managing Oracle Cloud by Oracle (management of upgrades and patches) or let you take over Cloud management tasks.




Unlike on-premises, Oracle is a very small player in Cloud, with less than 1% of market share with 16 regions, but is growing quickly (Oracle has opened 12 new regions in the past year, with plans to launch 20 new regions by the end of 2020).

Additionally Oracle Cloud targets mainly existing Oracle on-premises database customers and Java users. In case it doesn’t fit to your needs, you’ll need to take a look at other Cloud vendors, especially Amazon, Microsoft or Google.





GCP on the 3rd place with 4% market share and at the moment can’t offer on-premises Cloud like Microsoft, Oracle or Amazon.

This is probably going to change by the arrival of Thomas Kurian from Oracle, as his strategy is to add Oracle enterprise culture to a Google.

More on the following link:



All what Google can offer now are Anthos & GKE On-Prem which is not the same.

GKE On-Prem is hybrid cloud software that brings Google Kubernetes Engine (GKE) to on-premises data centers.

With GKE On-Prem, you can create, manage, and upgrade Kubernetes clusters in your on-prem environment and connect them to Google Cloud Platform Console.


Anthos deployed on VMware lets you take advantage of Kubernetes and cloud technology in your data center.

You get Google Kubernetes Engine (GKE) experience with quick, managed, and simple installs as well as upgrades validated by Google.

Google Cloud Console gives you a single pane of glass view for managing your clusters across on-premises and cloud environments.

Under the covers, Anthos is a combination of Google’s Kubernetes Engine (GKE), GKE On-Prem and the Anthos Config Management console for unified administration, policies and security across hybrid Kubernetes deployments.

It is hardware agnostic and can be run on customers’ existing servers or with VMware, Dell EMC, HPE, Intel, and Lenovo racks.

More you can find on the following links:






Google is the first Cloud first company with an excellent Open Source reputation.

Running your Kubernetes managed apps that are Cloud neutral (like Java) is interesting concept that can cut multi-cloud costs significantly.




Anthos is not on-premises Cloud that I’ve described previously.

If you need to catch up with regulations and security requirements, you should look at Azure/AWS/Oracle.




With acquisition of Red Hat (and its OIpenShift platform), IBM will become the largest hybrid-cloud provider, but it’s still to early to confirm that is really happening.




Main goal of this article is to give you option you have on your disposal in cases you want to combine scalability of a Cloud with on-premises security and full control.


You can find another great article which is covering the same subject on the following link:


Get notified when a new post is published!



There are no comments yet. Why not start the discussion?

Leave a Reply

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

This site uses Akismet to reduce spam. Learn how your comment data is processed.