Ways to Save Money on Cloud

Ways to Save Money on Cloud

Posted on 21. Dec, 2017 by in Cloud, Developement, Hosting, internet, linux, Open stack, Programming, Traffic, Web Hosting, windows


Quick Snapshot: Today’s Newsletter

Decreasing the Cost of Public Cloud. Now that cloud providers have addressed cost management, the next step is cloud cost optimization. Cloud optimization would ensure that enterprises pay the lowest price for the same level of cloud service.

OpenStack Steps Up. OpenStack faces stiff competitive pressure from commercial cloud offerings. This is driving their latest effort to deliver an integrated stack that would close the gaps across a number of open source projects.

Unlimited Plans Improving Customer Retention. A recent 451 Research survey of 3,840 primarily North American respondents found unlimited data plans appear to be improving the customer experience and creating stickiness among unlimited users, since they are less likely to consider switching to another carrier.

Microsoft Turning Its Sites on AI. The company formed a new AI research unit last year, bringing together a number of initiatives, including Cortana speech recognition, Bing search, robotics and other projects. The unit has grown from 5,000 employees in mid-2016 to about 8,000 today.


I. Ways to Save Money on Cloud

Understanding cloud costs is a complicated business. A cloud application typically consumes a variety of services, each measured and billed in different ways.

Originally, a number of third-party vendors tried to address this complexity by providing advanced reporting and cost attribution – referred to as ‘cloud cost management.’

However, the leading cloud providers improved their capabilities, leading to consolidation in the cloud cost management market.

Now that cloud providers have addressed cost management, the next step is cloud cost optimization – automatically ensuring that enterprises pay the lowest price for the same level of cloud service.

A number of startups have entered the cloud cost optimization market (and many are prime candidates for acquisition).

Generally, players in the cost-optimization market primarily act as either cloud broker-dealers or cloud waste managers, and in some cases both.

Cloud Broker-Dealers

The cloud cost optimization market started with broker-dealer ClusterK, which was acquired by AWS in mid-2015.

Since then, several other startups have entered the market. The Table below shows some examples.

Cloud Cost Optimization: Broker-Dealers

Third-Party Cloud Waste Managers

To date few of the big cloud vendors have bothered to address the issue of cost waste.

Because of the complexity of cloud applications, there is a substantial risk of waste (unneeded costs) – for example, forgotten VMs left running, expensive VMs being used where cheaper VMs would suffice and resources left running during quiet periods can all generate waste.

The need to ‘trim the fat’ started with AWS’s Trusted Advisor, which provided recommendations on what resources were underutilized.

Several startups subsequently entered the market, promising substantial savings by cleaning up resources that aren’t being fully utilized.

Cloud Cost Optimization: Waste Managers

Consolidation Likely

Cloud cost optimization is a key opportunity for service providers to add value and differentiate in a competitive market, and 451 Research anticipates consolidation in this young market.

AWS, Google and Microsoft are the most obvious potential acquirers.

Other potential buyers include MSPs building and managing applications built on third-party clouds. Cloud cost optimization would be a logical value-add.

Other potential acquirers include vendors that aggregate cloud services (e.g., Accenture, Vodafone and others).

Finally, Cloudability – the leading cloud cost management specialist today – is likely to expand into cost optimization.

II. OpenStack Embraces Other Projects

The OpenStack Foundation’s new mission is to close the gaps between a multitude of open source projects, with the goal of delivering an integrated stack that provides the functionality of commercial cloud stacks (e.g., AWS, Microsoft, Google, etc.).

Beyond the Base

From containers and container management to edge computing and serverless technology, providing a credible alternative to hyperscale clouds will require an integrated, full-stack approach that goes beyond the core OpenStack base.

What form that will take is an open question at this point, but the involvement of other entities – such as tools and projects from organizations such as the Cloud Native Computing Foundation and The Linux Foundation – will be needed.

Enabling an entire open source stack equivalent to AWS, for example, might mean OpenStack and other Foundation technologies forming the core majority of services, with other (non-OpenStack) groups participating.

A first set of measures toward this end – called the Open Infrastructure Integration strategy – was announced at the recent OpenStack Summit in Sydney.

The OpenStack Foundation will manage new open infrastructure projects as part of this integration strategy. The goal is to help organize the ecosystem and user community around domains such as:

  • Datacenter cloud infrastructure
  • Edge infrastructure
  • Container infrastructure
  • Continuous integration and continuous delivery (CI/CD)

Historically, all software managed by the OpenStack Foundation has been part of the OpenStack project, but as the community organizes around emerging domains, new projects may be managed independently with their own technical governance and branding.

The strategy is to provide greater focus and more defined scope to OpenStack core services while supporting new use cases in parallel.

Collaboration Strategy

A four-part strategy to address integration of OpenStack and other open source technologies includes:

  • Documenting cross-project use cases
  • Collaborating across communities, including contributions to and from other open source projects
  • Fostering new projects at the OpenStack Foundation
  • Coordinating end-to-end testing across projects

This means the OpenStack Foundation will collaborate with other open source projects and foundations.

Next Steps

One key capability that OpenStack requires is an integrated way to deploy and manage containers. Kubernetes currently has the leading position in the market, although there are a variety of alternatives. (Google originally developed Kubernetes, and contributed it to the Cloud Native Computing Foundation.)

Where possible, the OpenStack Foundation does not plan to duplicate work that’s already been done. One key goal for the Foundation is to deliver Kubernetes on OpenStack.

The Future

Going forward, the OpenStack Foundation will act as a facilitator for additional infrastructure primitives, enabling key open source environments to work on OpenStack. In addition to Kubernetes, examples include CloudFoundry and TensorFlow.

There is also a need for open source serverless technology that could be used as an alternative to AWS’s Lambda, which has a massive lead in the market. There are a handful of open source serverless mechanisms in the market, although none has much momentum at this point.

The OpenStack Foundation has not positioned itself as ‘looking for non-OpenStack projects,’ but that’s effectively what it is doing, and it’s open about operating them under non-OpenStack conditions.

III. Impact of Unlimited on Wireless Customer Experience

Unlimited data plans appear to be improving the customer experience as subscribers to major carriers report being more satisfied with their overall service than customers on tiered/shared plans.

A recent 451 Research survey of 3,840 primarily North American respondents found that unlimited data plans are creating stickiness among unlimited users, since they are less likely to consider switching to another carrier.

Satisfaction. More unlimited users say they are Very Satisfied with their overall wireless service than subscribers on shared or tiered data plans, but the differences are greatest for Verizon and AT&T customers.

Overall Customer Satisfaction – Unlimited vs. Tiered/Shared (Sep 2017):

Put another away, T-Mobile customers seem to be satisfied regardless of which data plan they’re on, while unlimited might not be as beneficial in improving Sprint customer satisfaction as it had hoped.

On quality of video streaming, T-Mobile and Sprint unlimited customers rate their experience better than metered customers, but for AT&T the opposite is true. At the time of survey fielding, the vast majority of Verizon unlimited customers were unaffected by recent restrictions on streaming quality, which explains the identical customer satisfaction scores on this aspect across data plan types.

Quality of Streaming Video – Unlimited vs. Tiered/Shared (Sep 2017):

On network performance measures, AT&T and Sprint unlimited customers report having a better experience than those on their providers’ shared/tiered plans. But for T-Mobile and Verizon, tiered/shared customers say they’re more satisfied with coverage and reliability than unlimited customers.

Coverage, Reliability and Speed of Network – Unlimited vs. Tiered/Shared (Sep 2017):

AT&T, T-Mobile and Verizon unlimited users rate the cost and value of their plan better than tiered/shared subscribers. But Sprint customers on tiered/shared plans report being more satisfied with the value of their plan than unlimited customers (31% Very Satisfied vs. 29%).

Likelihood of Switching. Unlimited is improving overall retention, as these customers are less likely to indicate they’re switching to another carrier. Naturally, they’re also less likely to explore other data plan options. Both cases are particularly true for Sprint customers more so than subscribers to other major carriers.

Likelihood of Switching – Unlimited vs. Tiered/Shared (Sep 2017):

Reasons to Pay for Unlimited. Best overall value continues to be a key factor encouraging consumers to pay for unlimited data plans, with a total of 38% of unlimited data users citing it. Peace of mind (34%), or not having to worry about incurring overages, is the second most important reason.

A total of 31% of unlimited users say their data plan is shared by multiple smartphones as part of a family plan. This has risen in priority over the past year, up 8 points since a year ago.

Meanwhile, grandfathered cheaper rates for unlimited plans (18%; down 8 points) is waning in priority as several of those customers are being rolled into new pricing.

At the provider level, T-Mobile customers are most likely to cite best overall value, while Verizon and AT&T customers are most likely to say their unlimited data plan is being shared by multiple users.

Top Reasons to Pay for Unlimited Data Plans – By Wireless Provider (Sep 2017):

IV. Microsoft Doubles Down on AI

Microsoft CEO Satya Nadella is refocusing the company with an ‘AI first’ mantra.

The company formed a new AI research unit last year, bringing together a number of initiatives, including Cortana speech recognition, Bing search, robotics and other projects. The unit has grown from 5,000 employees in mid-2016 to about 8,000 today.

In July, Microsoft formed a separate team – called Microsoft Research AI –to focus on the toughest AI challenges.


Cortana Intelligence Gallery is a community for developers and data scientists to create and share resources related to the Cortana Intelligence Suite, including projects developed in Azure Machine Learning Studio and Jupyter Notebooks. It also includes:

  • Reference architectures
  • Design patterns
  • Preconfigured solutions
  • Tutorials

Microsoft AppSource lists more than 20 AI applications built using Cortana Intelligence. And the Azure Marketplace includes more than 400 data processing and analytics products available as Azure services.

AI Breakthroughs

Microsoft has made some significant technical breakthroughs in machine learning.

For example, in 2016 it reached human parity in speech recognition. In a test involving transcribing conversations between people discussing topics such as sports and politics, it achieved a 5.9% error rate, considered to be on parity with humans doing the same transcription.

In August, the team surpassed that with a 5.1% error rate, effectively making it better than human performance.

The next step is to move from recognizing to actually understanding speech.

Enterprise Search

Enterprise search is a challenging technology. End-user demand is high, but so far there has been limited market growth and enterprise adoption.

In September, Microsoft promised to solve the complex problem of enterprise search with Bing for business – a new intelligent search experience for Office 365 and Microsoft 365. It uses AI and the Microsoft Graph to deliver more relevant search results based on your organizational context.

Bing for business represents the technical and commercial viability of AI and machine learning.

Search has long used machine learning, albeit in shallower forms of statistical algorithms rather than deep learning.

Enterprise search is difficult to set up, and requires constant maintenance and tuning. But that tweaking is another area where machine learning comes into play, with the search software learning what people want to know and adapting to it over time.

It’s too early to tell how effective Bing for business will be, but searching across both cloud and on-premises data with one tool is the next challenge and opportunity for Microsoft.


In addition to Microsoft, all of the major cloud providers (AWS, Google, IBM) are putting AI and machine learning at the heart of their offerings, as are many smaller vendors – both open source and closed source.

For developers, the products are roughly on par in terms of ease of use; the differences come in a variety of variables, including price, licensing terms, ease of data export, trained models and many other factors.

At the AI application level, Microsoft competes primarily with Oracle Adaptive Intelligent Apps, Salesforce Einstein and SAP Leonardo.

And in parts of its AI business, Microsoft also competes with OpenText Magellan, Infor Coleman and SugarCRM Candace.

SWOT Analysis

Rate this ➜

0 people like this.

Tags: , , , , ,

No comments.

Leave a Reply