Digital Transformation
Showing results for 
Search instead for 
Did you mean: 

On premises "as-as-service" - private "public" cloud


Two years ago, I presented at our DISCOVER conference in Las Vegas with the CIO of a publishing company from New Zealand. He talked about how he had worked with HPE to consolidate from seven datacenter to one. As you can imagine, before the consolidation he and his team were nervous - would they lose data?, would performance be as good after the work?, and so on.

The HPE lead consultant said to him, “I know this is your first data centre consolidation. But it’s not our [HPE’s] first data centre consolidation”. That was an important point. The experience that comes from doing something many times counts for a lot, and this principal comes to the fore when we consider the platforms that are required for digital transformations.

In this post, I’d like to explore the concept of the platforms required for digital transformation, and how we deliver those platforms to the people who need them.

The requirements of a DX-hero
Digitization projects, whether they are in mobile banking or omni-channel retailing or IoT in manufacturing, are a series of small steps, rather than the more traditional, 18 month, waterfall affairs. This is mainly because of the uncertainty of going into the unknown territory associated with digitization.


So, we need to give our DX-heroes (digital transformation heroes) platforms on which they can create these series of steps.

What do these platforms contain? Perhaps a platform on which to develop “cloud-ready” applications (i.e. containerised, mesh applications and services) that contains Docker and Mesosphere. Or, maybe a data & analytics platform containing an Hadoop “data lake”, Spark (high-speed analytics) and TensorFlow (machine learning).

You’ve probably noticed something in common about the software products that go into these platforms. They are all open source software. This is change over the past. In the past, open source was seen as something to play with, but, apart from perhaps Apache web servers and Linux, not something you’d actually use in production. Now, open source is regularly used in production systems, and Enterprise-grade support is available for it.

And one of the characteristics about the open source used by DX-heroes is that it moves quickly. One minute it’s Hadoop 1, then it’s Hadoop 2 and then there is Hadoop 3. Two years ago, Spark didn’t exist. Now, it’s the hot data & analytics download - until machine learning offerings take the top spot.

So, our DX-heroes need platforms. And on these platforms, they need collections of open source. And this open source is moving quickly.

How does Enterprise IT keep up?
Spare a thought for the Enterprise IT team. Because open source changes quickly, our DX-heroes are constantly asking Enterprise IT to give them the latest open source versions. This is really tough on Enterprise IT, because they are being asked to keep up to date on a huge range of fast-changing open source.

What about using a PaaS?
So, if your Enterprise IT department doesn’t have the resource to track open source’s every release, what else can you do? You can go to a PaaS - platform as a service - that offers the functionality that you need. PaaS cloud providers will be offering the same PaaS platforms to thousands of customers, and so they can use the power of scale.

SIDE NOTE : when some people see PaaS, they think, “a platform for developers”. I’m using PaaS in the more general sense. A PaaS could be a developer’s platform, but it could be a data scientist’s platform or an IoT developer’s platform.


Thus, today, you have two extremes. You can run your Enterprise IT department ragged by demanding the latest releases of a range of open source, or you can have your digitally-powered app or analytics run on a PaaS.

However, two pieces of recent research have shown that some workloads are not best suited to public cloud. This can be for a variety of reasons - absolute cost, prediction of cost, assuring performance, end-customers’ concerns over data control, or the need to do analytics at the edge.

The middle way - on premises platforms as a service
There is a middle way - “on premises as-a-service”. With on premises as-a-service, platforms are managed remotely on your premises by a “service provider”. So, Enterprise IT don’t have to keep the open source based platforms up to date themselves, but they can keep control by having the hardware on premises.

On premises "as-a-service" might give the best balance of control and admin costOn premises "as-a-service" might give the best balance of control and admin cost

Because the on premises “service provider” will be doing the same thing for many customers, it gets the scaling that PaaS cloud providers get.

What do we mean by a platform?
The platforms that are required will, of course, change over time. But today, these platforms might include a micro-services based app platform with Docker and Mesosphere on it. Or a data & analytics platform with Hadoop and perhaps Spark and a machine learning product.

We might see the SAP platform as a platform too.

And there might be IoT platforms for the edge too. While it’s often (43% of the time) not possible to do IoT analytics anywhere but the edge (I discuss this in another post - “The death of the cloud. Long live the edge”), it is possible for the on premises service provider to manage the IoT platform on those edge devices. Because 76% of IoT developments are led by non-Enterprise IT groups, the ability to have someone else take care of the IoT platform they need for them is appealing.

I’m a big fan of the concept of a platform being managed by someone else on your premises. Developers of digitally-powered applications and analytics are typically not from Enterprise IT (only 26% of digitally-fuelled development is controlled by Enterprise IT according to HPE’s own research), and they don’t want to get sidetracked from the job of using digital to help the business by keeping a multitude of open-source applications up to date. And yet, because of cost or performance or control concerns, or because the platform has to run at the edge, they don’t want to use public cloud based platforms either.

So, let some else do the remote management of platforms on machines on your premises. Because they will be doing the same thing across a lot of customers, they will be able to use the power of scale - it might be your first machine-learning platform, but it won’t be their first machine-learning platform.

What’s the future of Cloud?
Get inspired at Enterprise.nxt.
> Go now

Mike Shaw
Director Strategic Marketing
Hewlett Packard Enterprise

twitter.gif @mike_j_shaw
linkedin.gif Mike Shaw

Mike Shaw
Director Strategic Marketing

linkedin.gifMike Shaw

0 Kudos
About the Author


Mike has been with HPE for 30 years. Half of that time was in research and development, mainly as an architect. The other 15 years has been spent in product management, product marketing, and now, strategic marketing. .


Thanks a lot for your guidance and sharing information.

Jan 30-31, 2018
Expert Days - 2018
Visit this forum and get the schedules for online HPE Expert Days where you can talk to HPE product experts, R&D and support team members and get answ...
Read more
See posts for dates
HPE Webinars - 2018
Find out about this year's live broadcasts and on-demand webinars.
Read more
View all