Awhile ago I reflected upon my attendance at the recent Cloud Foundry Summit in Basel. In my post, I wrote about what I saw as a very real risk for the commercial vendors (and, in particular, the most successful commercial vendor to date, Pivotal) that organizations, after growing comfortable with Cloud Foundry delivered through a commercial distribution, would move away and operationalize Cloud Foundry usage by leveraging the open source product.
Now, of course, this situation isn’t one which is novel to Cloud Foundry – every open source project has this tension and the success of Red Hat shows that providing services on top of an open source project, be it Linux, OpenStack, OpenShift, Cloud Foundry or whatever, is possible. But what I perceive to be different is that these more modern open source solutions (in particular Cloud Foundry) are inextricably tied to a change in the way that organizations work. Whereas Linux, for example, simply swapped out one operating system for another one, Cloud Foundry both enables, and requires, a new, more agile way of working. Cloud Foundry is, in some respects, the key to unlocking enterprise agility, developer-centricity and a culture of speed and innovation.
A hypothetical case study
So if we run this through and think about a hypothetical example for a minute, you might see my gist. BigCo is a 60-year old financial services company. Most of their IT is based around virtual desktops and the Microsoft Office suite and their back office solutions come from a third party vendor and are proving inflexible and a barrier to innovation.
BigCo knows it needs to change, but is really unsure of where to start. Its CEO has been to a number of leadership lunches and has heard people wax poetic about business innovation through technology but really doesn’t understand what that means.
But the CEO is an open-minded woman and does a bit of digging around and decides to get Pivotal to come in to help make the transition to being a software-led organization. The fine folks at Pivotal come along and spend as much (if not more) time helping the CEO on the softer aspects of this journey (people and culture, systems and process) as they do on the technical aspects. Sure, Pivotal sells a bunch of technology solutions to the CEO, but more of the deal is around service delivery.
Fast forward a couple of years and the experiment is an amazing success. BigCo has really changed direction and is seen by many in the industry as a shining light for digital transformation. Their customers are happy and their staff are excited and challenged on a daily basis.
But then the board questions the CEO on the millions of dollars a year they spend with Pivotal on license fees. Some of the board members are savvy individuals and are well aware that the expenditure is, at least in part, going on a product which is freely available (ahhhhh – open source, what a beautiful thing.) The board instructs the CEO to look at this and consider the necessity of this spend.
Long story short and the CEO makes a far assessment that all of the culture-change stuff is pretty much done and the relationship with Pivotal has moved from one of being a trusted partner and service delivery agent to simply a provider of software. Since that software is freely available, the CEO decides to can the contract and leverage the open source versions of the software being used. The board is chuffed, the CEO gets a pay rise and some Pivotal sales staff feel kind of queasy.
But wait, it happens in real life
My hypothetical story was grossly oversimplified but, at least from my perspective, mirrors some of what might have happened within Fidelity. As I detailed in my post, Fidelity where big Pivotal customers but, after a few years, moved to the open source version of Cloud Foundry.
Now we shouldn’t underestimate what an impressive team Fidelity have and just how much buy-in they have from the executive team. But I can’t believe that Fidelity are a total outlier and this is where my concerns about the long-term prognosis for the open source distributions come in.
Which is why I wanted to talk to someone on the inside. Daniel Jones is the CTO of Engineer Better and someone in an excellent position to comment on my concerns. EngineerBetter was an integral part of Fidelity’s journey from Pivotal to Cloud Foundry, but they are also a Pivotal partner. As such, while they could be expected to be sensitive to the optics of the conversation, they’ve experienced both open source and commercial and are in a position to opine.
The partner perspective
Jones, in discussing the Fidelity case, in particular, was quick to point out some reasons why he feels their situation is one which is not easily replicated by other traditional enterprises. To whit:
EngineerBetter have had the pleasure of helping Fidelity International on their Cloud Foundry journey over the years, and their success is in part due to two factors that aren’t easily replicated by traditional enterprises. Firstly, Emma Hammond has done a fantastic job of creating a forward-thinking team and ensuring that they have breathing space to do adopt modern practices. The second is the maturity of the team’s understanding of DevOps practices – whilst many other enterprise users struggle with concepts like infrastructure-as-code and kanban, these folks are living and breathing it. Without a team like this, users would struggle to make a success of open source Cloud Foundry.
More generally, I asked Jones to comment on this tension between open source and commercial and he sent me the following via email:
Rachael Wonnacott from Fidelity International gave an awesome talk on the great achievements her team has been able to make in deploying open source Cloud Foundry. Fidelity International are looking after 26 individual Cloud Foundry installations, which is a substantial footprint and a sign of mature usage.
Cloud Foundry is the sort of technology that really gives benefits at scale – if you’ve got 50 developers, then the overhead of learning to use and operate it is likely to overshadow the benefits of improved delivery speed. For enterprises with tens or even hundreds of thousands of developers, the benefits are enormous and so this is why commercial vendors have typically focused their attention. This means licensing costs are typically prohibitively expensive for SMEs, turning folks towards the open source distribution instead.
The commercial distributions offer value in three key areas: quality assurance, support, and value-add services. To run something like Cloud Foundry directly from open source, you need to have a team well-versed in treating platforms as a software project: so all the good things like test-driven development, continuous deployment, and infrastructure-as-code. This is primarily as you’ll be missing a layer of testing that commercial vendors will have provided, so your own test pipelines must be rock-solid.
Many regulated customers insist that they need a support agreement in place in order to use a technology. That’s patently not true, they just need to have a plan for if there are problems. As was discussed in the talk, in Fidelity International’s case there is an international team ready and on-call in case of an outage.
Open source users won’t get all the bells-and-whistles of the commercial distributions, such is the ‘open core’ model. If customers don’t find these added extras valuable enough to warrant paying licence fees, then that’s a signal that vendors must pay attention to.
The vendor’s perspective
I reached out to Pivotal to get their take on the tension and their comment was quick and to the point. No comment. Ah well.
MyPOV
It’s not black and white and the sky isn’t going to fall anytime soon for Pivotal for two reasons. Firstly, the number of organizations still needing to make this transition is huge and will keep the Pivotal team in kitchen snacks and cool tee shirts for years. Secondly, and perhaps more importantly longer term, the fact is that large enterprise are risk averse and like the feeling (justified or not) that they have someone to blame when things go wrong.
Despite the huge number of vendor-driven software projects that go horribly over budget and horribly over time, there is still the default that big vendors bring comfort and this is a wave that Pivotal will be able to ride for years.
That said, the commercial distros need to think hard about the value they’re offering above and beyond that of the pure open source project. As these opens source projects inexorably increase the breadth of what they do, this becomes a harder and harder goal to achieve. Time will tell how this all ends up.
As a final word, while the IBM acquisition of Red Hat was obviously a great outcome for the open source company, there is a fairly widespread view that Red Hat’s growth would be hard to continue, constrained as it was by the very fact that it was simply offering surety and a few services on top of an open source product. A cautionary tale, perhaps?