The newly announced Portworx Data Services from Pure Storage sounds neat, but I’ve seen a lot of internal marketplace/service catalog attempts over the years, so I’m somewhat skeptical.
The big challenge for internal service catalogs is that they tend to leave out a major feature of their public cloud equivalents: pricing.
Without pricing, you’re missing a fundamental part of the marketing mix, and one that forces a great deal of discipline on both the customers and the vendor, in this case the internal IT team.
In a competitive market, pricing serves a lot of useful functions relating to efficient resource allocation. But internal IT tends to be a monopoly, so it prefers a centrally planned approach to the messier market-based emergent allocation approach. With no price signals to provide feedback, other mechanisms get used instead, like meetings, and these don’t work nearly as well.
A monopoly is a seller’s market, so project teams get charged whatever IT thinks the storage should cost, usually enough to pay back the cost of the array, sometimes including network ports and other supply-chain components, maybe an allocation of admin FTE. But cost recovery doesn’t drive efficiency, and it provides no incentives for good customer service or product design.
Instead of being able to pick whatever tools and services they prefer (provided it fits within a budget), developers are forced to use services designed by infrastructure specialists with capacity management goals. The result tends to be two unhappy groups of people, both of them incentivised to keep each other miserable.
To really get the value of something like Portworx, organisations need to adopt completely different ways of working, and this is where I become more hopeful.
To use Portworx at all, you already need to have bought into the idea of Kubernetes. Kubernetes inherently has a more developer-centric view of the world, and while this has its own challenges, the fact that you’ve already made a choice to think in this way removes many of the barriers that have held back internal IT teams in the past.
Rob Lee, Pure Storage CTO told me that their customers are less interested in the hand-carved, everything is special approach from years past.
“There’s just more demand to say I don’t want to have islands of special purpose infrastructure that are designed for this application soup to nuts and that’s the only thing that runs there. You know?” Lee said.
A more cloud-like mindset means you’ve already moved a long way into a more operational view of infrastructure, rather than a Build It Once mindset. The developer world is constantly changing. Software is always being recompiled, torn down, re-deployed. Containers are constantly in flux. Applications move around, scaling up and down with demand or whim. Being flexible and adaptable to customer demands is far more familiar to people in this world.
Which is why I think that, if any group can build a customer-centric managed-service-provider model on top of a product like Portworx Data Services, its these groups. The track record of internal service catalogs might not be great, but we know from public cloud (and managed service provider businesses everywhere) that is is possible to succeed.
Now we have to see if Pure Storage customers are better than the average IT team or not.
Disclosure: Pure Storage is a PivotNine client.