With Infrastructure Software, Go SaaS or Go Home

Originally published in The New Stack 

The old Chinese proverb, “May you live in interesting times,” is certainly bearing out for the infrastructure technology market.

Recently, Boris Renski, co-founder of OpenStack vendor Mirantis, argued that the whole Infrastructure Software market is now dead (and that he has made his peace with it). To make his point, he highlights challenges with the traditional software delivery model:

  • OpenStack and infrastructure software are full of bugs.
  • Infrastructure software is hard to upgrade
  • Infrastructure software is a nightmare to operate

Phew. These problems are well-known among those who have experience with OpenStack, so it’s almost a relief that a leading OpenStack vendor has admitted they exist.

But, it’s not just about OpenStack. From traditional enterprise software to innovative technologies like big data/Hadoop and emerging containers, issues around operational complexity and quality are a fact of life for most non-trivial distributed systems.

Old Processes, Changing Expectations

Operational complexity is ultimately caused by quickly evolving software that has not been adequately hardened. At the core, it is a quality problem. For example:

  • Enterprise software vendors such as Microsoft, VMware and Red Hat work around it by investing in thousands of engineers and in millions of engineering man-hours in building and stabilizing major software releases. However, this only works to a point: 90 percent of engineering cycles must be spent stabilizing software, which means quality comes at the expense of innovation.
  • Innovators working on projects such as OpenStack, software defined storage, software defined network, container orchestration, platform-as-a-service and so on, struggle to balance rapid innovation with a successful and easy customer experience.
    There’s nothing new here. What’s changed is customer expectations about working with infrastructure software. We live in a “post-AWS” world where users now expect extreme ease of use, accelerated time to value, greater efficiency and more. They are less and less willing to accept the operational hassles associated with using and managing infrastructure software.

The key question is: Is there a way to deliver infrastructure software without the tradeoff? Can an infrastructure software vendor provide a great customer experience and all the latest innovations?

Is Professional Services the Answer?

In his post, Renski says Mirantis has been able to do so with its “hordes of talented infrastructure hackers that made up for the gaps.” In other words, a professional services model that uses human capital to alleviate — or at least mask — operational and quality issues.

However, a professional services model is not without drawbacks:

  • The talent and skill sets of individual hackers vary, which means different customers can end up with different outcomes for identical problem sets.
  • These talented infrastructure hackers are expensive (and hard to find), so it can ultimately make the solution far more expensive than other alternatives (even AWS!).
  • And if talented individual hackers are solving the same problem over and over at various customer sites, there’s just a lot of time being spent and wasted.

In short, throwing people at the problem feels like a brute-force approach rather than an elegant, repeatable solution. There has to be a better way to truly provide a great customer experience.

Software-as-a-Service (SaaS): A Better Path Forward

The case for SaaS has largely been made by Salesforce and AWS. Both have used SaaS to deliver their software elegantly to enterprises. At first, both were dismissed by industry incumbents who believed (and touted) that security concerns would prevent adoption. Fast forward and today Salesforce is the de facto standard among enterprise customers, and AWS is growing rapidly and disrupting legacy enterprise solutions.

SaaS fundamentally eliminates all major customer experience problems we’ve spoken about:

  • Bugs can be detected much faster and resolved globally across all affected customers.
  • Upgrades are “free” and happen magically behind the scenes.
  • Operating the solution is “free” and happens magically behind the scenes.

In other words, the SaaS model takes core software bits and packages them with great automation built by a team of developers and site reliability engineers (SREs) whose mission in life is to delight customers with a smooth and transparent experience.

Not unlike the early days of Salesforce, detractors point out perceived security issues with SaaS-based infrastructure software. However, in a world where the only workable alternative is the public cloud (with even greater security exposure), SaaS is a more than reasonable alternative for customers who want the advantages of new technologies without complexity.

For infrastructure software, SaaS is where rapid innovation and great customer experience meet.

Tell me what you think. @sraghuram

Sirish Raghuram

You may also enjoy

Migrating Off AWS: How to Move Production from AWS to Enterprise Data Centers Running OpenStack and Kubernetes

By Platform9

Kubernetes FinOps: Basics of cluster utilization

By Joe Thompson

The browser you are using is outdated. For the best experience please download or update your browser to one of the following: