- Developer Tools
Brent Smithurst, November 5, 2013
Unless you're deeply embedded in the PaaS, OpenStack, or cloud computing industries (e.g. a vendor), you probably haven't heard of Project Solum yet. Many articles have been written about it over the past week and a half and most of them haven't been very complimentary. It's been described as everything from a home for failed PaaS vendors (by one of our advisors no less) to a Cloud Foundry competitor. The Twitter chatter I've seen has been almost overwhelmingly negative as well. Disdain for the project has even crossed over to criticism of OpenStack's lack of focus in an interesting manifesto of sorts by Andrew Clay Shafer.
Why would ActiveState want any part of this? Stackato is very successful; it's anything but a failed PaaS project. We have been winning large enterprise deals in direct competition with other PaaS vendors for the past 20 months and have no reason to change our direction. Furthermore, we're a happy member of the Cloud Foundry ecosystem and CAB (Community Advisory Board) and have no plans to switch to a different platform.
So, why in the world would we join Project Solum?
We had a talk with Rackspace
Adrian Otto and Rhett Dillingham of Rackspace were kind enough to get on the phone with us and talk about Solum and how it may evolve. They didn't describe it as a new PaaS project, but as a way for developers who use OpenStack to deploy applications and manage their lifecycle easier. That sounds similar to some of the benefits of PaaS but at a much lower level and it definitely piqued our interest.
Solum is just beginning
Solum has barely even gotten off the ground yet. Its architecture is still being defined and its community is still taking shape. Solum doesn't even describe itself as a PaaS. Is it going to be a PaaS or is it possibly something that will simply make it easier for PaaS vendors to leverage OpenStack? Why not get in on the ground floor to try and find out while influencing Solum's design? We don't see Solum as a threat to Stackato; rather, it's a potential enabler.
Solum has interesting partners
The Solum website currently has nine company logos on it as I write this. Four of them are already ActiveState partners in one way or another—Stackato v3.0 (coming very soon!) uses Docker from dotCloud, Rackspace is a Stackato hosting partner, Stackato is built on top of (among other things) Canonical's Ubuntu, and ActiveState and CloudSoft have collaborated in the past on hybrid cloud deployments.
OpenStack needs better Paas-readiness
Do you remember the "great OpenStack API debate"? If not, don't worry about it. I won't open up that can of worms again. The fact remains that if OpenStack doesn't support common APIs (such as EC2's), then products (such as Stackato) that support multiple IaaS platforms need to do extra work to support OpenStack.
Stackato is infrastructure agnostic
The key reason is that Stackato is infrastructure agnostic. It runs on anything. This isn't so much philosphical as it as pragmatism—Stackato has three main categories of customers: large enterprises running Stackato on their private cloud (almost exclusively vSphere with a smattering of CloudStack, OpenStack, or others thrown in); cloud hosting providers such as HP (almost exclusively OpenStack with some CloudStack); and smaller startups (often AWS or HP Cloud). We support all of these infrastructures equally but it is a lot of work keeping up with all the changes. If that could be made easier via better integration points, we'd be very happy (as would our customers).
Put it all together
Here's a quick example: we recently launched a Citrix CloudPortal Business Manager (CPBM) connector. It's primarily for use by Citrix CloudPlatform users. It leverages parts of CloudPlatform that make it much easier for Stackato to be easily provisioned and configured on it (plus more). We could build those capabilities into Stackato or we could leverage CPBM's capabilities. Leveraging CPBM is more scalable and sensible for us; we can more quickly, easily, and reliably support multiple IaaS platforms if we use their APIs. Deeper integration points are desired by us and Citrix; Citrix is helping with this by building new capabilities into future versions of CPBM so we can take advantage of them.
That same thing holds true for OpenStack or any other IaaS platform. If the integration points exist in their APIs, then there is not only less work for us to do, but we can do it quicker and more reliably.
OpenStack gets all the buzz but vSphere has the bulk of enterprise users. Our desire to support multiple infrastructures is rooted in the needs of our customers. They may currently use a single IaaS platform but most of them are at least interested in keeping their options open—either they may want to switch their IaaS in the future or they may want to use more than one at the same time (hybrid cloud). In fact, a good enterprise PaaS like Stackato can enable the ability to switch IaaS platforms. Stackato abstracts the infrastructure away and can span different infrastructures, whether private or public. That lets IT staff and developers continue to use the PaaS to be productive even if the underlying infrastructure changes.
That's why we're interested in Project Solum. We currently hook into several of OpenStack's individual projects but there is no central OpenStack project that manages the integration points a true PaaS system like Stackato requires. Solum's blueprints so far would make it easier and more efficient for Stackato to leverage OpenStack because Solum in turn leverages Heat, Nova, Glance, Keystone, and more. The promise of doing so in a standardized and predictable manner is enough for us to happily contribute our knowledge and code in order to benefit our customers and the OpenStack ecosystem. The only concern we have is that Solum may be a distraction to those who may be better off focusing their energies on the core issues that prevent faster adoption of OpenStack by enterprises.
We're not worried that Solum will compete directly with Stackato because Stackato solves a wider range of problems for enterprises than just deploying application code onto OpenStack. It may eventually compete, but that's a long way in the future. It's also important to remember that Solum is OpenStack-exclusive. For it to be a real threat to Stackato, all of our customers would need to be able to use it and that's certainly not the case today since the vast majority of them don't use OpenStack. In the meantime, we're keeping our minds open and look forward to contributing.