Announcing Wolfram Enterprise Private Cloud
September 19, 2016 — Conrad Wolfram, Strategic Director
Today I’m pleased to announce Wolfram Enterprise Private Cloud (EPC), which takes the unique benefits of the Wolfram technology stack—ultimate computation, integrated language and deployment—and makes them available in a centralized, private, secure enterprise solution.
In essence, EPC enables you to put computation at the heart of your infrastructure and in turn deliver a complete enterprise computation solution for your organization.
There are two strands to this blog post: what we’re delivering and why you’d want an enterprise computation solution and strategy.
Here’s how we got to today. For a few years, one of our key directions at Wolfram has been to build computation as a cloud service, so high-level computation (and computation-rich development) can be deliverable to everyone with the convenience of a cloud deployment. A couple of years ago we delivered our public cloud, a manifestation of core Wolfram technology delivered as a consumer service for professionals and individuals, but hosted by us.
EPC is the enterprise “privatization” with enhanced capabilities—taking this public Wolfram Cloud and packaging it up for hosting on any organization’s infrastructure or a designate like Amazon’s EC2. Instead of us offering the computation cloud service, you can, all within your enterprise. That means all the computation of Mathematica 11 and rapid application development of the Wolfram Language can now be server-side and cloud-based in your organization. High-level computation (for example, applied to your private data) can be an instant, ready-to-go, secure internal service for anyone you choose, with a wide range of interface modalities you can use directly for deploying from CEOs to developers, and instant APIs to go through other applications too.
Let me also point out the key principle that I believe marks out our technology as uniquely suitable for this centralized computation service model: we’re a unified, all-in-one system, not a collection of different systems for different tasks. We’ve put together all computational fields and functionality into one high-level, coherent Wolfram Language. We’re enabling complete interconnectedness. In a cloud-based service, lots of different systems means lots of separate “computational servers” to do different things—stats, reporting, modeling—causing huge switching losses, and that’s once you’ve got them and kept them playing together at all for a given task or workflow. Disparate systems are a real killer for broad, computation-based productivity.
That’s why our technology is in general so suitable for a private cloud manifestation.
We’re also adding many technologies we’ve implemented specially for EPC—from pre-warmed APIs to intra-node parallelization.
In the end, EPC has one objective: to enable computation everywhere in your organization, whether through ease of advanced access by traditionally computational groups or newfound access outside those groups.
So at one level, Wolfram Enterprise Private Cloud is an (exciting) new product. But at another, I believe it’s something much more significant: the start of a fundamental shift in how organizations see and deliver computation for the enterprise.
What do I mean by this? Until recently, the use of high-level computation has only been accessible to a small number of specialists in most organizations. If you’re not one of them, you really had three options: use basic computation (like Excel) yourself; rely on preordained, heavily collimated uses of computation; or seek out a specialist to build something custom or give you a one-off answer.
But computation is now very central to a huge number of organizational functions and the organization itself. It isn’t just for the specialist, many layers removed from the CEO; it’s too core for that. So likewise, it’s important to have an architecture for computation that matches this new reality. That means quality, security, command-and-control, coherence and consistent technology ability for computation need to be enterprise functions, not each decided ad hoc for each use or by each user. I’m describing the need for every organization to look at their enterprise computation strategy (which you can see explained more in our short video piece).
Here’s a typical example I come across. I’m visiting a bank and they ask, “Can the Wolfram Language make DLLs [dynamic link libraries] for Excel?” Digging into this request further, I find out that R&D is using the Wolfram Language for building prototypes that traders want to use through their familiar Excel interface. They’d like to package the DLLs up to hand to each trader instead of recoding. I ask, “But what happens if that R&D code has got a bug and the trader goes on running it? Or leaves and has taken a copy with them? How quickly can you even deploy this in practice? How is this wired directly to management reporting?”
The bank’s question betrays an “individual computation” way to think about the problem. The “enterprise computation” way would instead be for R&D to host an API on the private cloud to connect to an Excel interface. R&D can update the cloud deployment anytime—there’s no DLL to be updated and redistributed, there’s no choice needed between computational ability and interface, there’s no translation or installation; the quality, tracking, auditing and security models are much easier to govern.
One key driver for enterprise computation is big data—you could even say big data is a killer first reason for enterprise computation. So many organizations now state that failing to get the best answers from their data is a core business-strategic issue. They have amassed huge amounts of data but not effective, imaginative and broad-based analytics and visualization. Data and analytics mustn’t be siloed but need to be diced between groups; a data analytics hub is needed. (Watch my live and interactive 2015 Thinking Digital talk about decisions and data and computation.)
When data analytics was a specialist function in organizations, using desktop software—ours particularly!—matched up fine. But now data analytics is a shared enterprise problem; you need to match it with a shared enterprise computation solution—starting with EPC. Only an enterprise model, not an individual desktop one, can sort out data analytics failings.
This change to an enterprise model is new to general computation but not to previous technological progress. Often there’s a question of whether the powerhouse should be distributed or centralized.
Think electrical power. In the very early days (mid-19th century), each user pretty much generated their own. Then centralized power stations were found more effective and efficient at delivering the widely varying requirements of each user. But to reach most people, they depended on a network, technological and engineering progress (e.g. transformers) and standardization so everything interoperated (e.g. the power grid). It may be now with photovoltaic cells and other small-scale power generation opportunities that we’re entering a hybridized power generation future with an optimized mixture of local and centralized generation combined.
With computing, we’ve flip-flopped from mainframe to PC and now to the hybridization of local and cloud computing—the web providing necessary networking standardization to make this a practical reality.
Yet since the mainframe, the high-end computation part of computing hadn’t adopted an enterprise or hybridized architecture. That’s the change we’re starting today with EPC: enterprise as well as local computation—elevating computation to a core service. Much more will follow from us, including a complete hybridized enterprise computation ecosystem.
One consequence I’m very happy about: how EPC empowers our many Wolfram technology enthusiasts to get colleagues’ and management’s attention for their great, innovative work. Almost any Wolfram Language results that have stayed local can now be deployed (all within organizational security policies) as ready-to-use computational power and knowledge-based programs to anyone with a web browser. EPC gives our existing users (and me!) a terrific answer to questions like “How do I use my Wolfram Language code in a production environment?”
EPC can deliver many things we’ve been asked for, but it can go further by resetting thinking about computation.
In particular, I’m finding real excitement in early briefings to CTOs, CIOs and others concerned with technology strategy about this architectural shift and EPC. Not all couched their current infrastructural challenges in these terms, but most agree they do need a much more coherent enterprise computation strategy moving forward.
That’s what Wolfram Enterprise Private Cloud and Wolfram can get you started with today.