JuMP: A modeling language for mathematical optimization

Monday, October 18, 2021

The JuMP logo.

As an author of the paper JuMP: A Modeling Language for Mathematical Optimization, I am honored to have recently received the Mathematical Optimization Society’s Beale—Orchard-Hays Prize, an academic award given once every three years for work in the area of computational mathematical optimization. The award, in fact, is about the open source software project JuMP, which I started with Iain Dunning and Joey Huchette while we were PhD students at MIT’s Operations Research Center almost nine years ago. The humbling milestone of the Beale—Orchard-Hays Prize seems like a good occasion to reflect on JuMP, how it has matured and grown as an independent community-driven project, and Google’s role in enabling me to serve as JuMP’s BDFL.

JuMP was created—in the classical open source fashion—to scratch an itch. As graduate students, we wanted a software package that would enable us to write down and solve optimization problems, especially constrained optimization problems like linear programming and integer programming problems. We wanted it to be not only easy, but also fast and powerful. At the time, one was faced with trade-offs between ease-of-use, speed, and flexibility. For example, optimization libraries in Python were user-friendly but introduced noticeable performance bottlenecks. Commercial software such as AMPL was efficient but hard to extend. Low-level interfaces in C or C++ introduced complexities that were distracting for teaching and academic research. We weren’t satisfied with these trade-offs, and began experimenting with a new programming language called Julia that promised to provide the best of both worlds.

Our early experiments showed that Julia was indeed capable of impressive performance. While similar libraries based on Python could be slower to construct the data structure describing the optimization problem than to solve it, our prototype of JuMP was competitive with state-of-the-art commercial libraries. This gave us confidence that JuMP could be useful for the community, and we made the initial public release in October 2013.

Since then, it’s been a real ride! The first JuMP developers workshop in 2017 attracted thirteen speakers from four continents; this year’s workshop featured 32 virtual talks. Of the 800+ citations to the award-winning paper, we were surprised to discover that that about 75% of them were from outside the fields of operations research or optimization itself; about 20% are in energy and power systems, another 20% are in control and engineering, and the remaining citations are spread across scientific applications, computer science, machine learning, and other fields. These figures speak to the role of optimization as a fundamental technology that can be applied almost anywhere. One example application using JuMP of which I’m perhaps most proud is a study by Sepulveda et al. on cost-effective ways to decarbonize the power grid. This study is cited both by Bill Gates in his new book, “How to Avoid a Climate Disaster,” and by Google’s methodologies and metrics framework for its goal of operating data centers and campuses entirely on carbon-free energy by 2030.

As JuMP’s core development team grew beyond MIT and its original creators graduated, it was important for JuMP to find a new home for its long-term sustainability. We were lucky to find NumFOCUS, a nonprofit organization supporting open source scientific software (of which Google is a corporate sponsor). As a Google employee, I have continued contributing code for JuMP, traveling to workshops, and serving in leadership roles thanks in no small part to Google’s generous open source policies and support from my team and management chain. Last year, I was granted the honorific of Benevolent Dictator for Life (BDFL). I plan to use this power judiciously and rarely, relying instead on JuMP’s strong culture of consensus-driven development.

As for the future, JuMP’s 1.0 release is near on the horizon, and I look forward to whatever comes next!

By Miles Lubin, Algorithms & Optimization Team, Google Research

Open Source in the 2021 Accelerate State of DevOps Report

Wednesday, September 22, 2021

To truly thrive, organizations need to adopt practices and capabilities that will lead them to performance improvements. Therefore, having access to data-driven insights and recommendations about the most effective and efficient ways to develop and deliver technology is critical. Over the past seven years, the DevOps Research and Assessment (DORA) has collected data from more than 32,000 industry professionals and used rigorous statistical analysis to deepen our understanding of the practices that lead to excellence in technology delivery and to powerful business outcomes.
One of the most valuable insights that has come from this research is the categorization of organizations on four different performance profiles (Elite, High, Medium, and Low) based on their performance on four software delivery metrics centered around throughput and stability - Deployment Frequency, Lead Time for Changes, Time to Restore Service and Change Failure Rate. We found that organizations that excel at these four metrics can be classified as elite performers while those that do not can be classified as low performers. See DevOps Research and Assessment (DORA) for a detailed description of these metrics and the different levels of organizational performance.

DevOps Research and Assessment (DORA) showing a detailed description of these metrics and the different levels of organizational performance

We have found that a number of technical capabilities are associated with improved continuous delivery performance. Our findings indicate that organizations that have incorporated loosely coupled architecture, continuous testing and integration, truck-based development, deployment automation, database change management, monitoring and observability and have leveraged open source technologies perform better than organizations that have not adopted these capabilities.

Now that you know a little bit about what DORA is and some of its key findings, let’s dive into whether the use of open source technologies within organizations impacts performance.

A quick Google search will yield hundreds (if not, thousands) of articles describing the myriad of ways organizations benefit from using open source software—faster innovation, higher quality products, stronger security, flexibility, ease of customization, etc. We know using open source software is the way to go, but until recently, we still had little empirical evidence demonstrating that its use is associated with improved organizational performance – until today.

This year, we surveyed 1,200 working professionals from a variety of industries around the globe about the factors that drive higher performance, including the use of open source software. Research from this year’s DORA report illustrates that low performing organizations have the highest use of proprietary software. In contrast, elite performers are 1.75 times more likely to make extensive use of open source components, libraries, and platforms. We also find that elite performers are 1.5 times more likely to have plans to expand their use of open source software compared to their low-performing counterparts. But, the question remains—does leveraging open source software impact an organization’s performance? Turns out the answer is, yes!

Our research also found that elite performers who meet their reliability targets are 2.4 times more likely to leverage open source technologies. We suspect that the original tenets of the open source movement of transparency and collaboration play a big role. Developers are less likely to waste time reinventing the wheel which allows them to spend more time innovating, they are able to leverage global talent instead of relying on the few people in their team or organization.

Technology transformations take time, effort, and resources. They also require organizations to make significant mental shifts. These shifts are easier when there is empirical evidence backing recommendations—organizations don’t have to take someone’s word for it, they can look at the data, look at the consistency of findings to know that success and improvement are in fact possible.

In addition to open source software, the 2021 Accelerate State of DevOps Report discusses a variety of capabilities and practices that drive performance. In the 2021 report, we also examined the effects of SRE best practices, the pandemic and burnout, the importance of quality documentation, and we revisited our exploration of leveraging the cloud. If you’d like to read the full report or any previous report, you can visit

Learn Kubernetes with Google: Join us live on October 6!

Tuesday, September 21, 2021


Graphic describing the Multi-cluster Services API functionalities

Kubernetes hasn’t stopped growing since it was released by Google as an open source project back in June 2014: from July 7, 2020 to a year later in 2021, there were 2,284 new contributors to the project1. And that’s not all: in 2020 alone, the Kubernetes project had 35 stable graduations2. These are 35 new features that are ready for production use in a Kubernetes environment. Looking at the CNCF Survey 2020, use of Kubernetes has increased to 83%, up from 78% in 2019. With these many new people joining the community, and the project gaining so much complexity: how can we make sure that Kubernetes remains accessible to everyone, including newcomers?

This is the question that inspired the creation of Learn Kubernetes with Google, a content program where we develop resources that explain how to make Kubernetes work best for you. At the Google Open Source Programs Office, we believe that increasing access for everyone starts by democratizing knowledge. This is why we started with a series of short videos that focus on specific Kubernetes topics, like the Gateway API, Migrating from Dockershim to Containerd, the Horizontal Pod Autoscaler, and many more topics!

Join us live

On October 6, 2021, we are launching a series of live events where you can interact live with Kubernetes experts from across the industry and ask questions—register now and join for free! “Think beyond the cluster: Multi-cluster support on Kubernetes” is a live panel that brings together the following experts:
  • Laura Lorenz - Software Engineer (Google) / Member of SIG Multicluster in the Kubernetes project
  • Tim Hockin - Software Engineer (Google) / Co-Chair of SIG Network in the Kubernetes project
  • Jeremy Olmsted-Thompson - Sr Staff software Engineer (Google) / Co-Chair of the SIG Multicluster in the Kubernetes project
  • Ricardo Rocha - Computing Engineer (CERN) / TOC Member at the CNCF
  • Paul Morie - Software Engineer (Apple) / Co-Chair of the SIG Multicluster in the Kubernetes project
Why is Multi-cluster support in Kubernetes important? Kubernetes has brought a unified method of managing applications and their infrastructure. Engineering your application to be a global service requires that you start thinking beyond a single cluster; yet, there are many challenges when deploying multiple clusters at a global scale. Multi-cluster has many advantages, it lets you minimize the latency and optimize it for the people consuming your application.

In this panel, we will review the history behind multi-cluster, why you should use it, how companies are deploying multi-cluster, and what are some efforts in upstream Kubernetes that are enabling it today. Check out the “Resources” tab on the event page to learn more about the Kubernetes MCS API and Join us on Oct 6!

By María Cruz, Program Manager – Google Open Source Programs Office

1 According to devstats

Kubernetes Community Annual Report 2020