October 2017 Summaries

Dave Malouf presents his thinking on DesignOps, a growing area of concern for design teams aiming to increase the value they produce. Unlike DevOps, which focuses on operationalizing development processes, DesignOps seeks to operationalize the design process itself. It involves using tools and practices unique to designers, such as graphic tools, vector and raster formats, and iterative exploration methods, to improve collaboration and communication with engineering teams. However, this can lead to inconsistencies in usage and a lack of standardization, particularly when working with code and other development systems. DesignOps also encompasses human resource management, asset management, and research practices, making it a complex field that requires different solutions depending on organizational and team cultures. DigitalOcean's specific DesignOps philosophy and processes will be explored in future blog posts, and the company is hosting the first-ever DesignOps Summit in NYC this November.

DigitalOcean has been running a summer internship program for the past few years, which provides opportunities for students to work with various teams across the organization. The first group of interns, known as "minnows," worked together in teams and contributed to projects such as improving One Click Apps and developing a feature to provide users with webhooks for alerts related to their Droplets. This program was designed to pay it forward to young technologists who want to work in the cloud and help build the next set of great companies. The company has invested significant time and resources into monitoring for Distributed Denial of Service (DoS) attacks and creating a system to track this data, known as the Flood Monitoring System (FMS), which uses technologies such as Apache Kafka, Apache Hadoop, and Golang.

Click below to sign up and get $200 of credit to try our products over 60 days!Sign up` High CPU Droplet plans are now available in Singapore, offering improved performance for CPU-intensive workloads such as CI/CD servers, data analytics applications, and app servers. These plans are designed to provide more powerful underlying computing power, with options starting from $40/mo for two dedicated vCPUs. The new droplet plans have been optimized with Intel's most powerful processors, delivering reliable performance and up to four times the performance of standard droplets. Customers in the early access period have seen significant improvements, averaging 2.5 times the performance of standard CPUs.

The DigitalOcean network engineering team has been working on building a global backbone network to improve the stability, performance, and reliability of their cloud platform. The existing network infrastructure was composed of independent "island" networks, which resulted in communication between regions relying on public internet connectivity, prone to issues such as congestion, malicious attacks, and ISP technical problems. To mitigate this risk, they are building a backbone network using technologies like dark fiber and wavelengths, allowing traffic between locations to run over dedicated links managed by DigitalOcean. This gives customers predictable and reliable transport between regions, essential for mission-critical applications. The team has already interconnected several locations, including NYC, European locations, and transatlantic links, with plans to expand further, aiming to create a seamless and high-performing network infrastructure for their users.

The DigitalOcean team has been using a mono repo called cthulhu to organize their Go code for nearly three years. The repository, which started with a single directory structure, has grown exponentially and introduced various challenges, including tooling and organizational issues. Despite these challenges, the mono repo approach has provided several benefits, such as being able to make large, cross-cutting changes without breaking downstream repositories. To address some of the issues, the team developed a build tool called `gta`, which inspects the git history to determine which files must be tested for a given build. This tool significantly reduced the time taken by builds and improved the overall quality of their Go code through static analysis checks and custom linting tools. However, maintaining the repository's structure and tooling also presents challenges, such as establishing ownership and dealing with code vendoring issues. The team is actively working to improve cthulhu and its build pipelines, alongside regular project work.

DigitalOcean co-founder Ben Uretsky recently discussed the company's humble beginnings and its growth over the past six years in an interview with Mitch Wainer on The Deep End podcast. DigitalOcean was started by Ben and three other co-founders in 2011, initially managed by Jeff Carr, who single-handedly scaled the service to nearly 100,000 customers. The company has since grown to become a global leader in cloud computing, with a team of over 350 people. Ben shared that the engineering team's evolution was crucial to DigitalOcean's growth, from a small team of one engineer to a more verticalized structure around each product. As CEO, Ben faces challenges related to people management and painting the right vision for the business, but emphasizes that companies are collections of people first and foremost. The company has also launched its Hatch incubator program, which provides infrastructure credit and technical support to startups globally.