March 2020 Summaries

Meet our Product Design Team: Part 1 | MongoDB Blog` The MongoDB design team is composed of individuals from various backgrounds and disciplines, united by a shared design philosophy that emphasizes clarity, integrity, inclusiveness, consistency, and automation. The team collaborates on projects that aim to create user experiences for users worldwide, with a focus on rigor, passion, and impact. The designers featured in this blog post share their personal stories, career highlights, and favorite team activities, offering insights into the company culture and design process. They discuss their experiences working together, sharing knowledge and skills gained from college, and overcoming cultural differences to create inclusive designs for global customers. The designers also talk about their hobbies outside of work, such as cooking and reading fantasy and sci-fi books. Overall, this blog post provides a glimpse into the collaborative design process at MongoDB, highlighting the team's commitment to creating user-centered designs that make a positive impact.

The document data model has become widely adopted due to its intuitive nature, flexibility, and resilience, making it ideal for developers working on microservices-based applications. This allows for higher development velocity, reduced code to write, and improved performance. The flexibility of the document model eliminates complex inter-group dependencies that have traditionally slowed developers down, allowing them to work autonomously and move at speed. In contrast, traditional relational databases require coordination with database administrators, ORM layers, and schema modifications, imposing friction on the development process. Empirical evidence shows that this can lead to production problems, slower resolution of issues, and reduced developer autonomy. The document model's flexibility allows for self-describing documents, dynamic schema changes, and polymorphic fields, making it well-suited for agile and DevOps practices, where continuous deployment and autonomous teams are common. Various companies, including Travelers Insurance, HSBC, and Toyota Material Handling, have successfully adopted the document data model to improve developer productivity and autonomy.

The MongoDB Women's Group in India was formed by Neha Mukherjee and Palki Sood, two employees who shared a vision of creating a support network for women employees. The group aims to build trust, inspire each other, and address common issues faced by women in the workplace. With the help of their office leader Amit Babbar, they organized monthly events, including fireside chats with inspiring speakers, which have been well-received by employees. The group has also hosted various events, such as breast cancer awareness sessions, movie nights, and self-defense classes, to promote a positive culture and support women's empowerment. The feedback from employees has been overwhelmingly positive, with many appreciating the platform for building new friendships and addressing common issues. The group is proud of its partnership, diversity, and inclusive culture, which values employees' success and well-being.

Toyota Material Handling Europe, a leading manufacturer of warehouse equipment, has adopted a microservices approach to develop Internet of Things software for its smart factories. The company is building both the machines and the data platforms necessary for Industry 4.0, with a focus on quality and innovation. Filip Dadgar, Principal System Architect at Toyota Material Handling Europe, discussed the shift from monolithic codebases to microservices, citing MongoDB Atlas as the database platform that enabled this change due to its flexible data model, automatic scalability, security, and ease of use. The company evaluated eight key criteria before selecting MongoDB Atlas, including performance, developer friendliness, and cloud agnosticism. Toyota Material Handling Europe is now using MongoDB Atlas on Microsoft Azure, with access to MongoDB University for developer training and support from the MongoDB team.

The video discusses Toyota Material Handling's transition to a smart factory by adopting microservices, Microsoft Azure, and MongoDB Atlas. The company aims to become the industry leader in the fourth industrial revolution era. With integrated telepresence, they connected over 100,000 warehouse trucks, which are used in almost every business. The development team transformed their monolithic codebase into a microservices architecture to increase productivity and make the platform more user-friendly. They chose MongoDB Atlas for its complete managed global cloud database service, which provided a natural JSON document format for data modeling, allowing developers to work efficiently without being tied to specific technical requirements. The company evaluated various criteria before choosing MongoDB Atlas, including performance, scalability, security, regionality, backup and recovery, customer flexibility, developer-friendliness, and ecosystem. Toyota Material Handling's decision was influenced by the team's support and educational resources provided by MongoDB University.

The past two decades have seen significant advancements in storage hardware performance, with the introduction of solid state drives (SSD), transition from SATA to PCIe, and innovations in non-volatile memory technology. However, as devices' access times shrink, the cost of navigating software stack overhead becomes more prominent. Recent studies have shown that file system and other OS overhead still dominates I/O costs in very fast storage devices like Storage Class Memory (SCM). To address this challenge, academics proposed a new user-level file system, SplitFS, which reduces these overheads but is not viable for commercial products due to portability concerns. Fortunately, MongoDB's WiredTiger storage engine can be optimized to improve I/O performance by using memory-mapped files and batching expensive file system operations. By implementing these changes, WiredTiger achieved up to 63% performance improvements on mainstream SSDs.

To achieve optimal performance with MongoDB, it's essential to model your benchmark using representative data, queries, and deployment environment. This includes considering multiple parallel threads for sharded clusters and bulk writes to reduce network overhead. Pre-splitting chunks before loading data can also improve performance by allowing documents to be loaded in parallel into appropriate shards. Additionally, designing the data load such that different shard key values are inserted in parallel can help distribute writes across multiple shards. Other considerations include disabling the balancer during bulk loads, priming the system with representative queries for several minutes, using connection pools and configuring ulimits, and monitoring everything to locate bottlenecks. MongoDB Atlas provides features like charts, custom dashboards, automated alerting, and visualized metrics to help identify performance issues and optimize database cluster health.

Meet Some of MongoDB’s Influential Women Leaders in 2020, highlighting the careers and experiences of four women who have made a significant impact at the company. Yulia Genkina, Curriculum Software Engineer, creates online courses for MongoDB University and speaks at conferences; Shalena O'Connell, VP Operations, leads the Global Deal Desk team and has grown her responsibilities in operations and commercialization; Lorna Malone, Sales Development Manager, focuses on growing the footprint of MongoDB in Northern Europe and developing top talent; Jahira Martins, Automation & Marketing Operations Manager, manages technical program management and has helped shape standards for marketing automation and operations. These women leaders share their experiences, leadership styles, and what drives them, showcasing the diversity and inclusivity of the company's culture.

MongoDB has joined the GitHub Student Developer Pack, making its database service, MongoDB Atlas, available to students through this program. This partnership aims to provide students with a low-cost and accessible way to launch, manage, and scale their databases, as well as receive certification and credits. The MongoDB Student Pack offering includes $50 in free MongoDB Atlas credits and a free certification worth $150. By partnering with GitHub Education, MongoDB is making its technology more accessible to the next generation of developers.

Performance Best Practices: Hardware and OS Configuration` is a crucial aspect of achieving optimal performance for MongoDB, especially in large-scale deployments. To ensure that the working set fits within RAM, it's essential to have sufficient RAM size, as other optimizations may not significantly improve performance if there's insufficient RAM. Using multiple CPU cores can also lead to better performance, with each core represented by its own thread and background worker threads performing tasks like checkpointing and cache eviction. Running one mongod process per host is recommended for best performance, and using virtualization or container technologies can enable multiple processes to safely run on a single physical server without contending for resources. Configuring the WiredTiger Cache is also critical, with the size of the cache being tunable through the `storage.wiredTiger.engineConfig.cacheSizeGB` setting. Using multiple query routers and interleave policy on NUMA architecture can further improve performance, while network compression can reduce bandwidth costs in bandwidth-constrained environments. Storage and disk I/O considerations are also essential, with high-performance storage, SSDs for IO-intensive applications, and RAID-10 configurations recommended for most MongoDB deployments. Additionally, using MongoDB's default compression for storage and I/O-intensive workloads can provide significant performance benefits.