June 2014 Summaries

MongoDB was adopted by AHL as the single platform for all financial data, replacing traditional technologies and delivering significant improvements in speed, cost savings, and scalability. The adoption of MongoDB enabled AHL to support its quantitative researchers more effectively, allowing them to research, construct, and deploy new trading models in a more efficient and agile manner. With MongoDB, AHL was able to retrieve low-frequency data 100x faster, deliver consistent retrieval times, and reduce time to recompute trading models from hours to minutes. Additionally, MongoDB enabled the company to capture tick data at a rate of 250 million ticks per second, scaling quickly to meet the demands of high-frequency trading. The adoption of MongoDB also resulted in significant cost savings, with AHL reducing disk storage needs by 40% and realizing a 40X cost savings compared to its previous solution.

The key points covered in this text include modeling One-to-N relationships in MongoDB, which involves considering factors such as the cardinality of the relationship, whether the entities on the "N" side need to stand alone, and the ratio of reads to updates for a particular field. The three basic ways to model One-to-N relationships are embedding, referencing, and parent-referencing, each with its advantages and disadvantages. Denormalization can be used to improve read performance by duplicating data in multiple documents, but it also adds complexity to updates and requires careful consideration of the ratio of reads to updates. The text concludes that MongoDB provides flexibility and productivity for designing database schema, allowing applications to adapt easily to change and support queries and updates as needed.

The MongoDB Backup tool allows developers to restore their database from a corrupt or lost state by using backups to start over. Before restoring, it's essential to configure the backup delivery method and ensure sufficient disk space on the machine receiving the backup files. Once restored, data can be retrieved from stored snapshots or point-in-time restores within the last 24 hours. Additionally, for sharded clusters, cluster checkpoints can be configured to enable restore functionality. The process involves retrieving snapshots from the Sharded Cluster Status page, selecting the replica set or shard to restore, and clicking on the "restore this snapshot" link to initiate the restoration process.