Firstly, a huge thank you to all our customers that took the time to review ArangoDB for the Gartner Peer Insights “Voice of the Customer”: Operational Database Management Systems Market report. Without your help and assistance, the continued improvements and enhancements we make to our software wouldn’t be possible. Read more
Graph databases are often used to analyze relations within highly interconnected datasets. Social networks, recommendation engines, corporate hierarchies, fraud detection or querying a bill of materials are common use cases. But these datasets change over time and you as a developer or data scientist may want to time travel and analyze these changes.
While ArangoDB may not come with built-in support for managing the revision history of graph data, we’ll show in this article how to manage it in a performant manner for some general classes of graphs. Best of all, this won’t require any groundbreaking new ideas. We’ll simply borrow a few tools and tricks from the persistent data structure literature and adapt them for good performance within ArangoDB. We hope that this will help enable new ways to use everyone’s favorite avocado-fueled datastore, and power some useful applications. Read more
Many ArangoDB users rely on our
arangorestore tools as an integral part of their backup and recovery procedures. As such, we want to make the use of these tools, especially
arangodump, as fast as possible. We’ve been working hard toward this goal in preparation for the upcoming 3.4 release.
We’ve made a number of low-level server-side changes to significantly reduce overhead and improve throughput. Additionally, we’ve put some work into rewriting much of the code for the client tools to allow dumping and restoring collections in parallel, using a number of worker threads specified by
--threads n. Read more
Our recent release 3.3.9 includes several performance optimizations for data retrieval cases. Benefits can be expected for both storage engines, MMFiles and RocksDB, AQL batch lookup queries, and cluster AQL queries.
MMFiles index batch lookups
For the MMFiles engine, an optimization has been made for retrieving multiple documents from an index (hash index, skiplist index or persistent index) in a batch.
We were in search for some C++ reader/writer locks implementation that allows a thread to acquire a lock and then optionally pass it on to another thread. The C++11 and C++14 standard library lock implementations std::mutex and shared_mutex do not allow that (it would be undefined behaviour – by the way, it’s also undefined behaviour when doing this with the pthreads library).
Additionally, we were looking for locks that would neither prefer readers nor writers, so that there will be neither reader starvation nor writer starvation. And then, we wanted concurrently queued read and write requests that compete for the lock to be brought into some defined execution order. Ideally, queued operations that cannot instantly acquire the lock should be processed in approximately the same order in which they queued. Read more
In this article, we’re going to dive deeper into our recently released feature preview in Milestone ArangoDB 3.4 – ArangoSearch which provides a rich set of information retrieval capabilities. In particular, we’ll give you an idea of how our search engine works under the hood.
Essentially ArangoSearch consists of 2 components: A search engine and an integration layer. The former is responsible for managing the index, querying and scoring, whereas latter exposes search capabilities to the end user in a convenient way.
ArangoDB is a multi-model database written in C++. It is a sizable application with an executable size of 38MB (stripped) and quite some library dependencies. We provide binary packages for Linux, Windows and MacOS, and for Linux we cover all major distributions and their different versions, which makes our build and delivery pipeline extremely cluttered and awkward. At the beginning of this story, we needed approximately 12 hours just to build and publish a release, if everything goes well. This is the beginning of a tale to attack this problem. Read more