On Thursday 12th September I co-presented a webinar on how MySQL Cluster delivers the key benefits of NoSQL Data Stores without having to give up the features that people rely on from relational databases (consistency, SQL etc.). There was a particular focus on how to use the new node.js JavaScript API which was recently released as part of MySQL Cluster 7.3. If you weren’t able to attend the live event then the webinar replay is available here. If you scroll down to the bottom of this post you’ll see a summary of the Questions and Answers from the session.

The new MySQL JavaScript Connector for Node.js is designed to deliver simplicity and high performance for JavaScript users.

Enables end-to-end JavaScript development, from the browser to the server and now to the world’s most popular open source database.

Provides native NoSQL access to the MySQL Cluster storage engine without first going through SQL transformations and parsing

This session gives you everything you need in order to get started on building highly scalable, event-driven Web, mobile, and social applications.

It discusses:

•  Implementation of the connector
• Database connection and metadata management
• Creating and executing queries
• Use with InnoDB (via an internal SQL conversion) and MySQL Cluster (direct)

WHO:

• Andrew Morgan, Principal Product Manager
• John Duncan, Senior Software Developer
• Craig Russell, Principal Software Developer

Q & A

• What is the benefit of using an asynchrous run-time model? The async everywhere idea in Node.JS means the one execution thread is always doing work for a user. Never waiting on I/O. It allows you to minimize the cost of waiting on the network for i/o, and that’s a big part of how it can handle so many simultaneous requests.
• Can you please ellaborate more about multi-threading in node.js? Node.js has a just one thread that runs JavaScript. Then there is also a small thread pool that handles some background i/o tasks.
• Why can’t you use a synchrous model? Would there be any drawbacks? The async programming style can be hard to get used to … but in JavaScript, async is really “how things are done”. Just like browser-side JavaScript responding to a mouse event, this is server-side JS responding to a database event or a network event.
• Is there also a synchronous Session function? There is a synchronous version of connect(). There are some “immediate” calls on session (which don’t have to do any waiting), but there are no synchronous calls on session that wait for anything
• Most applications run as responses to requests and so most of the logic needs to be executed sequentially. How are asynchronous calls handled? Is there any mechanism to wait and join multiple asynch call for a particular section of logic? If you have multiple database operations to execute, you can use the batch feature which will wait for all operations to complete and then call back. If you have other operations you need to have your own synchronization mechanism.
• It was mentioned that you can use this APi to access the MySQL Cluster data nodes directly or MyISAM/InnoDB data via SQL and a MySQL server – how do I configure that? In the connection properties for your database connection, you use adapter=ndb for a native NDB connection, or adapter=mysql for a MySQL connection.
• Are results mapped into objects? Yes. When the inner layer reads a row from the database, it creates a JavaScript object (according to the mapping) and it calls your Constructor with no arguments and with this set to the newly created object.
• So there is seperate mapping for tables and for results? No, a single mapping.
• Does the object mapping support the entity relationships like in JPA implementations? Can it be referenced with JSON ojects with one-to-one, one-to-many, etc relationships? The current release does not support relationships/JOINs.
• JavaScript is weakly typed. How are different data types handled with the same reffrence? Dynamic typing means that values (objects, exspanssions) have types (i.e. a set of available operations and properties) but not variables, so a variable can, over time, refer to values of different types.
• Are there sql like transaction through the Node.js NoSQL API? Yes, the API looks something like this: session.currentTransaction().begin() … define operations … session.currentTransaction.commit()
• So, we can use session to track any transaction? Right, a session owns its single transaction. If you don’t call begin() on the transaction, then your operation will run “auto-commit”.
• Does adapter=mysql option use https://github.com/felixge/node-mysql‎ adapter adapter? Yes.
• Is this library similar to what mongoose is to mongodb? I’m not very familiar with mongoose — but it is broadly similar to many object-to-database mapping layers.
• Is there a working “Hello World” example showing all of these technologies as a big happy family? You could start with this… Using JavaScript and Node.js with MySQL Cluster – First steps
• So, just for clarification, the chief advantage of this is it’s moving the heavy lifting from the server side (like with PHP) to the client side? Not quite, node.js is server-side javascript. The main advantage is an asynchronous processing of requests that never block. In contrast, a large number of server-threads need to be provided with requests that can block.
• Node.js runs on V8 which developed by google, its run fast any browser or only in Chrome? Node.js actually runs in the app server rather than in your browser and so it will play nicely when accessed via any browser (or other client)