Streaming Music Services and Databases

Posted by Alex Slotnick on Jul 29, 2015 5:09:00 AM

The average American consumer isn’t usually interested in “buying” a big database. But thanks to the recent boom in the popularity of subscription music services like Spotify, databases are competing for folks’ cash more directly than ever before. After all, what are these services other than giant banks of media, made accessible and sleek? The summer of 2015 has brought the battle for music-lovers’ wallets to fever pitch, with the launch of Apple Music and Tidal, both poised to challenge the reign of top-dog Spotify.

Music_Services_and_Databases

In terms of content, all three services are pretty similar. They share a ton of overlap. Unless a consumer has an overwhelming love for Mrs. Carter or Miss Swift, or base your choice solely on cost, it’s tough to say what would sway somebody toward one service over another. Here at VividCortex, we can’t tell you which subscription service is sexiest or which offers most decibels per dollar, but we thought it would be fascinating to take a look at how each company presents its technical product to the public. How much does each service flaunt itself as a database technology? How much does each service attempt to keep its backend concealed (i.e. more like Taylor than like Beyoncé).

Apple Music is the most immediately interesting in these regards. Of the three services, Apple’s is the newest, launched just last month, though Apple Music comes as the latest development in the company’s long lineage of music database products. iTunes has been around well over a decade and has gone through many different iterations. Early iTunes libraries existed completely offline, requiring users to store their media locally and organize it however they saw fit. Because users have become accustomed to using an Apple platform to index their music collections, one might assume that transitioning to Apple Music’s streaming service would be smooth, painless, intuitive. Unfortunately, no, and early snafus with the new service have been all the more jarring by contrast. Whereas iTunes allowed a user total control over his or her database organization, inaccuracies within Apple Music’s iCloud database have actually interfered with some users’ personal organizational metrics when those users have tried to sync their libraries with the library in the cloud.

Additionaly, it is extremely difficult to find information on the structure of Apple Music’s backend database. Apple, of course, is a massive, massively profitable corporation, and much of the technology they use is proprietary information, kept behind a veil. However, in the midst of competition, a gesture toward transparency could appeal to customers, especially in the case of music, a form of media with which many people have personal bonds. Even if it’s only a peak under the hood.

The best public information we could find about the details of Apple’s database technology was in a ZDnet article from 2011, which, in turn, reports its information based on insights retrieved from a job posting for a position at one of Apple’s data centers in Maiden, North Carolina. “Database architects require skills in the following: RDBMS: Oracle, Teradata, MySQL, DB2; NoSQL DB: Hadoop, HDFS, mongoDB, Cassandra; Columnar DB: Vertica, SAP Hana.” Helpful? Not really.

Spotify stands in stark contrast to Apple Music: check out Spotify Labs. As recently as last month, this blog published an in-depth article detailing how the Spotify team switched its user database from PostgreSQL to Cassandra. It’s really an admirably detailed account, going so far as to explain the step-by-step method for migrating info, and to explore Cassandra’s LWT “Lightweight Transactions” feature, which Spotify employs to guarantee a user’s uniqueness.

Other articles on Spotify’s blog are equally knowledgeable. Here’s a piece from 2014 about how the Spotify database randomizes playlists (spoiler: it’s not actually random, because true randomness is not what customers actually want), and in 2013 they published a piece called “In praise of ‘boring’ technology,” which champions mature and proven tech such as MySQL and PostgreSQL, a sentiment we share here at VividCortex. It’s hard to say exactly how this kind of technical transparency might translate to the quality of Spotify’s final product, but it’s fun to look at nonetheless.

And what tech specs are readily available for Tidal? Beyond the fact that it’s managed by the nordic Aspiro Group, owned by Jay Z, absolutely nada. Perhaps maintaining magic and mystique goes along with Tidal’s VIP vibe. Hov probably knows best.

One thing is certain: streaming is the new way to listen to music, even if it requires the user to surrender some personal customization and control. The coming months will show which methods, features, and positioning the public finds most appealing. In regards to transparency, it comes as no surprise that we favor Spotify’s approach. We believe that as industry leaders share their methods, the community will learn and only grow stronger. Here’s a look at how we scale our own backend systems, and some insight on how to build your own database driven app with Go. And, of course, if Apple would like assistance monitoring their MySQL and MongoDB databases, we’d be happy to help.

Pic Cred - ParlourCafe.com

Recent Posts

Posts by Topic

see all