The databases of Web enterprises

Almost no one use Oracle, so what is their better solution?

Their solution
Google Percolator for the search engine. Successor of BigTable, another software created by Google.
Spanner. Distributed and replicated in real time NewSQL DB, for GMail and others.
Facebook HBase. Derived from BigTable and is part of Hadoop, a set of tools.
MySQL. For common functions. With TAO overlay for the social graph.
Haystack. For images and other media.
Hive. Adds data analysis to Hadoop.
Schema: Infrastructure of Facebook for its databases. (Source Malte Schwarzkopf).
Twitter Cassandra for statistics and geolocation.
MySQL. A fork of the dbms, for users and tweets. The source code is available on GitHub.
FlockDB. A graph database for social connections. The source code is on GitHub.
HBase and Hadoop. Probably to replace MySQL.
Wikipedia MariaDB. Clusters of wikis with a common frontend. Programmed in PHP and JavaScript (30%).
Yahoo MySQL customized. The company had a time the size record for SQL tables.
PostgreSQL whose query part has been customized.
eBay Hadoop with Kylin, a system of search on mass data.
eBay had also used Oracle.
Digg DynamoDB, a NoSQL solution provided by AWS, the host.
Netflix Cassandra.
Skype PostgreSQL. Used with PHP and other scripting languages.
Youtube MySQL + Speed to scale it on the cloud.
Uber Migrated from PostgreSQL to MySQL (Percona) + Schemaless which turns it into NoSQL.
Uses also Cassandra, Hadoop, MongoDB and Redis under Node.js (source).

This is only a sample of the best known companies. A lot of others use PostgreSQL, MariaDB (replacing MySQL), HBase and Hadoop tools.
While Cassandra was abandoned by Facebook which created this software, it is however used by many other companies whose needs are different.
Amazon uses only Oracle, but the company also has the habit of paying without discussing the patent trolls, it is a consistent attitude.

We see that for structured data, enterprises use MariaDB (possibly clustered) and PostgreSQL. But they also use them in combination with other tools for network data.
However, other types of software are more suitable for a variety of web applications: NoSQL tables, documents, graphs, key-value, etc ...

See also ...