OID Supports 400,000 (Four Hundred Thousand!!) Operations Per Second on 500 Million User Database. AKA OID Eats Facebook Database For Breakfast
By Mark Wilcox - CTO - Oracle Consulting Security-Oracle on Mar 28, 2011
- ability to scale to extremely large number of entries without needing to split the entry database into multiple instances (called partitions)
- can leverage Oracle Exadata database machines
And we have put this together into a new whitepaper. In this white paper we showed how we got OID 11g on an Exadata machine containing 500 million entries (e.g. roughly the size of Facebook's 2010 user population) - were able to get it to 400K operations per second. And that was only on a Exadata half-rack. With 10gb Ethernet (as opposed to Infiband). Meaning - we're not even close to maxing out the performance here. Yet we're easily lapping the field and doing it with less management overhead. Or in other words - if you find you need a new large-scale directory, there really isn't any reason to be looking at any other directory vendor. Because no other vendor gives you two proven options to scale to these numbers. You can either choose to scale horizontally by data partitioning using DSEE leveraging existing commodity hardware or you can avoid partitioning by utilizing the power of Oracle RDBMS with OID with or without the unique capabilities of Oracle Exadata.