The beauty of a "hybrid" database system is that it allows you to manage both XML and relational data at the same time, even side-by-side in the same table if you want. You can choose which data to store in XML and which in relational format. Best practices for determining the "right" mix of XML and relational in a database schema is an interesting topic, but I'll reserve that for a separate blog post in the near future.
The XML Database Blog
MatthiasNicola 120000E28R Tags:  xml relational update xquery database hybdrid purexml db2 3,333 Views
MatthiasNicola 120000E28R Tags:  db2 benchmark intel performance xeon westmere xml tpox purexml 4,652 Views
Intel has recently launched their new “Westmere-EX” CPUs that are now part of the Intel Xeon E7 processor family. For 5 years now Intel has a strong history of testing and showcasing many of their latest processors with the Transaction Processing over XML (TPoX) benchmark. So it comes as no big surprise that Intel has now also released TPoX benchmark results with their latest 10-core Xeon E7-4870 processor.
This benchmark uses TPoX Version 2.0 with 1TB of raw XML data, running on DB2 9.7 under SUSE Linux 10. DB2 compression reduces the 1TB of raw data to 537GB, which includes all required XML indexes.
The system under test is an Intel Xeon E7-4870 processor server with 4 CPUs, 40 cores, 80 threads, and a clock speed of 2.4GHz. Further details on the CPUs can be found here. The storage configuration consists of 54 Intel X-25E Solid State Drives (SSDs), 48 SSDs for the database tables and indexes plus 6 SSDs for the DB2 transaction log.
To showcase the maximum possible TPoX throughput on this hardware, Intel ran the TPoX workload driver with 440 concurrent connections (“users”) all of which issue a random stream of read and write transactions without think time. The result is a whopping 17,757 TTPS (TPoX Transactions per Second)!
This result constitutes 29% higher performance than on the previous generation CPU, the 8-core Nehalem-EX, which achieved 13,745 TTPS. Details on these TPoX benchmark results are available on the TPoX results page where you find a history of all TPoX results as well as examples of other TPoX usage.
The great thing about these benchmarks is that they continuously
push hardware, operating system, and database to the edge, relentlessly
exposing any inefficiencies that may exist and giving us the chance to
continuously improve the technology. This latest TPoX benchmark is also
one of the first TPoX tests that runs entirely on solid state drives!