Schooner announced initial benchmarks of SchoonerSQL™ against the new Percona XtraDB Cluster. Percona XtraDB Cluster is a distribution of Percona Server with the Galera HA toolkit.
Standard benchmarks are a proxy for the performance that users may see with their own workloads. Schooner ran the standard DBT2 benchmark, which is the open-source OLTP version of the TPC-C benchmark, using a heavy workload of 1,000 warehouses with 32 concurrent connections to the database and zero think time. Measurements in each case were made using a typical two-node master-slave configuration. Each node was a standard IBM x86 server with 12/24 processor cores/threads, 72 GB of DRAM, and two 320 GB Fusion-io ioDriveDuos, running Red Hat Enterprise Linux 6.2.
Standard benchmarks are a proxy for the performance that users may see with their own workloads. Schooner ran the standard DBT2 benchmark, which is the open-source OLTP version of the TPC-C benchmark, using a heavy workload of 1,000 warehouses with 32 concurrent connections to the database and zero think time. Measurements in each case were made using a typical two-node master-slave configuration. Each node was a standard IBM x86 server with 12/24 processor cores/threads, 72 GB of DRAM, and two 320 GB Fusion-io ioDriveDuos, running Red Hat Enterprise Linux 6.2.
SchoonerSQL Delivers Three Times the Throughput of New Percona XtraDB Cluster, with 1/4th the Response Time and Half the Total Cost of Ownership
Press Release Link
Thank you for the analysis. Can you share more details about how you determined TCO? All the other numbers look objectively reproducible but it would be good to understand things like the timespan and other assumptions that go into the TCO number.
ReplyDeleteRespect Baron,
DeleteIve got problems getting the other numbers also.
1. A Master-Slave Replication (Schooner) solution is compared to a Master-Master (Galera) solution.
2. I don't see any configs.