EDB Postgres AI, the newest PostgreSQL-based database from EnterpriseDB (EDB), outperformed 4 different databases, together with Oracle, SQL Server, MongoDB, and MySQL, on a pair of benchmark exams that span SQL and NoSQL workloads, the seller introduced right this moment.
EDB commissioned McKnight Consulting Group to run a pair of benchmark exams to measure the relative efficiency of EDB Postgres AI towards relational and non-relational databases. McKnight arrange a TPC-C-like take a look at to match the transaction processing prowess of EDB Postgres AI towards Oracle and Microsoft SQL Server, whereas it arrange a JSON processing take a look at to match the multi-modal database towards MongoDB and MySQL, which is managed by Oracle and may course of JSON paperwork.
On the brand new orders per minute (NOPM) transactional take a look at, McKnight measured EDB Postgres AI hitting a peak throughput of about 987,000 NOPM in comparison with about 844,000 for Oracle and about 759,000 for SQL Server. The exams had been carried out on equivalent AWS environments: ixi8xlarge cases operating in US East 1, which price $24,055 per 12 months to run.
Supply: McKnight Consulting Group
The value-per-performance comparisons had been starker. In response to McKnight, EDB Postgres AI was in a position to obtain an NOPM price of $0.21 per unit in comparison with Oracle’s $1.58 and SQL Server’s $1.26. The variations are because of the larger price of the Oracle ($47,500 per unit) and Microsoft ($15,123 per unit) software program environments in comparison with EDB ($2,780 per unit), in addition to larger assist prices ($10,450 per unit for Oracle and $3,327 per unit for Microsoft versus zero for EDB).
McKnight notes that, whereas the NOMP benchmark was arrange in accordance with TPC-C requirements, it was not an official TPC-C benchmark take a look at. To function the database driver, McKnight used HammerDB, which is “a broadly used and accepted implementation of the TPC-C take a look at,” the corporate says.
A wholly totally different take a look at was used for the JSON workload, however the identical AWS cases had been used. For the JSON take a look at, McKnight used PG NoSQL Benchmark, which is a benchmark software created by EDB to check PostgreSQL and MongoDB databases. McKnight modified the syntax of the benchmark to incorporate MySQL, which is also able to processing JSON information.
Supply: McKnight Consulting Group
The take a look at measured how rapidly every database might carry out batch a great deal of JSON paperwork, INSERT the information into the database, after which carry out a SELECT question on the information. The batches ranged from 5 million paperwork (13GB) as much as 100 million paperwork (266GB). The take a look at simulates the kind of database work that may be carried out as a part of a retrieval augmented technology (RAG) pipeline for a generative AI software.
EDB Postgres AI beat the opposite two databases throughout all three JSON exams. For bulk hundreds, EDB was 34% quicker than MongoDB and 63% quicker than MySQL. For database INSERTS, EDB was 4 instances quicker than MySQL and 150 instances quicker than MongoDB. For the SELECT question, EDB was on common nearly 3 instances quicker than MySQL and greater than 5 instances quicker than MongoDB. Extra particulars of the take a look at setup and outcomes may be present in McKnights’ report right here.
The exams pit the world’s 5 hottest databases towards one another. Whereas Oracle, MySQL, and SQL Server proceed to be the most well-liked relational databases, PostgreSQL has emerged to threaten them for dominance, based on DB-Engines rating. MongoDB has been the world’s hottest NoSQL database for nicely over a decade.
Associated Objects:
Microsoft Open Sources Code Behind PostgreSQL-Primarily based MongoDB Clone
Postgres17 Brings Enhancements to JSON, Backup, And Extra
Postgres Rolls Into 2024 with Large Momentum. Can It Hold It Up?
