We conducted a performance test comparing the performance of Fujitsu Enterprise Postgres backup utlity pgx_dmpall backup with the OSS PostgreSQL pg_basebackup method for database backups. The purpose of the test was to evaluate the backup speed and overall efficiency of these two backup solutions.
The performance test report compared the performance of Fujitsu Enterprise Postgres with the pgx_dmpall backup method and with the pg_basebackup method for database backups. The purpose of the test was to evaluate the backup speed and overall efficiency of these two backup solutions.
Test environment
- Hardware configuration
- Server: Azure VM
- CPU: 4 x Intel Xeon Gold 6240 @ 2.60GHz
- RAM: 16 GB
- Disk: Standard SSD LRS
- MAX IOPS: 500
- MAX throughput (MBps): 60
- Software configuration
- Operating system: CentOS 8.5
- Fujitsu Enterprise Postgres: Version 15.0
Prerequisites
- Install the required OS packages.
For details, refer to the Fujitsu Enterprise Postgres manual Installation and Setup Guide for Server > Operating environment > Required packages.
- Basic working knowledge of the pg_basebackup utility.
For details on this utility refer to the PostgreSQL documentation PostgreSQL client applications > pg_basebackup.
- Basic working knowledge of the pgx_dmpall utility.
For details on this utility, refer to the Fujitsu Enterprise Postgres manual Operation Guide > Backing up the database.
- Testing database of 10 GB.
To create the 10GB data use the command pgbench -s 700 --foreign-keys -p 27500 -i test_bkp, where test_bkp is the testing database and pgbench will create the sample tables and data inside this database.
Test scenarios
- pgbench test
Execute the pgbench test in two different sessions to generate some sort of transaction load into the database.
pgbench -c 50 --progress=60 -T 1800 -p 27500 test_bkp
- Backup speed
Measure the time it takes to perform a full database backup using pgx_dmpall and pg_basebackup for a database of size 10GB using the fast checkpoint option.
Backup speed test results
- pgx_dmpall
The backup of the 10GB database using pgx_dmpall was completed in an average of 3 minutes and 1 second.
- pg_basebackup
The backup of the same 10GB database using pg_basebackup was completed in an average of 4 minutes and 32 seconds.
Results using pg_basebackup
| Item | Test 1 | Test 2 | Test 3 | Test 4 | Test 5 |
| Start time |
Thu Sep 21 05:33:41 UTC 2023 |
Thu Sep 21 06:42:20 UTC 2023 |
Thu Sep 21 06:50:45 UTC 2023 |
Thu Sep 21 06:56:39 UTC 2023 |
Thu Sep 21 07:13:43 UTC 2023 |
| End time |
Thu Sep 21 05:42:04 UTC 2023 |
Thu Sep 21 06:45:33 UTC 2023 |
Thu Sep 21 06:54:00 UTC 2023 |
Thu Sep 21 07:01:25 UTC 2023 |
Thu Sep 21 07:16:48 UTC 2023 |
| Total time taken |
8 minutes, 23 seconds |
3 minutes, 13 seconds |
3 minutes, 15 seconds |
4 minutes, 46 seconds |
3 minutes, 5 seconds |
| Average time taken |
4 minutes, 32 seconds |
||||
Results using pgx_dumpall
| Item | Test 1 | Test 2 | Test 3 | Test 4 | Test 5 |
| Start time |
Thu Sep 21 06:14:27 UTC 2023 |
Thu Sep 21 06:47:10 UTC 2023 |
Thu Sep 21 07:18:32 UTC 2023 |
Thu Sep 21 07:21:46 UTC 2023 |
Thu Sep 21 07:27:14 UTC 2023 |
| End time |
Thu Sep 21 06:17:34 UTC 2023 |
Thu Sep 21 06:50:03 UTC 2023 |
Thu Sep 21 07:21:25 UTC 2023 |
Thu Sep 21 07:25:20 UTC 2023 |
Thu Sep 21 07:29:53 UTC 2023 |
| Total time taken |
3 minutes, 7 seconds |
2 minutes, 53 seconds |
2 minutes, 53 seconds |
3 minutes, 34 seconds |
2 minutes, 39 seconds |
| Average time taken |
3 minutes, 1 second |
||||
Benchmark key takeaways
- The backup speeds for pgx_dmpall constantly outperformed pg_basebackup in terms of backup time for the 10GB database.
- It's important to note that performance can vary depending on the specific workload and system configuration. Further testing under different scenarios and with larger datasets may be necessary to draw more comprehensive conclusions.
- Both tools offer similar functionality and ease of use. Users familiar with PostgreSQL’s native tool will find pg_basebackup intuitive, while those seeking advanced features and improved performance may opt for pgx_dmpall provided by Fujitsu Enterprise Postgres.
Making the right choice for your backup needs
Based on this performance test, Fujitsu Enterprise Postgres with pgx_dmpall offers good performance for the backup operation as compared to pg_basebackup.
Enterprise customers with large databases may benefit from adopting Fujitsu Enterprise Postgres for their enhanced backup capability.
It is also worth to note that performance results may vary based on hardware configurations and database size. Therefore, it is highly advisable to conduct a similar performance test in a representative environment before making a final tool seclection.
