Hyper Open Edge Cloud


Describe the rationale for burning test in RapidSpace servers
  • Last Update:2020-06-02
  • Version:001
  • Language:en

Q: Why do you do 3-day burning test although the hardware was tested at factory?

A: In our experience, even the biggest brand hardware has a significant rate of failure whenever it is used with real database loads, such as ERP or Big Data. We found that by running a very stressful test for 3 days, combining CPU, disk I/O, networking and RAM all at the same time, we can discover faulty machines quite efficiently. The test we use consists of running ERP5 unit test suite (6000+ unit tests) on all cores of the system at the same time. This test suite includes downloading source code, compiling, running database queries, all at the same time. It is more stressful than traditional tests such as cpuburn or memtest86. Those servers that pass this test will work reliably for a couple of years, even in case environment temperature becomes a big high (35 degrees) in summer.