Know that the server with ECC memory, by encoding can correct a few bit errors, but also always thought that Reg ECC is similar function
A recent incident has made me seriously feel that Reg ECC is not just a function of correcting bit errors, it can also switch the fact hot.
Recently bought the second-hand HP server to do experiments, installed is the ESXI5.1 system, one day running is suddenly found that the server access is particularly slow, PING ESXi server drops is very serious, but after a few minutes and normal, also thought is the switch problem, restart the switch everything is normal, until the afternoon, the same problem has arisen, After a few minutes and normal, this came to the machine room testing equipment, accidentally found on the server state and other bright red.
Use notebook to Exsi CLIENT, view, find hardware warning memory Alarm, inform an invalid memory corruption, security shutdown server, remove the alarm memory, boot server status lights to restore green, the system back to normal.
If it is using ECC memory, I would like to estimate the server will crash or spend the screen, REG ECC can let the server thermal failure migration, and alarm prompts.
So later I on the server can no longer according to the needs of how much memory to buy, must consider a more redundant memory space, such as the actual demand 16G with 4G single, a total of 4, then I will install 1 more, 20G, if you buy 8G a single word I will buy 3 24G
This article is from the "WINGS3" blog, make sure to keep this source http://57072.blog.51cto.com/47072/1660634
Benefits of Reg ECC memory for servers