How important is "Advanced" ECC mode?

[PliotronX]

Server is a PE R510 w/two Xeons and we will be upgrading its RAM today. Not sure what was ordered but from the sounds of it, 4x16GB is coming in and it currently has 4x8GB in Advanced ECC mode. According to the manual, having two banks, it won't run with all of these particular (mixed) DIMMs installed but I might be able to pull off running two of the 8GB with the four 16GB. This will populate all channels for maximum bandwidth, however it might have to run in "Memory Optimizer" mode. I found a much more detailed description than Dell's own whitepaper here about it, but I am not convinced that it should be running in AECC (a.k.a. SDDC).This is not a clustered server, but it will have a future running two or three VMs including a DC. What say you guys? 64GB able to "recover" from a full stick failure or 80GB of tri-channel goodness?

 

[Mark R]

Advanced ECC doesn't protect against a whole stick failure, it protects against a single *chip* failure.

Single chip failure is very rare, and will usually be rapidly detected by regular ECC resulting in a memory failure BSOD.

It really depends on how critical your data and uptime are as to whether you want to protect against the very rare event of a single memory chip failing.

 

[PliotronX]

Interesting, thanks! I was reading about findings by HP and the Googs that 24/7 servers see a few single bit errors annually in which case ECC helps check for, but would this AECC assist in actually correcting them? Will it BSoD as well? IOW does it behave differently than module based ECC? I came close to finding my first ever bad stick of RAM but I still have yet to run into a DIMM being the cause of memory errors, it's usually the mobo which is why if it were up to me, I would go with the faster and more capacious "Optimizer" config.