Quantcast
Viewing all articles
Browse latest Browse all 183865

ECC vs. non-ECC RAM (Xeon vs. I5 / I7 and AMD equivalents)

We're looking to replace a number of aging servers (file, Exchange, print, etc.) with a pair of redundant virtualized systems. We're fairly small (30-ish people total), and have about 8 TB of data.

The usual arguments of price vs. reliability aside, my management wants other alternatives besides used servers in excess of $5,000 each. (Convincing them of the need for almost $10,000 in software is hard enough.)

I've read various articles, both here and elsewhere, that plainly state that ECC is better than non-ECC for servers and therefore must have server-grade motherboards and Xeon processors. Some articles even quote an alleged study by Google that found error rates "much higher than expected", but the numbers vary from one bit in a megabyte to 4,000 correctable errors in a (size unspecified) RAM module per year, so I have my doubts as to the veracity of these articles. Then there's all these people who are trying to set up server farms with Raspberrry-Pi systems...

I (and probably most of us) have run used desktop equipment at home as servers for years, with few problems, if any, though of course they don't get anywhere near the use they would in a business.

Back to reality... Is *anyone* out there actually running (or have run in the past) non-Xeon, non-ECC servers, and what kind of results have you had with them over time? Are data corruption and crashes prevalent? I've seen my share of those with server-grade equipment, too.


Viewing all articles
Browse latest Browse all 183865

Trending Articles



<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>