Scale up vs. Scale Out

See http://weblogs.java.net/blog/malcolmdavis/archive/2006/07/scale_up_vs_sca.html

Reliability vs. Availability

Reliability: measured by time between system failures
o Distributed system should be more reliable than single system
o 5 machines with .95 probability of being up. 1 - .05*5 probability of
being up.

􀀂 Availability: fraction of time the system is usable.
o Redundancy improves it
o Need to maintain consistency
o Need to be secure
o Need to tolerate failures: mask failures, recover from errors.

http://www.cis.upenn.edu/~lee/07cis505/Lec/lec-ch7a-consistency-v3.pdf

Calculate in Google

Try the following search terms:

(120000+1500)*500 byte -> megabyte
(120 000 + 1 500) * 500 gram -> kilogram