The Origins and Future of Distributed Computing and Clouds - The Practice of Cloud System Administration - page 447

Appendix B. The Origins and Future of Distributed
Computing and Clouds
To me it seems quite clear that it's all just a little bit of history repeating
—Propellerheads
Modern, large datacenters typically consist of rack after rack of pizza box-sized computers.
This is a typical design pattern for any large cloud-scale, distributed computing environ-
ment. Why is this? How did so many companies arrive at the same design pattern?
The answer is that this design pattern was inevitable. To manage enormous datacenters,
hardware and processes must be highly organized. Repeating the same rack structure within
a datacenter reduces complexity, which reduces management and administrative overhead.
Many other factors led to the choice of hardware to fit into those racks.
However, nothing is truly inevitable. Thus the appearance of massive datacenters full of
machinesbuiltfromcommodityhardware,distributedcomputingreplacinglargecomputers,
and the popularity of cloud computing were inevitable only in the sense that necessity is the
mother of invention.
Earlierapproachestolargewebande-commercesitesdidnothavetherighteconomiesof
scale. In fact, scaling became more expensive per user. This caused the first dot-com bubble
to be unsustainable. What drove the distributed computing revolution was the need to create
the right economies of scale, where addtional users become cheaper to support.
Every order of magnitude improvement in the cost of computing enables a new era of
applications, each of which was unimaginable just a few years before.
Each requires new supporting infrastructure technology and operational methodologies.
These technologies andmethodologies donotalways arrive intime, andsometimes theyare
ahead of their time.
Understanding the historical context of all these changes gives context to the tools and
techniques described in this topic.If this doesn'tinterest you,feel freetoskipthisappendix.
Thehistoryofthecomputerindustryisquitelong,startingwiththeabacus.Thisappendix
will skip ahead a few years and focus on a few specific periods:
• The pre-web era: The years immediately prior to the web, from 1985 to 1994
• The first web era: “The dot-com bubble” from 1995 to 2000
• The dot-bomb era: The economic downturn from 2000 to 2003