1. A Brief History Of
Cloud
(..and a brief future of PaaS)
James Watters
2. • James Watters: @wattersjames
• Last 10 years: strategy, business
development, technology partnership roles
at Sun and VMware
• Projects: Cloud Foundry, vCloud, Sun
volume systems, Solaris
• Today is my personal take only....
3. • Agenda
• What were the ‘Cloud’ marketing claims?
• How are we doing against them?
• So what’s next?
11. • Big $ transaction, for big transactions
• RAS
• Highly specialized IO
• Small transaction, for small(er) systems
• Ease of +1 system scaling
• Commodity IO
17. ~2.4M x86 servers sold in
2q11
• In same period over ~5M
virtual machines and cloud
servers created.
• New tool chains being born
to manage new provisioning
and server lifecycle
philosophy...
22. “Is there a site that offers price history for various
Amazon Web Services such as EC2, Cloudfront,
etc?
I would like to be able to forecast that if something
like a small on demand EC2 instance costs $0.085
per hour today that it will likely half in cost to $0.043
per hour a year from now. Similarly if I have 10GB of
files in S3 storage how will the cost be affected over
a similar span of time? I can only imagine, that like
all technology, the cost will go down.”
23. “There hasn't really been a need for anyone to
make such a site since prices have only changed a
couple times in the many years EC2 has been
around. Less than once every other year! The
CPU-hour price of a small instance has only
changed once in the past 5 years -- from its
original $0.10 to $0.085.”
24. • RDS launch October 2009
• 2009 XL DB Instance: .88/hr
• Today: $.88/hr
• Aligned to RAM pricing declines = $.49/hr
25. EC2 does cut prices,
strategically...you are at
their mercy.
35. Story of Urban Airship
• August 2009: Sent over one
million production
notifications
• August 2011: Sent over five
billion production
notifications
• Tracking over 300 million
installations
• Started on EC2..
36. “What is the magic
bullet? Real hardware.
know, I know...”
--Adam Lowry, Urban Airship
45. Moore's Law is for
Hardware Only
• Does not apply to software
• Productivity gains not keeping up with hardware
and bandwidth
• Writing software is hard, painful, and still very
much a craft
Credit: @chanezon