Cloud Computing Larry Gottschalk Computer Science Faculty Metropolitan State University
Definition • The movement of computing from the desktop and corporate servers to computing services (on a myriad of hosts/servers). • Striking similarity to service bureaus and time-sharing systems of 40 years ago. • Biggest difference: now is internet connections, instead of dedicated phone lines.
Analogous to Elec Utility • Billed for services used • Separate bill from each utility used • Visibility to infrastructure costs (unlike now with x employees, y servers used for both development AND operations) • “on demand computing” says it all
Drivers • Cost to maintain applications on desktops, departmental and enterprise servers is astronomical, and not getting better. • Distraction of IT staff by doing maintenance • Service level management also is outsourced. • Reliability (service level can be contract item) • Cost: 18% lower; electric power 16% less • Abililty to experiment with one or two apps
Cloud Architecture • Large server farms • Linked by high-bandwidth connections • Carefully managed, highly tuned.
Cloud Major Players • The usual heavy hitters have entered: • Oracle • IBM • Yahoo • Amazon • Google • Microsoft • Qwest, and other phone companies
Terminology • Cloud Computing also referred to as: • On demand computing • Software as a service • The Internet as platform • (First two phrases worked back in time sharing days also.) • Virtualization: • Location of service hidden behind generic URL • Variable number of servers running your app(s) • Done with high speed links within server farms
Different Offerings • The next four slides list discrete market offerings of Cloud Computing: • Infrastructure as a service (Most important) • Software as a service (2nd most important • Desktop productivity tools into the cloud(minor) • “dumb terminals” serviced by cloud (minor)
CC: Infrastructure as a Service (IAAS) • This is the time-share of 40 years ago. You can move your corporate apps to the Cloud. • The Cloud provider is responsible for uptime, and for restarting apps when they crash. • Enterprise gains dynamic scalability (if have contracted for it) • Enterprise gets locked into vendor APIs • API== “Application Programming Interface”
CC: Software as a Service (SAAS) • Second most important segment • Useful applications: • Customer contact tracking • Market research • Engineering applications • ERPs • Industry specific apps in • Petroleum, Engineering, Medicine, Transportation
CC: Desktop productivity tools into the cloud • Minor segment • Desktop productivity becomes remote service: • Google docs • Buzzword (bought by Adobe) • Photoshop Express
CC: “dumb terminals” serviced by cloud • Minor segment • Some (10%?) of PCs will have no OS nor even browser. Simply a boot program to download and start the server apps. • Examples of dumb-terminal emulation s/w: • eyeOS system • AIR (formerly Apollo) from Adobe • Open-Laszlo, an open-source project
Restraints to adoption • Disruptive • Price appears as line item CEO can criticize • CIO’s desire to build empire • Lots of re-packaging without innovation, which is masked by excessive hyped marketing
Other restraints to adoption • Privacy: can leaks of data be prevented? • Security: will data ever be lost? (no FDIC for data) • Reliability: can level of service guarantees be met? • Ownership issues • If you terminate subscription and you discover you need a document, can you get it? • Can you really ever delete a document? • If gov’t subpoenas your data, will you even be told, and ever be told whether data was surrendered or not.
What does future hold? • Possible • Probable • Desirable
Possible • Worst case: there are one or two disasters written up widely, causing pulling back by clients. • Bad case: huge cost of outsourcing becomes better understood • Good case: 40 to 60% of Enterprise Computing gets outsourced • Best case: Enterprise computer centers wither.
Probable • Innovation will continue, with massive value for customers • Market may get overheated with everyone jumping in, and customers oversold • In lead now are Google, Amazon, VMWare, Citrix, Microsoft, HP, and IBM. • Innovators at this time are Google, Amazon, and VMWare.
Desirable • Cloud computing is seen as alternative to in-house processing. Each firm strikes a balance between the two depending on its own values. • Ownership of data • Ability to determine own destiny • Desire to get out from under maintenance. • Example: Target.com vis-à-vis Amazon.
Lots of new entrants • The startsups will or are becoming IPOs • McAfee and Verizon teaming up • Cisco/EMC joint venture (large storage) • Compuware (application monitoring) • Rackspace • Cetrom/ASCIIgroup joint venture (CaaS) • IBM SmarterPlanet
DOD • RACE (Rapid Access Computing Environment) Claims 99.99% availability (53 minutes down / year)
Discussion questions • What industries or market segments will become adopters beyond just the mission critical apps? • Large firms? Medium firms? • What are effects on vendors to IT: • Server market • Large storage market