JACOB LANGE CLOUD COMPUTING
Dec 17, 2015
J AC O B L A N G E
CLOUD COMPUTING
DEFINITION
• Colloquial expression• Multiple machines connected through real-time
communication network• Work together to perform a given task• Pooling of resources (processing, memory, etc)• Virtualization• Hypervisor• Optimizing effectiveness of shared resources
HISTORY
• Ideas go back to 1950’s• Mainframe computing• Terminals• IBM
HISTORY
• 1970’s• Virtualization• Greatest technological drive for communication and
computing evolution
• IBM – VMOS• Made up for limitations for physical memory by allowing
multiple virtual environments
HISTORY
• 1990’s• Telecommunication companies provide shared
access to same physical structure• Virtualization becomes standardized and cheap• Explosion of World Wide Web
HISTORY
• 2000’s• Huge price drops for hardware• Shift away from one server to multiple servers
HISTORY
• Today• Multiple services finding own niche in market• IaaS, PaaS, SaaS, and others• Remote storage/access• Businesses relying on third party cloud service for
all computing needs• Concerns about security
CHARACTERISTICS
• Application Programming Interface (API)• Physical Component Based• Representational State Transfer (REST)• Ignores details of component implementation• Focuses on roles of components and interactions• Client-Server model• Internet is largest example of REST
CHARACTERISTICS
• Multitenancy• Single instance of software serves multiple clients• No need to buy new hardware/software for each
user. Only buy when needs of all users exceed system’s limits
CHARACTERISTICS
• Remote Access• Device independent• Emulation• Remote data storage
CHARACTERISTICS
• Virtualization• Virtual Machine• Highly mobile• All software is emulated on virtual machine. • Can be ported to different physical machines very easily
CHARACTERISTICS
• Reliability• Failure of one physical node means very little• Very easy to back up data and store on another
machine• Very good for users that require continuous
access
CHARACTERISTICS
• Scalability/Elasticity• Just add another server to the chain• Diminishing returns
• Upgrade hardware to one or several nodes in the chain• Also results in diminishing returns
• Turn a node on or off depending on user demand
CHARACTERISTICS
• Efficiency – not so good• Having all resources in one super computer is
more efficient than having them spread out• Adding physical node to the chain results in
diminishing returns• New node means more security, more work for
hypervisor, more overall complexity, etc
CHARACTERISTICS
• Maintenance• “Easily” maintainable by third-party system• Wider cloud -> harder to maintain
• Businesses/individual users don’t need to worry
CHARACTERISTICS
• Security – double edged• Still a huge issue• Cloud handled by one organization tackles
greater security threats than multiple users• Wider cloud -> more difficult security• Wider cloud -> more difficult to log user activity• Trust
CHARACTERISTICS
• Legal issues• Ownership?• Confidentiality of customers of the business that
uses the cloud system• Private/sensitive data• Tracking/Abuse of data by cloud host
SERVICE MODELS
SERVICE MODELS
• Infrastructure as a Service (IaaS)• Most basic• Provide virtual machine, physical resources,
hypervisor• Storage, firewalls, IP addresses, virtual LAN’s• Remote storage• Little to no cost depending on amount and
specificity of data
SERVICE MODELS
• Platform as a Service (PaaS)• Provides computing platform (OS)• Environment built for programming language
execution (libraries)• Run software without worrying about complexities
of underlying hardware or software packages
SERVICE MODELS
• Software as a Service (SaaS)• Most sophisticated, most expensive• Infrastructure and applications• Ideal for non-technical business• Vast majority of work is on host side• Remote access to applications• Host has access to client work - Trust
SERVICE MODELS
• Other models (NaaS, CaaS, DaaS, BPaaS)• Everyone has their niche• 2012 – NaaS included in standard cloud model by
International Telecommunications Union• NaaS provides flexible VPN, bandwidth on-
demand
TRUST ISSUES
• Four components for trust in cloud computing• Security – classic computer security
• Privacy – Confidential data won’t be leaked
• Accountability – Taking responsibility for performance, • includes legal responsibility
• Auditability – quality of records that track user/host activity
TRUST ISSUES
• Methods to address concerns:• Preventative Controls – Controls number of times any
action is performed, and access to data.• Ex: Restrict Read/Write access, Firewalls
• Detective Controls – Identify privacy and security risks that break established polices/procedures.• Ex: Data logs, security audits
CLOUD ACCOUNTABILITY LIFE CYCLE
CLOUD ACCOUNTABILITY LIFE CYCLE
1. Policy Planning – decide what information needs to be logged
2. Sense and Trace – Software used to detect unexpected events in cloud environment
3. Logging – Log event with as much detail as possible
4. Safe-keeping of Logs – Encryptions, back-ups, etc5. Reporting and Replaying – Logs generate reports
that summarize event6. Auditing – Checking reports for irregularities7. Optimizing and Rectifying – Security loopholes
are closed and any disciplinary action is taken
FUTURE OF CLOUD COMPUTING
• Huge potential for non-technical business• Pooling of resources almost always makes them
more affordable for everyone• Sharing of resources• Ownership• Sharing of data – TRUST!