E2C continuum and cloud computing

AOverview

Lecture 01 asked where each task of an application should run. This lecture builds the places it can run. First it gives the Edge-to-cloud (E2C) continuum its shape, three layers ordered by latency, and then it opens up the largest of them, the cloud: what it is made of, what it costs, how it is sold and how reliable it promises to be.

As a PhD student you need it for three things. The exams ask you to compare the edge, fog and cloud layers, tell a Type 1 from a Type 2 hypervisor, recite the NIST definition and compute a Total cost of ownership (TCO) or a Compound SLA. The research project needs the same vocabulary to state precisely what layer, what virtualization and what service model a proposed system assumes. And any real system you design will be judged on whether its cost and its availability hold up, not only its latency.

The path has five stops. You order edge, fog and cloud by latency and see why the fog was invented. You go underneath the cloud to Virtualization, from hypervisors to containers built on namespaces and cgroups. You price legacy IT against the cloud with a three-year cost case study. You learn how the cloud is defined, deployed and sold as IaaS, PaaS and SaaS by hyperscalers and private platforms. Finally you turn an Service-level agreement (SLA) into allowed downtime and chain services into one compound guarantee.

Success looks like

• Place a workload on the edge, fog or cloud layer and justify it by latency, capacity and the scope each layer serves.
• Explain how a hypervisor and a container isolate workloads differently, naming what namespaces and cgroups each control.
• Compute a three-year TCO from CapEx and OpEx and explain why the cloud came out more than 35% cheaper in the case study.
• State the five NIST essential characteristics and match a scenario to its deployment model.
• Draw the IaaS, PaaS and SaaS stacks and mark where responsibility passes from customer to provider.
• Convert an SLA percentage into allowed downtime and compute a serial compound SLA, such as 0.9999 × 0.999 ≈ 99.89%.

How to study this lecture

1. Choose your route. Read the full guide in one sitting for the big picture, or work through one part at a time when you want depth.
2. Answer every recall prompt in your head, or on paper, before you reveal it. The effort of retrieving is what makes the idea stick.
3. Take each quiz and read the explanation even when you are right.
4. Mark a concept as understood only when you could explain it to a classmate without looking. Unmarked concepts show you where to return.
5. Keep the glossary open for terms, and use the reference sheet for the layer comparison, the cost figures and the downtime table when you review or solve problems.
6. Come back after a few days and retry the recall prompts and quizzes cold. Spaced practice builds the long-term memory an exam needs.

BThe 10 parts

1. 01The latency continuum and the birth of the fogWhy the cloud cannot simply move closer to the client, how mini data centers along the latency continuum form the fog, and which points need networking at all.5 conceptsSlides 1-930 min
   1. 1.1Edge, fog and cloud as distances on one latency continuum
   2. 1.2Why the cloud cannot move closer, but small copies of it can
   3. 1.3The fog: a layer of mini data centers between edge and cloud
   4. 1.4Does every point need networking? Let the function's scope decide
   5. 1.5Edge on the LAN, fog on the WAN, cloud on the Internet
2. 02Roles of the edge, fog and cloud layersWhat each layer of the continuum provides, why the fog helps, and two worked architectures (IoT hierarchy and a vehicle platoon).5 conceptsSlides 10-1430 min
   1. 2.1The fog: one mini data center serving many edges
   2. 2.2The edge: the client's front door to the continuum
   3. 2.3The cloud: big compute, big storage, big networking
   4. 2.4Reading an IoT hierarchy: sense, process, coordinate, analyse
   5. 2.5A vehicle platoon: edge cars, roadside fog, distant cloud
3. 03From legacy IT to virtualizationStarts the cloud computing section with the utility computing vision and the legacy IT stack, then defines virtualization, its history and its main forms.6 conceptsSlides 15-2136 min
   1. 3.1Computing as a utility: McCarthy's 1961 idea that became the cloud
   2. 3.2Legacy IT: when the customer owns every layer
   3. 3.3Virtualization: one physical computer, many independent ones
   4. 3.4Transparency, and why virtualization is not emulation or simulation
   5. 3.5From VM/370 to a family of virtualization techniques
   6. 3.6The hardware virtualization stack: isolated above the interface, shared below
4. 04Hypervisors, containers, namespaces and cgroupsCompares Type 1 and Type 2 hypervisors, then explains OS-level virtualization (containers) built from Linux namespaces and control groups.6 conceptsSlides 22-2836 min
   1. 4.1Where the hypervisor sits: bare metal or hosted
   2. 4.2Type 1 versus Type 2: overhead, control and isolation
   3. 4.3Containers: many isolated user spaces on one kernel
   4. 4.4Two problems, two kernel features
   5. 4.5Namespaces: what a process can see
   6. 4.6Control groups: what a process can use
5. 05Cost of on-premise operation and TCOLists the cost drivers of legacy IT and works through a three-year total cost of ownership calculation for a 50-employee news streaming SME.5 conceptsSlides 29-3430 min
   1. 5.1Where the money goes in legacy IT
   2. 5.2The news SME case: assumptions and hardware
   3. 5.3CapEx versus OpEx: buying assets versus paying to run them
   4. 5.4The upfront bill: 68,824 EUR of CapEx
   5. 5.5Running costs and the three-year TCO
6. 06Cost case study, NIST definition and deployment modelsFinishes the traditional IT versus AWS cost comparison, then introduces the NIST definition of cloud computing, its five essential characteristics and the four deployment models.6 conceptsSlides 35-4436 min
   1. 6.1Pricing the same workload as a monthly cloud bill
   2. 6.2The three-year verdict, and why nobody deletes their own IT
   3. 6.3The NIST definition: one sentence, three layers
   4. 6.4Five essential characteristics: the test for whether something is cloud
   5. 6.5Deployment models: who may use it and where it lives
   6. 6.6Hybrid and community clouds: composing clouds and sharing one
7. 07Service models: IaaS, PaaS and SaaSExplains how IaaS, PaaS and SaaS split responsibility for the nine-layer stack between customer and provider.4 conceptsSlides 45-4824 min
   1. 7.1One stack, one moving line: who manages each layer
   2. 7.2IaaS: renting virtual hardware and running everything above it
   3. 7.3PaaS: bring code, the platform runs and scales it
   4. 7.4SaaS, and the responsibilities that never leave the customer
8. 08Public cloud hyperscalers: AWS, GCP and AzureSurveys the history and service portfolios of the three biggest hyperscalers.5 conceptsSlides 49-5630 min
   1. 8.1Why three companies became hyperscalers
   2. 8.2AWS: an internal platform turned into a product
   3. 8.3Reading a portfolio: six building blocks
   4. 8.4Google Cloud: platform first, infrastructure later
   5. 8.5Azure: the enterprise franchise moves to the cloud, and one map for all three
9. 09Private cloud platforms: OpenStack and OpenNebulaIntroduces self-hosted private IaaS platforms, focusing on the open-source OpenStack and OpenNebula projects.4 conceptsSlides 57-6224 min
   1. 9.1Building your own cloud: what a private IaaS platform is
   2. 9.2OpenStack: a NASA and Rackspace project that became a foundation
   3. 9.3The OpenStack service map: one API per resource, glued by Keystone
   4. 9.4OpenNebula: a lean front-end that manages hypervisor clusters
10. 10Service-level agreements and availabilityDefines availability, maps SLA nines to allowed downtime, and shows how serial services combine into a compound SLA.4 conceptsSlides 63-6824 min
    1. 10.1Availability: the share of time a service can answer
    2. 10.2From nines to a downtime budget
    3. 10.3Chaining services multiplies their availability
    4. 10.4Turning an SLA into allowed downtime you can check

CGlossary and reference

• GlossaryEvery term from the lecture, defined once, with links back to the slides that use it.
• Reference sheetThe lecture on one printable page, for revision before a quiz or exam.