Understanding XaaS Concepts

Advanced Software Engineering Internet Applications - Cloud Computing

• Cloud computing is a model enabling convenient, on-demand network access to shared computing resources (e.g., networks, servers, storage, applications, services).
• It quickly provisions and releases resources with minimal management and service provider interaction.
• Key ideas in this model include: on-demand resource consumption via the internet, self-service, rapid elasticity, resource pooling (location independence), and usage-based billing (utility computing).

Cloud Computing Definition - 1 & -2

• Cloud computing is a distributed system for computing as a service.
• Cloud providers supply IT resources on demand for users.
• Scalability and load balancing are key features aimed at higher efficiency, lower cost, and greater reliability than in-house solutions.
• Service virtualization allows for abstraction from resource details.

Cloud Computing Stack

• Client: End-user access (e.g., browser, desktop/mobile).
• Application: End-user applications (e.g., CRM, SRM, ERP, CMS).
• Platform: Execution environment (e.g., Java, .NET, SAP).
• Infrastructure: Virtualized resources (e.g., CPU, storage, network), hardware, and operating systems optimized for cloud use.
• Server: Hardware and operating systems optimized for cloud use.

Cloud Computing Definition - 3

• The cloud computing stack consists of various layers:
  • Hosted Applications/Suites of Services
  • Development tools, Database Management
  • Operating Systems, Virtualization
  • Servers and Storage, Data Center infrastructure
  • Networking and Firewalls/Security
• Network as a Service (NaaS)
• Infrastructure as a Service (IaaS)
• Platform as a Service (PaaS)
• Software as a Service (SaaS)

Core Cloud Technologies

• Virtual Storage: A distributed file system or database combining multiple physical address spaces to create one virtual space.
• Virtual Machines: Packaged execution environments that migrate between machines for applications without modification. Each VM has an isolated virtual address space on the host machine, making it easy to create and deploy.
• Multi-Tenancy: Supporting numerous, separated users within the same address space and process, saving resources over single-user instances.

Cloud Ownership

• Private Cloud: IT resources controlled by an organization for internal use within its data center or infrastructure (internal/partner usage).
• Public Cloud: Resources offered to third-party consumers with varied pricing models (e.g., pay-per-use, subscription).
• Public/Private Cloud: Public cloud resources leased when private cloud capacity is insufficient.

Cloud Benefits

• Cloud Consumers: Low entrance barrier with pay-per-use, lower capital expenditure (CAPEX), and only operational cost (OPEX).
• Cloud Providers: High resource utilization, and leverage the long tail effect.
• Technical: Fast and seamless scaling, infinite amount of resources, automation, increased efficiency via virtualization and encapsulation.

Cloud Computing Challenges

• Economics (providers): N-sided business models with high profit margins; Cost savings versus risk in consumer use.
• Economics (consumers): Cost savings may have risks; SLAs may have limits.
• Other players: Marketplaces and cloud brokers; value-added services.
• Technical: Near-zero effort deploys with migration through inter-operability between layers; Hide technical details; Seamless and automatic scaling.
• Technical: Security and data integrity; Integration of public and private clouds.

Cloud Challenges: Vendor Lock-in

• Consumers become dependent on providers; migration costs are often higher than cost savings; provider knowledge and data format may lead to vendor lock-in.

Hype Cycle for Cloud Computing (Gartner 2011)

• A graph showcasing the market anticipation and disillusionment for various cloud computing methods over time.

Infrastructure as a Service (IaaS)

• IaaS provides storage (SSD, HDD, backup, file-based access), computation (CPU, RAM, VMs), and networking/bandwidth.
• IaaS consumers are responsible for OS/VM configuration and application scaling.

Data Center

• A diagram showing the layers involved in a data center, including: SAN/Storage/Backup, Computing, Networking, Management and Orchestration.

laaS Provider Challenges

• High cost of building and running data centers.
• IaaS as a commodity business limiting differentiation.
• IaaS consumers easily change providers, demanding high resource utilization to maintain providers' profitability.
• Handling demand peaks presents a hurdle.

laaS Pricing and Costing Models

• Pricing options encompassing availability, support hours, quality of service, software licenses.
• Cost elements involving processes, support, licenses, infrastructure (CPU, DC-LAN, SAN, storage back-ups, facility, tooling).

Cost Optimization Approaches

• Increase storage utilization rates, use high-capacity disks efficiently, implement data compression and de-duplication.
• Increase server utilization rates, use fewer different architectures for servers.
• Simplify software architecture to reduce dependence on specific components.

The Helix Nebula Project - Science Cloud

• Huge amounts of data are commonly generated in scientific experiments.
• Storage and analysis locally may be expensive.
• Cloud computing (e.g., Helix Nebula) can help by providing scalable storage and computing resources.

Helix Nebula: Research Aspects

• In-house IT resources and hybrid cloud solution are central to research.
• Massive resource requirements limit local solutions.
• Transparency in inter-cloud scaling and cloud interoperability (APIs) addressing various infrastructure types are necessary.
• Business aspects of profit, stable consumer relations, vendor lock-in prevention, trust, and data privacy are also crucial.

Helix Nebula: Use Cases

• This technology assists with the computing capacity of experiments such as ATLAS and genomic analysis.
• CERN ATLAS sensors generate massive data each year; a computing center with over 15,000 CPUs is insufficient for processing the volume.

PaaS

• PaaS execution environments abstract from underlying hardware.
• Provides scaling support, additional services (e.g., authentication, DB, integrations).
• Focus on Web applications and includes development environments/IDEs and parallel processing.
• Popular services offering these include Google AppEngine, Windows Azure, and Force.com.

Big Data & Large-Scale Data Processing

• Enormous data volumes, high velocity data streams, and diversified data types.
• Traditional scaling-up approach by adding more storage/processing power.
• Modern scaling-out paradigm by distributing data/processing across multiple machines for enhanced efficiency.
• MapReduce is a common distributed computing method to process large amounts of data on multiple machines.

MapReduce

• A distributed computing pattern to run large data sets on multiple machines.
• Processes large datasets including sorting, indexing, machine learning, and pattern detection.
• Apache Hadoop is a popular open-source implementation, frequently used by Google, Facebook, Amazon, and others.

MapReduce Idea

• Splitting data, distributing among nodes, performing map operations, rearranging results, data processing and combining results in the reduce step, collecting results.

MapReduce Programming Model

• Data handling in the form of hashmaps.
• Separating into map and reduce functions for scalability.
• Distributing data and results among nodes to enhance performance.
• Combining results to generate overall outcomes.

MapReduce Example: Word Count

• Counting word occurrences in a text to illustrate MapReduce concepts.

MapReduce Algorithm with Tree Distribution

• Data partitioning among worker nodes, and results combining through the map-reduce processing steps.

Migration among PaaS Providers

• Abstractions through blueprints, specifying provider features, and supporting provider configurations.

Dynamic Application Deployment

• Using blueprints, the execution environment and deployment using APIs offered by underlying laaS and NaaS layers.

Platform Scalability through Load Balancing

• Handling applications across several servers to improve resilience and scalability.

Other PaaS Features? Quantitative User Study

• Essential features (availability, automated scalability, APIs, security, backup).
• Quantifies essential components of cloud platforms.

SaaS

• SaaS is software provided on-demand to end-users. End users interact via a web browser or thin client interface.
• SaaS providers offer customization options without code changes and restart requirements.
• Challenges include multi-tenancy, elasticity, privacy, and security.
• Examples include Google Apps, Salesforce.com, and SAP ByDesign.

Everything as a Service (XaaS)

• Categorizing cloud services (SaaS, PaaS, laaS) based on the level of abstraction.

Cloud Players Interact via a Marketplace

• Central hub for cloud service interactions (e.g., service discovery).
• Supports provisioning, contracting, and handling transactions.

Marketplace for Services

• Diagram displaying various components (e.g., initiator, contractor, enforcer).
• Details processes such as creating quotes, conducting negotiations, managing billing, tracking orders and delivery status, receiving ratings and feedback.

Examples

• List of marketplace software systems highlighting popular solutions.

4CaaSt Sample Scenario

• A specific scenario outlining the use of cloud marketplace infrastructure to provide services.

Summary of 4CaaSt Technical Challenges

• Key technical challenges related to XaaS, blueprint interoperability, customization, deployment, scale-in/scale-out processes, business terms, and composition-based architecture.

Summary: SOA Extended

• Explanation of the roles involved in SOA (e.g., market maker, service provider, service operator) and their interactions within a service-oriented architecture (SOA) system.

Additional Literature

• Relevant articles and literature on cloud computing.