Cloud Computing and SLAs

• Study notes about Cloud Computing and Service Level Agreements (SLAs)

Cloud Computing Syllabus Overview:

• Cloud computing encompasses definitions, characteristics, history, and various deployment models like Public, Private, Hybrid, and Community clouds.
• Cloud computing include service models such as IaaS, PaaS, and SaaS, as well as cloud architecture.
• Cost-benefit analysis of cloud adoption is also part of Cloud Computing
• Virtual machine provisioning and migration services cover virtualization concepts, types, hypervisors, VM creation, and containerization (Docker, Kubernetes).
• High availability, disaster recovery, cloud migration concepts, and techniques are covered.
• SLA management, cloud security, and AWS services delve into Service Level Agreements (SLAs), their management in the cloud, and automated policy-based management.
• Security fundamentals, cloud security challenges, vulnerability assessment, security, and privacy are explored, along with cloud computing security architecture.
• Amazon Web Services (AWS), AWS services - Identity and Access Management (IAM), and Virtual Private Cloud (VPC) are covered.
• Advanced topics include serverless computing, edge computing, managed databases (RDS, NoSQL), and data warehousing solutions like Redshift and BigQuery.
• AI/ML services in the cloud, such as AWS SageMaker and Google AI Platform and real-world cloud computing case studies with a discussion on cloud adoption in various industries.

Services and Service Oriented Architectures:

• In the early days of web app deployment, application performance at peak load was a key provisioning criterion.
• Provisioning involved deciding hardware configuration, determining the physical machine count, and upfront acquisition.
• Web applications were hosted on dedicated individual servers within enterprise server rooms.
• SLOs for applications were typically response time and throughput of end-user requests.
• Capacity buildup catered to estimated peak load and determining server and capacity that can serve peak load requests which is called capacity planning.
• Two web applications, A and B, are housed on separate dedicated servers within enterprise server rooms.
• Each application requires three servers to run effectively.
• As web applications increased, server rooms became data centers owned and managed by companies themselves.
• Companies realized it was more economical to outsource app hosting to third-party infrastructure providers.
• Enterprises do not need to invest in expensive hardware without knowing viability.
• Hardware and application maintenance became non-core part of business.
• Managing growing number of web applications increased sophistication which increased cost of maintenance.

Service Level Agreement (SLA):

• Third-party service providers get needed hardware and make it available to deployed web applications in the infrastructure.
• Enterprises enter into a legal agreement with infrastructure providers to guarantee minimum quality of service (QoS).
• QoS parameters pertain to system CPU availability, data storage, and network execution at peak loads.
• A legal agreement is known as the service-level agreement (SLA).
• SLAs state the application's server availability for 99.9% core time and 85% non-core time.
• Service provider would respond to an issue in less than 10 minutes during core time, and in one hour during non-core time.
• These SLAs are known as infrastructure SLAs.
• Infrastructure service providers are known as Application Service Providers (ASPs).

Infrastructure SLA Scenario:

• Enterprises applications are hosted on dedicated servers belonging to an ASP.
• Tools for monitoring and measuring infrastructure availability were developed.
• Availability of infrastructure doesn't guarantee app availability to end users.
• Tools help in tracking SLA adherence.
• Companies IT team performs capacity planning, and the infrastructure provider procures the same.
• Dedicated hosting caused redundancies in ASP data centers because of server underutilization.
• Applications were not fully using server capacity at nonpeak loads.
• To maximize server utilization and reduce redundancies, ASPs began co-hosting apps with similar workload patterns.
• Co-hosting involves deploying more than one application on a single server.
• Co-hosting created cost advantage for ASPs and enterprises.
• The enterprise and third parties perspective are shown before after co-location.
• System is multiplexed between application A and B and capacity of server used by enterprise owing application A is total amount consumed.
• The ASP's perspective of capacity use of application A and B is shown when their workloads are similar.

Challenges of Co-Hosting

• Application performance isolation and security assurances have emerged.

• Performance isolation implies an application should not take resources from co-located application.

• For example, if application A need more from a resource, the amount of resource available to application B is reduced.

• Application B performance affected.

• One application should not destroy or access any information about co-located applications.

• Guarantees for security and performance need appropriate measures.

• ASPs cohosting benefits were limited as a result of these issues.

• To overcome above challenges, virtualization technologies have been proposed.

• ASPs exploit containerization for virtualization to guarantee application performance isolation and data security.

• The applications are encapsulated using virtual machines instead of physical servers.

• The virtual machines were mapped to the physical machines.

• Conserving and noconserving modes can be used to make system resource allocation to vitual machines.

• In conserving mode, a virtual machine that needs more resources than quota cannot be allocated more by un-utilized resources.

• In nonconserving mode, the spare resources are fully used by virtual machine that need it.

• Virtual machines are migrated to another physical device capable of more resource usage if resource requirements are not met.

• ASPs can allocate system resources on demand because of new development.

• Capacity planning becomes redundant because resources are allocated based on application need.

• Enterprises and ASPs do not need to provision resources for peak load.

• Adoption of virtualization technologies required ASPs to get insight into characteristics runtime patterns with high accuracy.

• ASPs allocate system resources effectively to apps on demand and level metrics so the metrics can be monitored.

• Request rates and response times are the level metrics.

• Application SLAs are required because infrastructure SLAs aren't enough.

• Managed Service Providers (MSP) service providers are reponsible for application availability.

• Both Applications A and B are shown to share virtualized server.

• SLOs of application SLA requires greater MSP understanding of its IT infrastructure which must be elastic.

• Elasticity implies scaling up the IT infrastructure to accommodate increased workloads.

• Billing is based on application usage of resources for a period only.

• Infrastructure is augmented by sourcing resources from multiple MSPs if datacenter resources are limited.

• Cloud platforms are a new infrastructure for hosting.

• Difficulties in meeting SLOs occur because of cloud platforms, agreed upon by cloud and application owners.

• Cloud owners treat customer application with third-party components like a black box because of the lack of available high-level design documents.

• Service providers must follow a process for understanding runtime behavior.

Types of SLAs

• Service-level agreements provide seller and buyer framework for profitable service-oriented business.
• They outline the understanding between business and service consumer, forming a basis for relationship maintenance.
• SLA that formally defines legal terms that binds service provider to provide consumer services with continuous access.
• SLA modeled using web service-level agreement language specification.
• WSLA is applicable for hosting applications despite being created to web services.

Key Components of WSLA

• Service levels parameter metric- Observable property of a service whose value is measurable.
• Definitions of values of properties that are computed or measured from a service.
• Metrics that describe what SLA parameters mean when specifying how to compute or measure.
• Function- Specifies how to compute metric values through values, other functions, and constants. Functions are central to describing exactly how SLA parameters are computed from resource metrics.
• Measurement directives-How to measure the metric.
• There are two types of SLAs from hosting perspectives:
• Infrastructure SLA
• Application SLA

Infrastructure SLA Details

• The infrastructure SLA that providers manage and guarantee availability for server machine, server machine, power, and network connectivitiy, etc.
• Enterprises that manage their application that is utilized on the server machine.
• Machines that leased to customers are isolated from machines that served other customers.
• Table 16.2. Key Contractual Elements of an Infrastructural SLA shows practical instances of assurances provided.
• key elements include Hardware availability 99%, Power availability 99.99%, Data center network 99.99%, or Backbone network 99.999%
• Refund of service credit prorated on downtime period and notification of customer within 1hr of complete downtime.
• Internet latency not to exceed 60 msc or Packet loss that will exceed 1% in a calendar month.

Application SLA Details

• In the application co-location hosting model, server availability is given for application that has high recourse request.
• Services which are allocate or allocate can be flexible with co-located appliances.
• Service providers that meet customers that have an customer application SLOs.
• Enterprise application SLA is for a service that is shown in Table 16.3.
• Key service-level include Web site response time, Latency of a web server (WS) or a DB, or an Average latency.
• Measurement include DB latency available via http://mgmtserver/em/latency of WS latency available via http://mgmtserver/ws/instanceno/latency.
• penalty will include service assurance or web site latency < 1 sec when concurrent connection < 1000 and 1000 USD for every minute while the SLO was breached.

Challenges

• Customers that decide upon different SLAS and are cost effective, rather than a single SLA.
• Customers have flexibility to choose any or multiple SLAS.
• Customers at runtime can switch between them.
• From SLA, there is a multi challenges for demand.
• Black box, not know about application runtime parameters therefore proper computer components need to be determined for apps with various workflow.
• The MSP needs to understand scalability and performance.
• Operations enhancement by the update can impact the SLA.
• Capacity planning is the responsibility of the service providers instead.
• It may not be sufficient to meet obligations of the customers the grade may not be simultaneously selected.

Lifecycle of SLA

• Identification starts form the identification of conditions that are stated, as well contract ends.
• Following phases that include Contract definition, Publishing and discovery, Negotiation, Operationalization, De-commissioning.
• SLAs can be generally defined through setting of templates and offerings of service.
• This SLA has offering through templates is from catalog
• Enterprise that customization derived from templates.
• Through service media is advertized and is located by customer by searching of the catalog.
• customers can search for that fulfill.
• application that is for need and is signed that meets.
• Application can be automated that is packaged.
• Manual process for application in cloud and MSP needs to analyze performance.
• At the end of customer and service that agree.
• SLA uses the Specification for [8].

Operationalization

• The three SLA aspects consist of SLA operation, SLA accounting, enforcing.
• SLA monitoring includes measuring parameter values that determine value.
• The parties that are on notification of deviations.
• Adherence has compliance by SLA accounting.
• Applications a s part that's actual is part that guaranteed.
• Penalities is breach also is a apart form the data.
• Action occurs by when monitors it detects.
• other things that needs to be notified is that the sides need some needs actions that has concerned.
• CIM [9] model open expresses management.

Commisoning

• The particlaur happens in SLA to hosting that finishes.
• termination and specifies to which providers that can be ended.

SLA Management In Cloud

• Consist of Feasibility, On boarding, Pre- production, Production phases and termination.

Feasibility Phase

• MSP analyzes application on cloud platform.
• It includes technical infrastructure and financial feasibility (1, 2, 3).
• technical feasibility, that ability need is that applications that scale output.
• Application software's and platform are for compatibility in MSP data.
• Software or hardware and need is available that require running those types of applications.
• Met and preliminary the app and info

Infrastructure feasibility

• Projection that availability to resources is sufficient.
• Financial that customer must pay in the MSP and MSP determines cost approximate.
• 3 studies are conducted that result.
• The customer may agree to proceed.

Onboarding

• Report and findings occur in this activity.
• detailed runtime that are part.

Production

• On agreement that application is that MSP makes of move to hosting as result.
• Platform is moving that called [10] onboarding
• that include profiles that are from part.
• Customer to the to this SLA in offer.
• That guarentee SLO has rules that require help and policies.
• On finishing accessibility needs the applicAtion.
• Onboarding is as follow

OVF standardization

• On deploying there is application and packaging.
• Validating functional has analyzing server's.
• Performance values that non environment gives and provides point on expectation if SLA can provide .
• Identification that intensive the performance needs is CPU to potential needs.
• Virtualization needs characterization is important.
• Characteristics have performances and are scale to bounds with output and up.
• SLOs needed is the characteristics.
• MSP starts set that different policy for automation.
• system deallocating that automatically to set resource management and appropriate component.
• business, operational, and provisional (1, 2,3) policy.

Operational Policies

• Access that resources get in case it helps and prioritize. Weights form group that customers that in case policy.
• conditions that are reached that are followed such logging.
• Application tier is notified addit.
• Actions representation is as policy
• conditions that happen to action happen or has that action taken is for [OP = collection]. Scale as threshold needs of MSP to be greater and web for automatik.
• The instances for such needs should take to increase the that application.

Provisioning Policies

• Defined request and are external. Scale is an example of.
• Actions represented is for PP and request that collection {request and action}
• Provision start must need to server in to create site that contain web.
• On identification with needs needs the.
• That all application are in capturing is onboard.

Preproductiponn

• Hosting starts that is from production.
• Validate MSP has characteristic performance or application runs.
• Agrees that have of that application and is successful.
• Allows that phase or application.

Production Phase

• Accessible from the on start up.
• Different when has application, pre that environments have.
• Terms can be of that breach a non requested or customer to new in include.
• Analysis what respect there policies may or reanalyze.

Temrination Phase

• Application is which to withdraw in end, the activity.
• application related from initiation is to is customer
• information is retention essential with transfer
• It's ends what 2 applications a of end to with signature obtained

Automated Policy-Based Management

• That explains detail and operationalization of the policy.
• Action is specified under the different.
• That functional with and system.
• Relationship helps to identify amount that the quantom and quantity that resource that helps