Dell Technology World - IT as a Business - Multi-Cloud Strategy is your Product

Editor's Notes

• #5: We’re seeing most of the leading organizations going to a multi-cloud model. Why? Let's look at the differences between hybrid and multi-cloud. Hybrid Cloud endpoints leverage the SAME core technology on and off prem. So migration from cloud to cloud does not require VM conversion. Multi Cloud leverages cloud endpoints with DISSIMILAR core technology. So migration from cloud to cloud is more difficult and requires VM conversion. The architecture you choose will drive the decision of which of these models are best for you Starting from the bottom, Traditional infrastructure is a scale up model. These monolithic applications are usually dependent upon infrastructure to provide availability A Cloud Tolerant architecture looks and feels like traditional infrastructure, but adds with ample redundancy so that cloud architectures can service greater needs A Cloud-native architecture delivers scale Out Application Resiliency because it is handled in code/architecture
• #6: Cloud is not a single destination, and not everything goes to the public cloud. If you look at the 2018 RightScale State of the Cloud report that just came out, you’ll find some interesting statistics. More than a thousand organizations responded to this survey.   81 percent of enterprises have a multi-cloud strategy,   Organizations leverage almost 5 clouds on average. The “portfolio of supported clouds” idea is taking hold as the new role of an IT organization. IT organizations must plan for a future that includes the ability to deploy the right workload to the right cloud. ------------------additional notes ----------------------- • Respondents are already running applications in 3.1 clouds and experimenting with 1.7 more for a total of 4.8 clouds. • Among those using any public cloud, the average is 2.7 public clouds used. • Among those using any private cloud, the average is 3.0 private clouds used.
• #7: Cloud is not a single destination, and not everything goes to the public cloud. If you look at the 2018 RightScale State of the Cloud report that just came out, you’ll find some interesting statistics. More than a thousand organizations responded to this survey.   81 percent of enterprises have a multi-cloud strategy,   Organizations leverage almost 5 clouds on average. The “portfolio of supported clouds” idea is taking hold as the new role of an IT organization. IT organizations must plan for a future that includes the ability to deploy the right workload to the right cloud. ------------------additional notes ----------------------- • Respondents are already running applications in 3.1 clouds and experimenting with 1.7 more for a total of 4.8 clouds. • Among those using any public cloud, the average is 2.7 public clouds used. • Among those using any private cloud, the average is 3.0 private clouds used.
• #8: Cloud Management is the “Control Plane” that enables deployment options.   Applications are the workload. A cloud operating model requires deploying the right workload to the right cloud – depending on business and technology considerations. We’re going to discuss how the right cloud management platform enables organizations to maintain control and flexibility over workload deployments.   <click> Cloud-native applications are developed by leveraging a Cloud-native Platform or Platform as a Service (PaaS). The platform provides support for the development and deployment of cloud-native workloads through language support, and by providing connectors to backend services that may be required.   <click> Traditional applications can be deployed and managed more efficiently using the cloud approach for Infrastructure Operations, commonly known as Infrastructure as a Service (IaaS).   <click>There are pools of compute, network and storage resources that are controlled by the cloud management layers above.   <click> In fact, the connection between the cloud management layer and the resource providers is enabled through a series of APIs that allow the cloud management software to manage the resources and provide cloud services. This may be from a software-defined infrastructure such as a vSAN data store or from more traditional components like SAN storage with API-level integration with the cloud management stack. HCI, CI and traditional best-of-breed components all work here. However, most large service providers are delivering their cloud services on scale-out software-defined building blocks like VxRail or VxRack from Dell EMC.   <click> Dell EMC recommends Pivotal Cloud Foundry as the right Cloud-native platform for your multi-cloud infrastructure and VMware as the right IaaS layer for your multi-cloud architecture.   <click> Choosing the right multi-cloud management – VMware for IaaS and Pivotal for PaaS – will enable true control of workload deployments so you can provision the right capacity, performance and service levels.   <click> One of those deployment options is private cloud. VMware and Pivotal can deploy workloads into a private cloud.   <click> Another deployment choice is a hyperscale public cloud. VMware and Pivotal can also deploy workloads into a hyperscale public cloud – giving control over workload placement.   <click> Additional choices include Enterprise Cloud services such as Virtustream, Dell EMC cloud service provider partners, the IBM Cloud etc. Both VMware and Pivotal can deploy workloads into Enterprise Cloud services giving yet a third deployment choice for workloads.   The right cloud management enables true multi-cloud deployment flexibility.  
• #9: Today we are going to talk about 4 pillars to building a muti-cloud strategy There are many more , and again our hope is this becomes a vehicle for dialog and sharing best practices , maybe grow this to be a bird of a feather session next year So today we are going to review why your multi-cloud strategy should be BizOps focused Marketing enabled App centric Migration ready Some of these pillars are technical in nature and some of these are business and service in nature But they all intersect They all urge us to think of IT more than a service , more like IT as a Business bc they incorporate startegies that were missing from the ITaaS narrative Additional bullets what we see in the field aligns to business incorporates different application needs covers the new and old world aligned to what is available today is honest about the limitations and opportunities Incorporates on premise Incorporates off premise incorporates different deployment models – SaaS, PaaS, IaaS covers all layers of infrastructure – yes including backup too covers people and process ---- The most important!!!!!
• #10: What is a “practical” multi-cloud strategy? Let's go over the blocks briefly Practical also means that we also look at trends related to customer deployment of cloud operations and align to better serve our customers. And one that has been validated by the State of Cloud report and – is that the industry is moving to a multi-cloud model , Here are some trends we see among our customers A Multi-cloud strategy better aligns to the needs of the business From an application perspective it easily incorporates different application needs, covering the new and old world and aligning to what you have available today Incorporates on premise and off premise applications and different deployment models such as Software-as-a-Service, Platform-as-a-Service and Infrastructure-as-a-Services It covers all layers of the infrastructure – including backup – and empowers your people and process ---- The most important point!!!!!
• #12: What are the essential functions that drive value in a cloud model?? It needs to provide a catalog for self-service and control Tools for orchestration and automation must be included The financial impact and cost of IT services must be measured, so for business control And orchestration and administrative tools must be embedded
• #13: What are the essential functions that drive value in a cloud model?? It needs to provide a catalog for self-service and control Tools for orchestration and automation must be included The financial impact and cost of IT services must be measured, so for business control And orchestration and administrative tools must be embedded
• #15: Let’s talk about service costing. Everything starts with the service catalog. The service catalog contains the service reference architecture, which identifies everything needed to deliver the service. This includes hardware, software, networking, facilities, personnel, external services and overhead assets. The service reference architecture is input to an asset alignment process, where IT financial data assets and physical inventory assets are mapped to the services to determine the component costs. Performance data from monitoring systems (e.g. Dell EMC ViPR SRM, VMware vRealize Operations, BMC monitoring tools, IBM Tivoli monitoring tools) is input to a capacity management process which performs historical trending to determine projected service utilization. After you’ve determined your service component costs and projected utilization, that information in input into appropriate cost models to determine service TCO.
• #16: What are the essential functions that drive value in a cloud model?? It needs to provide a catalog for self-service and control Tools for orchestration and automation must be included The financial impact and cost of IT services must be measured, so for business control And orchestration and administrative tools must be embedded
• #18: There are two perspectives that we see shaping your cloud products one thinking about the applications and the deployment artifacts they require, and two the customer value stream We will look at the customer value stream in more detail , and ask the fundamental question – who is you customer - but first lets go into the application perspective Note that we find these perspective key to developing cloud Products regardless if you are improving exiting cloud products or just starting to build your cloud team
• #19: When we say artifacts what we mean are the mechanisms for abstracting, encapsulating, and running applications. This is the where of the applications Think virtual machines , containers, etc Interestingly enough these become the contracts between applications and the infrastructure they run on
• #20: In general, there are two high level application patterns we see in our environments There are probably three – if we include mainframe The first is the heritage architecture we know and love – the 3 tiered architecture , sometime deployed as a monolith , sometime partitioned into 3 pieces with some horizontal scaling – especially in the presentation tier; The presentation layer or web layer – typically a stateless layer The Business logic layer or middleware Then the data tier the hardest to make cloud ready or native This pattern is usually more difficult to scale horizontally at the middleware or data tier However more applications are now being built with a new architecture pattern that we call event driven and non-blocking We call it event driven bc this app pattern is optimized for ingest of events/transactions from devices or customers into a high capacity low latency queueing mechanism, to facilitate scaling in a data driven world, and we call it non-blocking bc the receiving of events is not limited or blocked/bottlenecked by the underlining components processing the events – sometimes these patterns are also referred to as asynchronous or streaming patterns This pattern will contain some key components A distributed event queue/bus – technologies such as Kafka Distributed data stores for operation data access or long term retentio semi-ephemeral analytics engines like SPARC to process both data from the queues and from more long term distributed data stores And of course stateless microservices – or mini services as some customers are calling them
• #21: The reality for modern It enterprises is that you will be deploying both of these architectures patterns and some cases intermixing the components in the patterns for some application domains
• #23: So lets dig in to what I mean by infrastructure artifacts that deploy and run our application components For the last 15+ years we have been encapsulating the tiers of heritage / tradition application patterns in VMs (whether x86 VMs or Lpars) This has driven a lot of efficiency in our operations by giving us isolation of application over baremetal and will be around as a key way of abstracting and encapsulating infrastructure for a long time So however we think about cloud products we will need some form of API driven target for VM instantiation , I call this the VM bucket Of course most ISVs still use VMs as the artifact for delivering and running their applications VMs can also be used to deploy the components of the newer cloud native, event handling application pattern
• #24: However we are always trying to find a better way of doing things in technology – and especially how to deploy and run complex scaled cloud capable application components [CLICK] And in 2013 we saw linux containers – that have been around for more than a decade - get a major boost from Docker. Because of the new tools from Docker that let us package and run containers and make deployment and configuration mgmt much easier efficiencies, it started making sense to put the stateless presentation layers in containers [CLICK] We get encapsulation and isolation but also get better standardization and speed in deployment It also makes sense to put and run stateless logic and middleware tiers in containers Note that the choice of whether to run a VM or container still depends on the application – whether it is built or bought – if its COTs- and whether the component can be refactored into a smaller form factor Looking at the second application pattern [CLICK] Makes great sense to think about stateless microservices components that under pin it But what about the stateful components? [CLICK] The growing maturity of persistent volume management of container orchestrators is a real thing 7 of the top 10 docker public registry downloads are data stores and DBs Containers are being strongly considered for deploying the stateful components of new event driven patterns such as distributed DBs, event queues, and analytical layers Also we are starting to see the ISVs and COTs apps , especially in the network world , start to deliver their applications through containers So just as important as the VM API endpoint bucket is also having an API driven container orchestrator / bucket [CLICK] Of course the undisputed winner , by unanimous decision of all container platform vendors is Kubernetes Great thing for us , is you have two options now…
• #25: There are many possible deployment approaches using VM and container artifacts – lets assume that you your deployment model looks like this You are leveraging VMs for most of your heritage 3 tier applications – some containers for the stateless tiers - and lots of containers for your next gen cloud native apps This is where you potentially have another optimization option - remember we don’t stop in the industry thinking about how to improve Who here runs more than 1000 container pods For those that do you can probably share some great stories about the operational concerns of running containers – thing like setting up a registry governance model, security scanning , multiple pipelines for containers , load balancer integration/deployment, credentials management Containers provide great efficiencies, but not without adding new operational models especially in Enterprise So the question I ask myself is – What is the best artifact to use for their deployment and management? What if I could just have developers push code , and abstract all the container management, and have the governance built in Yes you could also use containers just like we discussed in the 3 tier app’s presentation and middle ware layers, however there may be a better ways to deploy and run stateless applications Your goal should be to minimize the management of any artifact including VMs and containers I would also like a bucket for just deploying code – an API driven orchestration layer that abstracts container and security management – and optimizes flow
• #29: When thinking about value streams – we have to think about perspectives of the stakeholders in that chain Lets start with Infra/ops – VP of infra - Linus’ perspective is from the server request and delivering/supporting environments
• #30: There is another perspective – the VP of Apps/Dev - Margaret Hamilton – and she is focused on managing feature and app requests – code/integrate/request/deploy/test
• #31: Maybe there is a better perspective - the perspective of CIO Sir Tim and/or LOB James G - lets be honest they want to facilitate ideas and get to value and happy customers – and everything in between is just stuff
• #32: When building cloud products , especially if you are not mature at IaaS, you should think about the entire value stream
• #35: Goal is to be customer focused Share story of $2MM ITSM cloud project for Linux VM fully automated vs prototyping sandbox statement from some influential IT product person – Design thinking focused and customer focused
• #36: Design thinking is a method for creative problem solving that focused on outcomes and customer value through a a data driven iterative approach We can trace the roots of design thinking to several design , engineering, architecture and science works from the 1950s, 60s, 70s , and 80s However it was Stanford professor and business man – David Kelley who adapted Design thinking for business purposes and popularized the method by creating the design consultancy IDEO “coolest picture of a Stanfard Prof - ” What I like about design thinking is that it strives to identify and investigate the unknown and ambiguous aspects as well as the more known and visible aspect – of a current situation - in an effort to discover parameters and alternative solution sets which may lead to one or more satisfactory goals In a cloud world moving so fast – where developers and applications are adjusting monthly – we need a new method to develop cloud products
• #37: A typical design thinking approach to cloud products may look like this Share story of $2MM ITSM cloud project for Linux VM fully automated vs prototyping sandbox collaborative learning
• #40: You cant improve what you don’t measure
• #44: What are the essential functions that drive value in a cloud model?? It needs to provide a catalog for self-service and control Tools for orchestration and automation must be included The financial impact and cost of IT services must be measured, so for business control And orchestration and administrative tools must be embedded
• #45: The reality of enterprise is that you have to run at least two stacks , maybe even three Each of these stacks are aligned to the different instances being used – virtual machines and containers And each aligned to a different application model or architecture Both of these approaches, regardless of the technology or application, benefit from a cloud model.
• #46: And different application architectures and stacks align to different off-premise cloud endpoint options depending on their availability and compliance needs
• #47: Here is a summary of the Maturity Model for cloud management functions We see that our customers are at different maturity levels in each function , however in general the move toward a better ITaaS model around multi-cloud will drive customers to the right
• #49: What are the essential functions that drive value in a cloud model?? It needs to provide a catalog for self-service and control Tools for orchestration and automation must be included The financial impact and cost of IT services must be measured, so for business control And orchestration and administrative tools must be embedded
• #50: What are the essential functions that drive value in a cloud model?? It needs to provide a catalog for self-service and control Tools for orchestration and automation must be included The financial impact and cost of IT services must be measured, so for business control And orchestration and administrative tools must be embedded
• #56: In context of Workforce transformation we want look specifically at the workforce catalog – ordering new devices, new peripherals, new applications. We’re also looking at scenarios where you’re managing your existing assets such as requesting to add memory to your laptop to improve performance or adding more storage. In that context, we’re providing all the components for that part of your catalog in order to deliver a well defined catalog to users, with an automated fulfillment process and tying it to the factory and/or your supply chain of how you are procuring things as well as ties to your procurement systems. You also need to make sure this is integrated with your licensing and entitlement solutions and that the device and asset procurement is synchronized with your asset management inventory.
• #57: Customer is JPMorganChase. Was originally referenceable but main stakeholders have since left the company. But still a great story to tell. Script: This customer is one of the largest banks in the world, with hundreds of thousands of employees. They had a few pain points they were looking to solve. First and foremost, they wanted to reduce their help desk and call center costs. Second, they wanted to begin to change the perception of IT and essentially offer up a consumer-grade marketplace for employees to understand what IT offers and also to provide self-service or “Tier 0” transactions. So what we did was we identified the Top 10 IT help desk issues that were costing the customer money, and put those on an agile product backlog. We started with password reset, and created a simple mobile web app hooked into their identity systems through micro-services. This small mobile web app was so successful in reducing help desk calls and costs in the first 3 months that the customer ultimately funded the broader IT self-service portal to advertise most of the service catalog and to provide self-service for top transactions and requests. The technology stack was a custom, modern JavaScript web front end, a micro-service architecture, and then a bunch of REST APIs provided by underlying platforms and systems. This whole solution is based upon a solution offering we have which we call “Unified Service Portal”, and we can use that framework to accelerate a similar journey for other customers.
• #58: First, let’s talk about the multi-cloud infrastructure. There are three main options for cloud solutions from the Dell EMC portfolio: VMware, Pivotal and Microsoft.   Clearly VMware and Pivotal are in the Dell Technologies family and Dell EMC has been a leading partner with Microsoft for many many years. Each of these offerings feature the Cloud Management Platform software from their respective companies.   <click> Each of these flavors has an on-premises deployment option powered by hyper-converged infrastructure from Dell EMC. We’re focused on providing the best HCI for cloud solutions. These solutions on premises provide a true private cloud that can be part of a customer’s multi-cloud ecosystem.   <click> The software stack for each of the offerings is provided by the partner – who is responsible for the lifecycle management of that part of the stack. VMware provides the best IaaS for traditional applications. Pivotal provides the best cloud-native platform for developing and deploying new apps. Microsoft provides an Azure ecosystem that has a consistent experience across public and private infrastructure. These cloud management platforms are the control plane for managing workload deployment.   <click> For each of these environments, Dell EMC provides enhancements that bring additional value by solving common challenges such as storage, data protection and security. We also provide a complete set of services to help organizations plan, deliver and manage their cloud solution.