Hybrid Cloud Journey - Maximizing Private and Public Cloud

Editor's Notes

• #2: Hello – and welcome Thank you so much for taking the time to join me today – to talk about private, public and hybrid cloud Introduction – introduce yourself
• #3: I love this photo. . . Do you ever feel like this? 1 case – you’re asked to move fast and accomplish a lot with very little (i.e. travel to space with a tin rocket and skateboard) 2 case – it could be that you’re a dreamer – you know the destination and can imagine getting there Final case – you – last weekend after a few too many drinks . . . The technology world we live in requires us to move at rapid speed, sometimes without the right equipment and without a known destination.
• #4: Accomplish in short time My hope is that you leave today with fresh ideas on your journey toward hybrid cloud while transforming and maturing your existing private and public cloud environments
• #5: Public cloud – burst onto scene 15 years ago Young, inexperienced, dreamer – much like our young friend with the rocket ship Crash / burn – but hasn’t – it’s showing us new ways the business wants to use technology we should not look at this as an “or” discussion – but an “and” Goal should be the beautiful hybrid between what public cloud can teach us – and evolving and modernizing private cloud while exposing the unique features that each offers.
• #6: Let’s explore the state of private cloud and some of the use cases
• #7: Why is on-premise private cloud not growing? Are the use cases shrinking? In conversations with clients – that’s not the case.
• #8: Controlling Performance, Security, Privacy and Solidifying Data Ownership
• #9: Legalities and regulation restrict or prohibit public cloud – with a preference on on-premise private -> Healthcare, Insurance, Financial, Utilities, SPs
• #10: Investments into real estate, lifecycle refreshes of infrastructure and operational investment
• #11: Some believe that private cloud is more cost effective Not always true – but not always false Capacity optimization and automation lead to some cost savings
• #12: Challenges with moving, refactoring or modifying core business applications
• #13: Pushing applications closer to the edge -> enhancing user experience CDN -> paved the way Smart Cities Automated vehicles AR and VR Mobile Edge Computing – reduce network bandwidth usage -> 5G
• #14: Still a real issue Fear keeps companies from exploring public cloud BUT – also – FOMO drives companies to hasty decisions
• #15: Let’s explore some of the use cases of public cloud
• #16: Rapid ability to move or change directions -> help business move with the market
• #17: Quickly being able to expand or contract resources On-demand usage Not carrying any more than you need
• #18: Services are iterating faster on public cloud Opportunity to test and validate these services with minimal investment
• #19: You can focus on application and services with main goal of value to users Not distracted with low-level infrastructure overhead, operations or maintenance Rightsizing your environment
• #20: Efficient usage of resources and benefits of cost structure OPEX model may be better for you Automation leads to even more cost savings
• #21: So we know there are use cases for private Certainly know the benefits of public cloud Again – the advantage is private cloud AND public – true hybrid Let’s take a look at some of the most popular hybrid cloud models
• #22: Best venue – also called optimal placement Assign applications a set of filters (or questions) that lead to the workload being placed on the best cloud
• #23: Lifecycle based – follows the software development lifecycle approach Various groups are placed onto the cloud that matches their specific needs for speed/agility, scale, etc.
• #24: Disaster recovery is one of the most popular Production running in private cloud – warm or cold standby running in public cloud Active / active is also an option (although very dependent on infrastructure and application)
• #25: Application is logically divided into tiers Tiers are then placed on the cloud that meets the tier’s requirements (scale, performance, security, etc.) Latency is a big concern here Not for everyone
• #26: Similar to tiered approach Used for tiers that see spikes in usage – not consistently Use another cloud to augment that demand by bursting Loosely coupled applications are better here
• #27: I’m often asked for insight on how others are distributing applications and on which clouds Survey by 451 in 2017 Types of apps and where clients are running them On-prem –> heavy back-office, batch and customer facing applications Public and SaaS -> cloud native, collaboration and E-Business Fairly balanced with hybrid comfortably in the middle
• #28: You might be sitting in the audience thinking “ok – so we should have massive adoption of hybrid cloud”
• #29: We’re not there yet. Clients have deployed virtualization + automation and called it private cloud It’s a great step – but that’s all – a step Continue to understand the benefit of public consumption -> transform private cloud to match
• #30: Reminder of cloud characteristics We do #7 very well. But think about the others and how mature these are in your orgs Modernization means – embracing these principles and planning your strategy around them
• #31: When transforming your private cloud – there are things you can focus on to help with the maturation process.
• #32: Modernizing the infrastructure means bringing our compute, network and storage up to certain levels in terms of performance, capability and operations Architecting and designing our systems for demand and failure Software should begin to enable the business to consume resources at their pace while making it easier for operations Finally – optimizing the environment with automation and efficiency to lower costs Goal: We have to remember that the business need is what is driving here.
• #33: Let’s zoom into the infrastructure for a few minutes and look at a few areas of modernization that we’re working with several clients on I’ve separated into 4 areas – although this is not an exhaustive list.
• #34: Processing Traditional x86 procs -> cores, clock speeds Instruction sets adding more features – enabling higher network IO, advanced visibility (packets), AES encryption, video encoding ARM processors -> smartphones , tablets -> now IOT Co-processing (GPU and FPGA) -> parallel processing - machine learning and analytics CHOICE!!! HCI and Composable Similar -> both Software Defined and flexible infrastructure HCI – distributed system that pools compute and storage together to be managed as one Easier integration points for as a service consumption Modular -> easier lifecycle and overall maintenance Evolving to leverage analytics for things like workload optimization CS – software defined infrastructure Use software to define the hardware components of a system -> cpu, memory, storage and fabrics On-DEMAND Uses unified API to control and provide integration into Think of self-configured infrastructure -> application dictates what it needs – what platform – composable makes that happen
• #35: Capacity and Footrping 14TB HDDs -> continue to improve magnetic recording 30TB SSD with 100TB SSD just announced -> THINK ABOUT THAT Goal – to pack 1PB into 1RU Whether you’d want to. . . Memory/Storage Technology All Flash is mainstream SATA and SAS bottleneck -> NVMe NVMe -> initially servers – now arrays -> what does this do to networks? Bridging gap between storage and memory - 3DXpoint – Non-volatile memory – speeds between NAND and DRAM Being built into arrays Several exciting partners mixing and matching these technologies to offer several options
• #36: Programmatic management Southbound APIs -> YAY! New ways to operate and monitor networks Automation platforms Vendor automation Roll your own – ansible, puppet, scripts Platforms Intent-driven networking Describe desired outcome Algorithms that translate to device commands “Create new VLAN on Boston Core Switch 1” Interesting companies -> Chatbots, NLP - - Also – operational tools that can theorize changes
• #37: Workload or processing unit Containers At EVOLVE – we performed a live crowdsource of container usage – here’s what we found Stand up if active conversations or usage of containers (~50%) Stay standing if you have any workload (dev/test/prod) using containers (~%40) Stay standing if you have any production workloads (~25%) Stay standing if you have a revenue generating application on containers (~10%) EXCITED -> but practical Great tool for speed, distributed apps and HYBRID (portable) Let use case drive usage Work to do -> Security and Networking Serverless Award for the worst technology name goes to -> Serverless Instead of serverless – we will furthermore refer to it as “Mike” So what’s really happening here? Entry point into the code (VM = app level, Container = process level, Mike – function level) Abstraction down to the function level Triggered events – more transaction based – IOT devices RIGHT TOOL, RIGHT JOB
• #38: Making our clouds available, scalable and resilient starts with properly aligning needs of the applications with the architecture of the infrastructure Public clouds have had the idea of regions and availability zones Private data centers should adopt the principles here Regions only Failure domains are at the data center level (physical) Data centers are a region Operate in active/standby mode Regions + Availability Zones Subdividing the region into availability zones Availability zone = a data center Zones work together to protect a region
• #39: Maturing our private cloud environments to offer more automation, orchestration and self-service consumption We start with the infrastructure and determine how many interface touch points we can live with Individually – means there could be a whole lot of integration work System – means we’re relying on a vendor to have built an ecosystem Spectrum of choice Create your own (ansible, puppet, chef, scripts) Other end – vendor ecosystem – possible lock-in Platform players in the middle (as we discussed – modeling used to abstract) Develop our strategy to automate the infrastructure Think about each area of the infrastructure and what functions or services need to be enabled Define your services and the workflows needed to activate those services Orchestration (will be discussed in more detail in hybrid cloud) – coordinates all of the automation needed to enable a service This starts off as an art form (end state of consumption) and ends with science (detailed steps)
• #40: Another aspect of maturing with private cloud – is elevating operations The foundational level is active awareness -> this means knowing the running-state of the system – and relying on alarms, sensors or anything else to tell you something is not running correctly Next level is getting proactive – which involves turning operational data into information – trending – things like capacity and performance management Finally – applying a level of analytics and data science to our environment to have algorithms help us predict situations that may occur One possible path to maturing operations would be: Analyzing the service (define those functional areas, identify dependencies and key metrics for health) Analyze the components that make up the service (their purpose, interdependencies and key metrics for health) Optimize the environment continually – seek ways to simplify – normalize your data
• #41: Looking at optimizing the costs of private cloud – we pulled some research on the ways that most clients are accomplishing this The top way is through capacity planning – maximizing your use of the resources you have – this is what public cloud does very well Automation tools are number three – which allow you to reduce that operational overhead Reusing data center hardware is an interesting one – and attainable if you can properly leverage software to help somewhat “commoditize” infrastructure Walk through anything else you feel is pertinent
• #42: Public cloud – while certainly mature in certain areas – continues to challenge clients with properly architected environments, cost, governance and usage control
• #43: Infrastructure parity really means that in a hybrid model – we have capabilities that are similar across private and public. More than that – we are getting parity in how we manage these environments Has our public cloud been properly architected according to best practices for all services? Is it documented? Do we have proper systems in place to deal with usage, governance and control. Can we introduce new services into public cloud without impacting existing services? Once again – are we optimized to have the best cost structure possible given the needs of the application Same goal – provide what the business needs
• #44: Remember our private cloud automation story? We start the public cloud discussion with that infrastructure automation already done for us The public cloud has exposed APIs to us in each area AND they have already written native workflows for using those services If we needed nothing more – we would be done. If we have needs outside of what the public cloud offers OR we are using a hybrid architecture – we have some developing to do on our own Once again – we need to define what our service workflows look like Leverage what’s provided to us when we can
• #45: Ok – so we have a mature private cloud We also have a mature public cloud These precursors will make it MUCH easier to attack hybrid cloud efficiently Let’s put these pieces together
• #46: We have our clouds – in this case – two public and two private Each of them has maturity to offer IaaS and PaaS with exposed APIs to operate and provision in those clouds We now introduce the idea of Cloud Management Platforms CMPs are there to perform a lot of the hybrid cloud functionality including: Workflows Orchestration and some low-level automation Optimizing the cloud and workload placement Controlling and reporting cost usage Ensuring compliance Finally – we need to expose our capabilities to consumers to use We do this through a service catalog
• #47: Here is an example of a cloud maturity model This model can help show the maturity of nearly anything – we could focus exclusively on private cloud or public cloud – or a specific area of the infrastructure Here we are showing private, public and hybrid cloud maturity We’re listing strength of a scale of 0 to 5 – where 0 is non-existent and 5 is mastery Inside of each area – we have different elements that we’re measuring – where we are today – and where we’d like to be in the future (aspirational) If we look at this example – we can see that our client currently has mastery goals for infrastructure modernization – but currently is only at a 3 The visual shows them areas they need to develop in order to reach maturity goals Walk through other examples on this diagram Point is: We can help clients build these as part of vision/strategy
• #48: Spend time visualizing and documenting your plan – if you don’t have a plan – you’re guessing. You may be right – but statistics do not favor you Ensure that your infrastructure is truly open Talk with partners about parity across clouds Hardware is harder to change than software Low value lifecycles -> commoditize Stay close to industry trends Use closed only if you have to Invest
• #49: So we’ve walked through a lot here. How can we help? Walk them through our unique value prop