actionitem

generally - homogeneity of infrastructure to drive simplicity - lower costs and consistency of process; specifically - Azure Stack for Microsoft shops; Cloud at Customer for Oracle shops and eventually Outposts for VMware workloads

Let's start with fear tempered by awareness of application rearchitecture difficulty.

The use cases for hybrid clouds are any app where data is distributed between public and private clouds. Think of database roles (systems of engagement, insight, record, etc.). That describes many e-commerce, multichannel marketing, and machine-learning/AI apps.

operations!!! In a short one word statement. This stuff is hard with a capital H. Between training and governance, customers want a packaged way to consume and manage hybrid. VMware and Microsoft have responded to the call and are ahead of the learning curve.

A key driver is a business requirement to have the data onsite. This can be driven by latency requirements, compliance requirements (GDPR, etc.) and the cost of connection to the public cloud.

If you look at AI/data science pipelines, it's clear that they're often begging for hybrid clouds. You ingest/process/prep data in public clouds, perhaps do modeling and training there, then serve it to apps in private clouds that combine transactional and protectd data

If you can't replatform, running traditional apps as IaaS is a financial non-starter (although may take some AWS bills to realize that). Hybrid is your next option that's more realistic. It's part of why I joined @PureStorage frankly - feels like an achievable vision.

@CTOAdvisor - True. Again - I still wonder what life would have been like if @eucalyptus was bought by someone other than HPE. It would have continued to mature after years of real-world use cases.

It's also clear that many B2B supply chains require hybrid/multicloud architectures for distributed transactions.

Hybrid clouds are often a necessary stepping-stone to more complex reliance on public clouds, providing the migration path for moving data, workloads, apps, and jobs to AWS, Azure, GCP, etc.

- One of the real world use cases we hear from customers is the natural lifecycle of apps. Firms want to start out with Dev / R&D stuff where it can be spun up quickly (public cloud) but may want to bring a mature workload onprem for tighter control and visibility.

@dfloyer I don't think that's why the solutions are chosen. I think data is a driver for Hybrid. In theory, end-users could build on-prem/off-prem solutions based on existing products. It's making it easy to manage. IE we've had hosted vSphere close to the cloud for years.

@CTOAdvisor So much "YES!!" to operational challenges. Learning always has a price tag...but it's ironic/almost humorous when the price tag shows up really fast in the form of monthly bills.

@plburris @eric_shanks also had a great recent post about architecture and cost in the cloud. https://theithollow.com/2019/03/19/its-up-to-you-t...

(edited)

- Enterprises want to prove whether something works quickly vs. investing to build a foundation for something that may not deliver business value. Public sandbox evolves to private permanent landing zone in many cases.

@CTOAdvisor this is spot on. If you make your cloud environment stable, cheap, and most importantly easy to use - it will monetize itself.

One key emerging use case for hybrid cloud is one where the data flows in a "pipeline" from edge to core. Collect data at edge from IOT devices, enrich at edge, minimize in transit, gravitate to the public cloud for long-tail analysis. Lather/rinse/repeat.

(edited)

@ballen_clt True. They were the best in-class when HPE purchased them.

Is the question whether hybrid clouds are the chosen approach, or whether one or another hybrid-cloud management solution is chosen, assuming that you've committed to that approach?

- Cost vs. value is also part of the conversation. Most enterprises focus on run-rate costs but not real transformation and re-training costs. Hybrid cloud stories give them confidence that the non run-rate costs will be more predictable.

VMware users found it very valuable to have some virtual machines running appropriate applications in the cloud, and some on-premises that need low latency links to other applications on-premises.

@eric_shanks also had a great recent post about architecture and cost in the cloud. https://theithollow.com/2019/03/19/its-up-to-you-t...

@dfloyer also costs. High frequency, simple transactions are cheaper to run in servers in the basement. There's a reason old IBM mainframes are still around.

Top use cases for #hybridcloud, @plburris, are:
- Legacy #app modernization/migration
- Risk mitigation
- Connecting on-prem #datacenters to the #cloud for #IoT and other edge scenarios
- A host of enterprise integration/SOA/API/service bus scenarios
- Maximizing old #capex

I find it depends on which direction you are starting your thought process from. Often if users are starting with on-prem going out to public cloud, what they want is to drive out capital costs.

Users starting from public going to on-prem often are dealing with issues like data sovereignty and compliance. These are examples.

(edited)

@dhinchcliffe perhaps #hybrid cloud is just what we're seeing as companies want to innovate but don't want to go through the trouble of replacing old servers. Equifax kept all their data on a single 'giant disk', yet no doubt had products and the website on the cloud.

@Dee_Marketing - I disagree. I see enterprises who make this mistake all the time. "If we build it, they'll come". Many CIO's have already lost the perception battle. Even if they're cheaper than AWS, there is no cachet with deploying to private cloud.

Any distributed cloud architecture is more complex than a centralized cloud architecture. The value of distributed is overall it will lower cost, and with enable applications that are impossible with a centralized approach.

"Hybrid clouds are often a necessary stepping-stone to more complex reliance on public #cloud, providing the migration path for moving data, workloads, apps, and jobs to #AWS, #Azure, #GCP, etc." - @jameskobielus. Bingo

@CTOAdvisor Never underestimate the value of inertia and existing vendor/ecosystem relationships. Sometimes, you just want to bury from someone you know and trust.

@Dee_Marketing - The other aspect is that most CIO's may have the skills to build a cloud but not the accumen to market or price that cloud. Showback? Chargeback? Differentiating features compared to AWS and Azure? They need help to make that case to the apps teams.

@ballen_clt perhaps, but lots of CIOs have lost their jobs over expensive moves to the public cloud or not understanding the security risks.

Another factor is what type of business you are in. Many companies have highly distributed environments (retail, logistics, manufacturing) and those business environments benefit more from hybrid than would a business with centralized logic.

@ballen_clt Also depends on what the developers/users are looking for. Private clouds are light years behind public cloud in terms of innovation and breadth of services. If developers are just looking for VMs, that could work.

@ballen_clt If they are looking for more advanced services, telling them to switch to private cloud is like asking someone who drives a BMW to switch to horse and buggy.

@Dee_Marketing - Agreed 100%. But the fact remains that CIO's have to be able to do more than say "use my cloud because it's better / cheaper. Trust me". The conversation is more about speed and value than cost IMO.

@kenhuiny Really agree with perspective here being about starting points. Having said that, I'd posit that few true public cloud apps/customers are thinking hybrid (multicloud for them). It's only on-prem folks that feel stuck and need something realistic that are thinking hybrid

Ultimately the most responsive/flexible options to the business at the lowest cost by avoiding lockin at whatever level that's important to you.

The full application potential is treating your network of hybrid cloud venues as a new type of datacenter. Think Mesosphere-style abstraction across the global WAN.

(edited)

- Dynamic workload placement. In theory, folks won't even know / care where things are deployed if things are truely hybrid. Communicate your needs and let automation / intelligence put your app in the best place in the context of some guardrails (latency, geo, etc)

@daveshac I love this both from a data management perspective as well as application perspective - you can figure out where things are running but should and eventually can be policy-driven so you're not moving things reactively/manually but planning ahead of time.

My take: The overall potential of full #hybridcloud, @plburris, is:
- Workloads can optimally near the data for the lowest cost
- Security is maximized
- Regulatory issues are minimized
- Offers the higher control topology vs. public #cloud
- Workload #arbitrage possible

The full application potential of a tightly coupled/true hybrid cloud is when it abstracts away the underlying public/private cloud platforms and enables transparent, automated movement of data/workloads/microservices for optimal lights-out performance via AIOps.

I gave a use case yesterday. When I can create Serverless code hosted in the cloud on my Oracle-based CRM in the DC and my security policies are enforced. https://youtu.be/htj0DdvTTVA?t=167

@andriven The Network Is The Datacenter (TM) ;-)

But from an application potential of #hybridcloud, @plburris, you get:
- Longer shelf life of existing #apps
- Improved access to data across/within apps
- Ability to connect external #analytics/#AI/#ML to data anywhere, easily + safely
- Higher performance in many scenarios

The full exploitation of hybrid clouds will come from hybrid applications. If you can send services and code to where the data is, data driven applications can operate in real time or near real-time, and these hybrid applications can be richer in data and functionality.

@daveshac Ha! Maybe "The Network is the Computer" too? ;)

to @kenhuiny point, it feels like we're deciding on definitions as an industry much less maturity. If you ask me what hybrid means, I'd ask first what vendors you as a customer prefer or have been talking with most recently.

Early early days - the most interesting IMO (Outposts) isn't shipping until later this year. Azure Stack finally looks like it's ready(ish) for prime time and Cloud at Customer...hmmm hard to say - we've talked to several customers but still learning how to exploit

We are still waiting on critical parts of the ecosystem to come together before any single solution is mature. I think the poster child for maturity is Netflix. They combined Mesosphere, their container orchestration, and culture change to achieve maturity. /1

Ginni from @IBM called this Chapter 2 of #cloud (with a $34B bet on @RedHat) - we are very early in real hybrid solutions that are not just pieces put together. Management, networking, security, data protection, identity, integration w SaaS... This will take years.

My take: In terms of #hybridcloud maturity, we are getting close to the end of the beginning. But it's taken surprisingly long to get to a point where workable, enterprise-ready solutions have had basic capability that can be relied upon long-term.
#cloud

As well, I'm not wholly sure that the landscape won't shift enough within 3-5 years that increasing mature (before getting fully mature) will be overtaken by landscape shifts. I realize that's more denying the premise of your question than answer it though - apologies.

Maturity could be defined in part as time-on-market. Considering that the oldest of these solutions (Microsoft Azure Stack and Oracle Cloud @Customer) have been on the market for only 3 years, I wouldn't declare them or the market mature.

@stu I won't pretend to be wholly informed or a fan of IBM....but I do like the Chapter 2 construct.

@CTOAdvisor Kubernetes is a long way from being what Titus is for legacy applications. And the enterprise has resisted data center schedulers. /2

@CTOAdvisor Mesosphere is fantastic. But Netflix only started streaming back in 2010, and when they did they started with AWS.

@andriven of course AWS says that it is always "Day One" - I like that one too :)

Technology solutions themselves are relatively mature, what is still evolving and emerging is the EA imagination to consider how/where to deploy business logic across a range of hybrid execution venues. Seamless HC hasn't been around long enough...

Maturity of hybrid-cloud management solutions, as a market niche, will come when each major public cloud vendor has a GA solution that has sufficient number of reference customers and has been through 2-3 enhancement cycles. Perhaps 2021 will be that year.

- No where close to maturity. I always tell customers that cloud is a teenager at best that just started shaving and driving. Able to do some new things is different than maturity.

It's difficult to determine where we are in the maturity curve until we know what is the desired end state for hybrid. Folks often talk about cloud as though it is this static thing which it is not.

customers are not a monolith like Netflix. I can't even get some customers to accept SaaS is Public Cloud. So answering when will a solution become mature? I'd say when AWS itself is mature from a feature perspective. That's who the entire industry uses as a pacer.

(edited)

Maturity will also come when the hybrid-cloud solutions on the market generally support at least all 3 major public clouds and at least 3 major private/on-prem cloud platforms in multicloud deploys. That may be 2024, at earliest.

If all you are looking to do is to connect on-prem to public cloud, then we are just about done. If you are looking to achieve parity of services, then we are at least decades away it's likely unachievable.

Some segments such as tightly coupled and true hybrid clouds are in full swing, with VMware, Azure, AWS, Oracle & Nutanix as key platforms

@Dee_Marketing Just a few years ago. Which makes them the perfect poster child for Hybrid. They actually had to go through the pain of moving monolithic applications between clouds. A lot of lessons for the enterprise.