openEMC

For me, 2016 should be the really interesting year ... private cloud seen as increasingly niche, focus on more on public + public hybrid, concerns of investors in OpenStack space, AWS clearly seen to be running away with it / followed by MSFT then GooG.

Growth of Cloud Foundry as the defacto standard for PaaS (in much the same way that AWS has become the defacto for IaaS) but this time with a strong open source play around it i.e. co-opting where needed, focused on creating competitive market etc.

The work at Azure is really quite impressive, particularly on hybrid. I do think the window of opportunity may be closing, but it might be more like 2017 or 2018.

Cloud Foundry is good, but Docker, Kubernetes, CoreOS are all out to compete with it piggibacking mindshare around container standard

great insight it's the year of "rubber meeting the road"

Back in 2008, I said 2016 was the key date for when private cloud would head towards niche ... I don't see a reason to change but you may well be right, it could be slightly longer.

Its a simple discussion; Cloud Foundry brings you container scheduling, container image management, buildpacks/configuration management for containers, IaaS automation, embedded operating system, high availbility data services..

and in Scheduling, containers, config management, IaaS automation, data services there are competitors

Well, it's not very far away so we're going to see. :)

but only Cloud Foundry does all of that, backed by an insanely well funded enterprise focused startup, with an $80M/year R&D budget

@zer0tweets CloudFoundry != Docker+Kubernetes. Different parts of the stack for the most part. And CF isn't a "container standard"

. @swardley @stu weighed in with #IaaS piece http://premium.wikib...

Cloud Foundry's challenge will be to avoid becoming contextual and locked down over time. That said, I think @wattersjames is right, there is not yet a competitive project that has put it all together.

Absolutely ... of course, I take the view that the fortunes of OpenStack could have been very different if the co-opting AWS play had been stuck with at the very beginning rather than being thrown away of the launch evening in pursuit of differentiation.

By now, I would have hoped for a mass of extremely large AWS clones powered by OpenStack and competing in the public market with syntactic and semantic interop over key APIs ... alas, not the case.

@furrier Stu is off-base. Hyperconvergence is a niche product. Blog posting out after this crowdchat to debunk that myth. It's CI + HCI for certain workloads. Hyperconverged won't displace standard converged.

@randybias CF is not a container standard, yes. It is container orchestrator and manager originally build for a container standard that has been overtaken by Docker.

So, I feel the differentiation play has led to the collective prisoner dilemma that has trapped OpenStack in a private space requiring it to focus on niche ... a pity, a lot of great talent but IMHO misguided play at the beginning.

CF owns the management APIs, Docker owns the container standard and is making a push up the stack

@zer0tweets No it's not. Containers are tangential to it's purpose. CF is more like Java Containers 2.0 (weblogic/websphere) in a modern cloud world.

@randybias it will for specific scale, ie sub 1000 workloads/VM's above that its all RackScale

On the upside, Cloud Foundry has been able to learn from this and hopefully we will finally get that competitive market based upon an open source reference model ... just at the PaaS level of the stack.

I agree on the Private Cloud focus and we've discussed about this on Twitter before. The reality is that there's still Billions of dollars to be made in niche markets. Niche != The Death of OpenStack.

@zer0tweets Containers are a simple technology tool that enable PaaS. Container orchestration is another layer of that. But container orchestration isn't going to create abstraction layers around data storage, messaging, etc.

@zer0tweets Cloud Foundry has positioned itself as a the defacto API for those who are seeking application portability amongst many disparate IaaS technologies (AWS, GCE, OpenStack, VMware, etc...). It's challenge will be service catalog development.

@randybias @zer0tweets would you say that the majority of workloads are container friendly/supported?

my prediction is that companies will do Openstack like distros and foundation will be reduced to rubber stamp.

PaaS is a "load your code and go" system. The developer points at a repo, loads a commit, and the code runs. It's an app runtime for modern cloud use cases. Docker+Kubernetes does not replace that. In fact, it's likely to be integrated.

Containerization is a principal driver for abstraction of storage, messaging, compute

Experience has shown that many such layers, in addition to compute workloads, are container ready

Containerization is a significant step toward PaaS adoption

@shakamunyi and that abstraction seems to be happening in the Docker nascent extension space as otherwise each element was treading on the toes of others - hence Powerstrip for wrapping the API for prototyping

and of all of these very small startups mentioned, only Cloud Foundry has record setting software while maintaining a very open standard: http://www.theregist...

[record setting revenue that is]

James share those numbers again

@blueboxjesse : Oh, I don't disagree that there are valuable niches e.g. Telco / Network Equipment Providers / NFV space has potential. Just that the overall potential of OpenStack (a competitive market of AWS clones) was far greater but never realised.

Prediction:Security finds out what is going on. #Panic. They get steamrolled. Breach happens. More panic. #profit?

@shakamunyi : containerization is an extremely useful subsystem of PaaS, it should be an invisible underlying component ... it's the wrong place to focus on.

it's an enabler - addressed and continues to address many issues at the IaaS, PaaS and application sphere

Focus should always originate from an app-centic perspective

Containerization, like IaaS before it, will need focus, then it can become an invisible underlying component

Blue Box Cloud. ;)

Ubuntu ;)... but I have a horse in that race - all kidding aside it is the most used one per the user studyhttp://superuser.openstack.org/articles/openstack-user-survey-insights-november-2014

does anybody really have the time and resources to try out more than a couple? Or do we just pick the community people we like and by inference the firms they work for?

Cloudscaling. Still the best by a long shot. :)

I believe over time we will stabilize to a handful

Red Hat ;) - but my natural bias aside, what end customers should look for is an OpenStack solution that is completely open with no proprietary extensions

...avoids customizations that lead to non-standard deployments, and one that has a broad ecosystem of hardware and software solutions from which to choose

Where does everyone think VMware Integrated OpenStack fits into this mix?

@cpswan Great point - so perhaps the question is 'What do customers look for when selecting a distributor'? Is it cost, speed, nepotism?

I still think this is the wrong question though. Distributions are the method for end users to consume OpenStack. It needs to be consumed as a service for the vast majority of deployments.

@shakamunyi Disagree. Pushing customers away from anything proprietary is a recipe for failure for many shops. They just need to go into it with eyes wide open.

@shakamunyi Probably not possible if you need real scale and a production environment: http://cloudscaling....

Alex I agree the standard will be a broad ecosystem, and the ability to supply and LTS model that allows for more production stability

Interested to know why one would pick one distribution vs the other. Has anyone built the pros and cons for different distribution vendors?

ecosystem ecosystem ecosystem

@blueboxjesse I meant to say "Distributions are the wrong method..." or my favorite line: "Distributions are doing it wrong."

pro and cons are somewhat tied to use case but I agree ecosystem is the key driver.

with multiple deployments in hybrid clouds, openstack distributions are clearly driving the enterprise value proposition

@shakamunyi Also the term proprietary gets abused. It doesn't mean "closed" source. It means ownership. And technically a company like RedHat "owns" something like Ceph. So Ceph is "proprietary". https://www.google.c...;

@randybias Intersting. Amazon owns EC2 APIs, S3 APIs. So Amazon is proprietary. Why advocate a proprietary API for anything?

scale, hardware choice, existing vendor/distro relationships - these all shape the envelope - I doubt anybody is doing an across the piste bake off

@hui_kenneth From my perspective, VMware's OpenStack pricing makes it largely a non-starter.

@hui_kenneth I view VMware Integrated OpenStack as a very serious contender

Ubuntu's is the most stable and polished in my experience

@shakamunyi For interoperability. OpenStack has got no chance in public cloud.

@hui_kenneth VMware Integrated OpenStack has a stable networking story via solid integration with NSX, something many other distro's lack

@randybias OpenStack already is the true interoperability driver for cloud based on enterprise adoption.

@Gatuss amen brother

@randybias I'm not sure I understand what you mena that it has no chance in public cloud. It already is in public cloud.

@shakamunyi Right now OpenStack doesn't do anything for interoperability. Here's my classic piece on this: http://www.slideshar...

@Krumpe Not at any scale. Azure, AWS, and even GOOG dwarf the OpenStack public clouds. Together they have 80-90% market share. That wont' change easily.

@Krumpe Not only that, but OpenStack public clouds simply don't have the structural advantages and economies of scale that the hyperscale guys have.

@Krumpe Public cloud is a development race: http://www.cloudscal...

OpenStack has no business playing in the Public Cloud space. Agree w/ Randy on this one.

@blueboxjesse Jesse, what's the main reason? Can't compete with AWS, Azure, Goog?

@bvibhu The engineering efforts that go into the hyperscale architectures required for public cloud dwarf the efforts that go into OpenStack.

OpenStack is interoperability for Public and Private Clouds

Only one Public Cloud is positioned beyond SMBs.

And the number of development resources on OpenStack across the community dwarfs pure-public-play cloud

@shakamunyi Disagree with that statement. Cloud Foundry or CMP like Scalr provider interoperability. OpenStack simply provides the best foundation for Private Clouds.

@blueboxjesse disagree. OpenStack provides IaaS level interop. PaaS like OpenShift provides DevOps level interop. CMP like CloudForms provides management level interop

@shakamunyi OpenStack provides IaaS interop between existing OpenStack instances, but certainly not amongst public cloud providers.

All Good points here, however if you view from a Customer's standpoint, other than than their use cases, what will help them to pick the right ecosystem/distribution of OpenStack? Too difficult to try out different POC's.

: pick your favourite OpenStack distro ... gosh, tough ... there's so many incompatible versions to choose from ... oh wait, that's the problem.