cloudAI

Scott, to answer that we need to acknowledge how difficult a problem that is and why traditional solutions have deficiencies

a) make the IT infrastructure aware of application requirements and business goals, b) "observe how the application is used and how that impacts the infrastructure then c) make placement and optimization decisions on that basis.

@AshtonMetzler Thanks Jim -- what are the biggest deficiencies, in your opinion?

Key is to understand that best practices cannot just be blindly applied, but need to be viewed based on their impact on business goals and other applications that will be influenced by a placement or optimization decision.

intelligent software algorithms are used all the time in anomaly detection (credit cards). Helping people troubleshoot can save time / money. Customers tell us cloud is complex, helping them find a faster means to performance issues is valuable

So many apps are highly distributed and so multiple pieces of software need to be monitored. Also, the infrastructure is comprised of physical and virtual resources, and or course, software can move around.

@TorstenVolk Are these best practices being applied, in your opinion?

At a conceptual level, the value of AI and ML for optimizing app performance is huge, as AI and ML help you uncover correlations between app performance and infrastructure utilization and configuration, even across enterprises, that you'd miss witho

Also important to realize that machine learning is being widely used in many disciplines, notably to improve sales and marketing. Point being, this is not some untried or unproven approach.

Yes, the problem is that they aren't context aware. They don't know the consequences of blindly applying them and they don't know how important these consequences are to strategic or tactic business goals. Example: It may be OK for a certain

How available is this technology today? It's early days, no?

application to be slow, in favor of another set of key applications to hold up to the pressure of highly strategic key customers. Once these customers are done, the lower prio app can be taken care of

I won't call Ml a mature technology, but it is building a track record.

Customers mostly aren't aware of it and vendors are only now starting to think about it. There isn't typically even a clear definition of how CloudAI and ML differ from traditional Ops management.

@AshtonMetzler thanks Jim

#ZeroStack comparing CloudAI to a self driving car is exactly right. We strongly believe that cars will be self driving in a few years, but by far not enough thought went into the Business Centric Self Driving Hybrid Cloud.

And Amazon does offer ML as a service - so it is certainly available

@AshtonMetzler Jim we agree, we have a crawl, walk, run mentality, we are rolling out capabilities in phases

It's available in its generic form. You can try it on #Bluemix, #AWS, #Google, etc. but there isn't a "plugin": that makes vCenter business aware. Or that at least evaluates your vSphere management decisions based on business impact.

it all is based on useage, workload size and workload characteristics. We see the world moving to both a mix of public clouds and on prem clouds -- hence hybrid cloud as a thesis. Place workloads where they fit in terms of cost and performance

There is no doubt that public cloud can be a good way to get started with an app, or to house an app that does not have heavy usage.

There is also no doubt that the costs of a public cloud can definitely add up based on the characteristics of the app.

So, buyer be warned. Do some arithmetic on costs and don't just assume that public cloud saves money. It might and it might not.

define 'larger'

I think that hybrid cloud (put what you want where it makes sense and possibly move it around when things change) is a great goal. Not easy, but great.

I have seen a cutoff number of $15K per month. @9muir muir your costs may be higher LOL

A problem is visibility with pricing models. New pricing models can quickly inflate costs. Ask Oracle users...

Unfortunately Chris is right.

But that goes back in part to my comment that one of the downsides of public cloud is losing control.

current employer is definitely above the cutoff. Not so clear for previous, but I'd say >>$15k/mo

Agreed w Jim. But will SP's offer long term pricing contracts? Portability is often not a (cheap) option?

Just to clarify, I think that so far we have used "public cloud" to refer just to IaaS.

#Security concerns, cost-benefit analysis over time, success of SecDevOps maybe?

I also include PAAS. party foul?

We don't think the world is that 'digital' - public cloud can have many personas -- think DR as an example, or as Jim noted below, getting started in cloud. The issue is do you need a multi-cloud strategy, and do some of your workloads need to be on prem?

They have huge economies of scale and a very strong differentiator each (local integration for @Azure and APIs for #AWS). What can stop them is security breaches or, as happened to VMware, getting greedy with their pricing.

For a lot, if not most, small to mid sized companies, public cloud in general, and SaaS in particular, will be popular for a long time

@TorstenVolk Nice callout on VMware!!

@ZeroStackInc Yup, we need a "brain" for the software defined data center / hybrid cloud that makes business policy driven decisions instead of technology decisions. There's no good cloud bad cloud. Just suitable or unsuitable cloud.

Stop cloud? A screw up as Torsten mentioned, or the coming to life of effective hybrid cloud solutions.

Agree with Torsten.

@AshtonMetzler that's what I'm getting at if these solutions "come to life" ... there is demand building you think?

well, hard to deny it... And of course striving for lock-in is also another killer for any cloud, public and private. I like Azure's strategy in that regard. Some lock-in, but well disguised :-)

@TorstenVolk We agree with you here Torsten - and we like your thinking around a all-seeing cloud brain! :) That is how we think about our direction long term

@ZeroStackInc 1984 in a good way :-). There shouldn't be secrets when it comes to relating a LUN to a business goal...

I think the market would be very receptive to hybrid clouds, but need to feel comfortable that it works well.

Yes, it is a race for all cloud vendors against getting commoditized. And leveraging AI to make the right management decisions and to monitor your infrastructure in a business centric manner is tomorrow's biggest differentiator in cloud.

Having spent over a third of my career running networks at a couple of F500 companies, I believe that organizational issues always impede implementing a significant change.

the agony of choice - going for proven proprietary cloud environments like VMware's vRealize Automation or following the OpenStack hype - and ZeroStack might be not so well known in Germany yet

For private cloud to be successful all of the various groups have to be in real alignment, not just nodding their heads in various planning meetings.

Is the momentum for #AWS and @amazon self reinforcing -- people are afraid of DIY?

Jim spot on culture is one challenge, trusting the 'machine' is another. And frankly vendors developing the tech so it really works as well :)

My comments are not intended to diminish the technical issues, including how to orchestrate and manage dynamic resources in a secure, highly performing, cost effective manner.

And it is very difficult to demonstrate the ability of a solution to learn over time - but to me that is a key part of the value prop.

Indeed big companies tend to believe in DIY, that said, AI democratizes cloud for any org, and that changes the adoption economics if you will

@ZeroStackInc makes sense ... for the big organizations, the economies of scale are definitely in favor of DIY

Yes, you can't tell people: "hey, now a robot is making the decision for you." It's about transparency and assisting the admin, by showing trade off and business impact.

How can I turn my vSphere environment into a ScaleOut cloud? You don't want to leave vSphere, but you also don't have a viable solution to use vSphere for ScaleOut without a third party solution. OpenStack didn't work out... lower OPEX is needed.

@AshtonMetzler Right, and companies that tried it got themselves black eyes and bloody noses. It's the divide between the VM based scale up IT and the scale out paradigm. Pets versus Cattle.