DXCTechTalk

You can go through NIST definition of Cloud which can make thing clear.

Data Classification and Legal Requirements may shape what you consume.

@ratneshkush Cool. Let me know if you have a link handy

@JerryAOverton, or how can my workload via analytics/ML tell me it could run better on a different cloud, with better operational growth, and outline the commercial value it could bring.

@ckinson Interesting. I've seen best practices for deciding, but allowing the workload to tell you what it needs takes that to the next level.

I had used it my cloud thesis. Can mail that.

If we log everything, then we can apply a hypothesis based in several variables that could identify workloads that have a high chance of success in working in other clouds.

Dby your requirements, Sovereignty rules, compliance may dictate=Private. Price drivers=Public(they have the economies of scale to be able to beat pricing) Most will want hybrid. Develop CNA in Public and move to Private when mature/ready. So Right->Left

@JerryAOverton one of the best early decision calls for "which cloud" can be data classification of the subject system.

DXC Advisory services are a great starting point to analyze workload requirements to determine the best landing zone or platform (public and private cloud).

@ckinson Gr8 idea. It would be better if we had historical performance of workloads across different cloud types. We could train a machine to make recommendations

I think it really depends a lot on your starting point. Is it an older mainframe system today? Can that be moved or should it? Can we add Cloud capability, expanded through APIs without massive shifts? I think we can.

@JamesMi82285791 Cool. Got a link to share?

workload orchestration based on enterprise governalance rules and policies, rules can be driven by AI/ML

we are heading in that direction, suggest a talk with Chris Swan, using the data to collect to make informed decisions about the operation of the IT estate.

@ganeshwrite #PolicyBasedEverything #SoftwareDefined :-)

@ganeshwrite Does this orchestration call for dynamic migration across public, private or hybrid cloud?

Jay Keyse leads Advise & Migration for DXC Cloud Platform and ITO. We use key partners such as Virtual Clarity and Racemi as part of our delivery model.

We could do what we did for devicescape. We could curate the type of data that needs to be processed and can be based on the compliance rules (in-house, public cloud or hybrid based on geofencing, data type, transaction type).

@ckinson This approach would apply to IaaS/containerised workloads, correct? It would be difficult to apply this approach to PaaS currently.

Dynamic launching of workloads based on policies and rules. Pre-designed platform agnostic blueprints to avoid migration hassles

smaller steps, the identification is a critical stage, if i could look across a clients estate, and with confidence provide a heat map of workloads to move, i can then invoke a transformation. Depending on the underlying tech, i then have options.

Learning is definitely possible. But we need to operate more to generate data and learn out of it.

I think it's a simple balance between risks of exposure and the need to collaborate/share with outside entities with ease.

Heatmap of workloads is OK if they are relatively independent. What about linked and interdependent workloads?

Also, there is not special magic with cloud IMHO. It's just another platform (or several). Don't think we want to limit ourselves or think some workloads are better than others. We might be surprised.

@mvita1297 What about considerations like CapEx and speed to market?

@NewInstEcon the problem is that heatmaps move, and thus the services must move #fastfollowers much of the research in #distributedcaches #distributedrouters will become substantially relevant as optimizations in these new hybrid models

I would go for the following parameters : Data Classification ; TCO; required functionalities from the infrastructure (auto scaling, temporary available infrastructure, specific functions you find with public cloud providers); Integration requirements

Process moving to data at the edge, limit the need to migrate very large data sets

@jpmorgenthal So the "gravity" is the data sitting at the edge, such as location data, other mobile data...

#DataGravity is a property that suggests that data is heavy/expensive, becomes difficult to move, and a data corpus if valuable enough/big enough attracts other data. Think anchor stores in shopping malls. @davemcrorry

@jpmorgenthal Maybe think of it as the driving force behind the emerging edge computing?

@DanHushon Ah, so the net is to go to the #data instead of bringing it to you

Another way to think about is a planetary system in which all planets have gravity, data just happens to often get stronger than apps and services

It is multi-speed transmission - In our Makkah crowd example, we analyse for patterns at the edge and transmit the actions near real time. We send the rest of the data separately as it has lots of mining value (e.g. Videos, point cloud etc)

sure #edge & #core but we need a federated/distributed query optimizer/director that includes movement strategies based upon intensity of #outerjoins and #shufflesorts

Also who has a consistent API across Public/Private? Agility helps bridge some of that. Azure Stack is basic compared to Azure Public but gives consistent platform for Hybrid cloud app deployments. VMware on AWS needs to catch up.

I'd say when you have monolithic and entrenched systems (e.g., mainframe, dense data, maybe older tech like VSAM) that is difficult to move (has a lot of gravity) so what can we do to maybe just orbit that planet?

@jpmorgenthal apps are just ways to access/append data, Model[information]Controller / ViewController remain divisions of labor?

Does the analysis generate a #data exhaust? I'd love to analyze it for patterns-- train a machine to estimate the data's gravitational pull

@DanHushon Apps do have their own gravity especially if you consider that it might operate over multiple data sources, then it would mostly likely gravitate toward the larger of those

@OhneMoose you could build a #solidindex across VSAM, or you could finally bite the bullet and use a modern scale out information base? that you could stretch w/ appropriate consistency dynamics #BASE [vs. #ACID]

@OhneMoose I think you captured #DataGravity nicely with the idea of "difficult to move"-- which makes the comparison to gravity make even more sense

@danhushon "Go to the data" therefore is even more important if by association other valuable data becomes available

@Faisal_Siddiqi #GoToTheData vs. #MakeTheDataEasierToCombineForValue

@Faisal_Siddiqi ...another nice gravity analogy-- the idea of stuff attracting more stuff and gaining pull in the process

@Faisal_Siddiqi move away from classic ETL approach to combining to create a structure when required for the #BusinessMoment

@sukhigill #API #discoverability is giving us an alternative to ETL +1

Doesn't #DataSupplyChain turn data into useful business information?