LocalData

workloads that are very latency sensitive simply don't need it, the cost of coordination between nodes isn't worth it

Adding the same hardware to a solution without being able to use it. Think DataDomain. You would buy a solution for the capacity you needed then when you needed more capacity you would have to buy a whole new DataDomain and use it as a new target

Most of EMCs current products.

Adding more disk shelves.

My network closet.

Oracle Databases... Most businesses simply throw TB's of RAM at them to make them run better and avoid licensing charges.

Legacy batch HPC approaches.

Anything where you say "I loved my first; I hated my ninth" - e.g., NetApp, Data Domain, etc.

Truly, most cloud services are not transparently scaleable. You still need to provision individual servers and services incrementally.

@dinisco going back to my EMC days, I do remember candid latency comparisons between Isilon, VNX, and VMAX with some solid technical explanation about the latency involved in any scale-out architecture.

the need for a large namespace warrants it, but there are cases where isolated islands of data make sense

Sounds like the argument about what is truely hyper-converged. :)

It's not scale out if your workload is has to be sharded and mapped to specific nodes so that each time a node fails or is added, you have to manually remap everything.

It's not true scale-out if adding nodes does not provide linear or near linear increase in performance.

When one of us uses Azure and the other uses GCS, and we have to interoperate.

it's the happy, buzzy, credibility-building successor to "hybrid cloud"

it's the reality that we live in now that there are legacy cloud environments.

Hybrid cloud is a compromise. Multi-cloud means you've removed the distinction between on-premises and public cloud.

Pricing arbitrage, feature arbitrage

@jhughes I know zero people who actually do pricing arbitrage between cloud vendors.

Go to LHC1809bu at VMworld to find out! I have been asking people this question for a while. Most say it is when they use multiple clouds. Not a fan of that response.

if I'm a bit less cynical, I see it was companies living in multiple IaaS or PaaS environments, doing so intentionally, and thinking about how to move workloads back and forth. SaaS sits outside of that.

@fdmts this is my pet peeve bc many just think multi-cloud is apps on one cloud and others on another. Not really multicloud in my mind. Latency is huge issue

@johna_white Well what's a better one? Wide area ESX?

AWS ☑️ Azure ☑️ GCP ☑️ - all the ☁️s

Cloud is not one destination. Most enterprises will rely upon a collection of Multi-Clouds.

why use one cloud provider when you can spend twice as much with 2

Smearing a single app across multiple cloud providers is almost always a mistake.

@fdmts It's a "nice to demo" feature but even if you do live costing the portability is too hard to be worth it often due to the weight moving data. It almost cries out for a cloud data management layer. ;)

Hybrid cloud will be the way of the future once you get common data and network planes working across multiple clouds. It will allow for much better decision making and curation of cloud.

There is a spectrum implied in multi-cloud - to some its JBOC (just a bunch of clouds) to others it's federated hybrid cloud with application logic spanning clouds

Data at rest in the cloud is cheap. Why lock yourself into a single cloud provider.

@patrickrogersca Lock yourself into a single cloud provider and get on with solving business use cases. Not worth the time to optimize between the major public providers.

@fdmts Actually, I don't know if the multi cloud application here is really arbitrage but rather using other clouds like Azure when you want to integrate with Active Directory or Google when you want to use some special capabilities (AI / ML) primitives.

Some users are treating multi-cloud like they did with multi-OS. Choosing different clouds for different apps based on bit or preference.

For example, most workloads in AWS run in Linux while most in Azure, it's Windows.

@fdmts Could be acceptable... NSX plus Linked vCenters will allow you to use multiple cloud locations as one.

service providers, microservices, hybrid cloud, and IoT applications to name a few

There's not enough CPU,mem,storage in a single system to solve just about any problem

Unpredictable yet fast infrastructure growth along with decreasing administrative time per unit of things to be managed.

Many applications are now being born "cloud-native." That brings with it an assumption of horizontal scale and other architectural properties.

humblebrag: Wikibon nailed the #TruePrivateCloud market analysis cloud is in high demand onprem

Also never know how fast an application might grow from the outset

dev ops

data growth ... ridiculous exponential data growth driven by digital transformation, cloud, IoT, etc

As a macro trend, data analytics / Machine Learning / AI / Deep learning / whatever you want to call it ... it's here and it demands massive scaling of data coupled with CPU capacity.

Data growth! http://inside.igneou...

There's also a whole set of scale-out technologies pioneered at FaceBook/Google/etc. that are now looking for for problems to solve + monetize in the mid-market and enterprise.

so I think partly of it as being driven by customer demand but also partly driven by tech searching for a problem in other markets

@steve_pao Agreed... Businesses are generating a ton of new data and are afraid to delete any of the old. All are looking to a world where they might get answers from big data one day.

Virtualization, I believe, actually stagnated innovation for about a decade. This has been needed for a while.

Data explosion demands new architectures. 6 Zetabytes today, growing to 90+ Zetabytes in 2025.

Another trend is the shift of IT capacity planning from OpEx to Capex. Business leaders demand agility and to pay only for what they use.

- #TruePrivateCloud is a departure from "private cloud" claims of legacy providers. https://cube365.net/...

Andrew: @andriven what problems are being solved by these new architectures

the need to independently scale resources will always be there, any time you purchase resources that aren't absolutely required you're missing an opportunity for cost optimization

people (mobile) and data (sensors) at the edge

One driver is the flexibility that comes with knowing you can grow as needed with little or no disruption. You don't have to size it correctly the first time.

Another driver is the amount of data being stored. Legacy architecture is not capable of storing this much data in a useful way. Users want to do more than pump and dump data into some repository.

@fdmts So business leaders want to take the risk out of decisions they have made?

A lot - I'd argue hyperconvergence is possible due to scale-out tech. So is containers/Docker/Kubernetes. Obviously what I do at my day job is made possible there too.