eWeekChat

A9: Do not go multi-cloud because someone else is doing it. Know your workloads well, your need to be multi-cloud and plan/ design for it. Public cloud is like a Swiss army knife that can help you – but the snowflakes and cost overruns can harm if you don’t know how to use it.

(1) Apps and data rule the choices today - but infra & shared services make those choices unbearable at-times... consciously move those security and networking choices to the fore-front, and not "after" the fact

A9: your choices are not longer just limited to x86, even in a multi-cloud world. arm-based processors like Ampere are pervasive across all clouds today! driving the needed performance efficiency

Q9: your dollars give you power, ask your top cloud service providers to open their platforms more with secure APIs to make your lives easier managing your workloads, ecosystem and federation is a good thing

A9: If a workload isn’t running in your #datacenter it doesn’t mean you can abdicate responsibility. You’re still responsible for your application performance & #customerexperience – and need to invest in technology to manage #cloud applications and optimize resources and costs.

A9: Multi-cloud is not the same as the traditional "dual vendor" strategy, and I would urge against choosing that simply as a defensive move. It should be about more than that--otherwise you are missing the true value

(2) Careabouts are different at each stage. What you look at during the first 20 workloads - very different when you are operating at 100 workloads, and very different at 500+. Cost, latency, performance, operational choices, skill-set gap are all quite real. Hard to fix after

Q9: Just like all of us, the threat actors are taking their business to the cloud, invest in a security platform with deep hooks into your cloud apps and especially your data, side doors and back doors are hard to lock down

A9 Understand your present and future multi-cloud use case(s). Put in the work early to design an architecture that supports and encompasses the ability to deliver desired use case(s) across multiple environment.

A9: Multi-Cloud will not be same for any two companies. Choose wisely, start small based on your business and users's needs now. If Security , Speed, and personalization are the needs. start with edge. For large volume data and compute hosting, choose a provider that offers that

@bernardgolden Because designing in requirements is far, far easier than retrofitting them post-launch.

A9: ask the hard questions about efficiency and sustainability. not all infra is created equal and not all "green" power is the same. find out the footprint of what you're actually running on

@jwittich ARM is totally awesome, and I expect ARM-based apps to become an increasingly large percentage of application portfolios. The price-performance is compelling.

A9: As a buyer, build clear Opex model , and the value you believe you are looking at from Cloud, ask your teams for some comparisons on cloud efficiency targets. Invest in tooling to track those business metrics

A9 we haven't really talked about data/app sovereignty in this chat, but this issue will be more prevalent and drive more multi-cloud use, including use of non-big 3 providers.

@bernardgolden Avoid (sometimes ego-driven) DIY efforts to build your own layers across multiple clouds. With the vendor and OSS energy in this area you likely can't hire/retain/dedicate enough engineers to that kind of effort for enough duration to succeed.

A9: For Executives, Organize hackathons with some DevRel teams from your providers to better understand the value and how to use those tools. You as an executive have to spend time. you can not be hands off.

@bernardgolden Agree

An initial cloud cost estimation is usually 10-100x off. Portability is key. We are in the phase of many UNIX operating systems and the Linux equivalent hasn't been created yet. Plan accordingly.

A9: Very important to look at common Identity and Authorization system across all your systems, tooling, data and use that as a way to simplify multi-cloud design

A8: We’ll continue to see growth for some time with specialization as cloud platforms mature. Plus, vendors like BMC are expanding #cloud use to power #SaaS offerings, so there’s room for growth.

A8: There will be a handful of big enterprise software players like IBM RedHat or VMWare who will be “Switzerland” powering multi-cloud solutions across large hyperscalers. Polycloud workloads and Cross cloud network connectivity will be tablestakes as part of a cloud offering.

A8. A lot more diversity - I mean the right cloud for the right workload. Today the app teams are happy to run in the cloud of their choice. Its Infra, operations and common services that poses as an impediment. Some abstractions can help.

A8: sustainability will become a consideration in multi-cloud choices. hosting workloads in clouds with the most efficient infrastructure for the type of compute/storage/network needed. multi-cloud enables those choices and flexibility

Also it will be interesting to see if the hyperscalers will whole heartedly start building common services for other clouds. Seen that in GCP Anthos, Oracle to Azure interconnects etc - but nothing mainstream.

Q8: explosion, more workloads moving, new apps being written for the cloud with different needs...tools for observability, optimization and security will be critical and talent to manage these will be challenge

@GaneshJKRam Poly-cloud - now I have to look that one up :-).

@jwittich Sustainability will be a big part of decision making

@jwittich Resource optimization (and as a result cost optimization)  will also be key to achieve #sustainability

@GaneshJKRam Well, we (VMware) certainly hope so!

@GaneshJKRam @jwittich Sustainability is THE growing concept.

will be interesting to see how tools evolve over time to account for efficiency. providing better granularity into specific infra choices by workload

A8. Predicting 5 years into IT future is always fun :). More and more SaaS no doubt. Any cloud service will be available in all places, almost nothing will be offered in a way that isnt a service--and this includes on-premises.

@codyhosterman Hard to argue with that.

We should visit this thread in a couple of years :-)

A8: I expect multi-cloud will be somewhat easier for end users, but still quite challenging. After all, we're still early in cloud adoption with plenty of challenges there, and multi-cloud presents additional challenges to the base level of cloud adoption challenges

@ramsba - yes - past statements can come back to haunt :)

We are likely to see more specialized multi-cloud providers that offer vertical specific cloud services (data lakes, analytics, AI/ML, ...) while using the combined buying power of entire verticals to lower costs for their customers.

A7: Yes and no. There are still workloads in private #datacenters that are ripe for moving. Most customers are hybrid/multicloud - i.e., they have some applications & infrastructure in data centers that they manage w/ the rest spread across multiple public #cloud platforms.

A7: dont believe the hype :-), data centers will be around for a while, we see many customers requiring hybrid architectures especially in finserv and healthcare

A7: No. There is significant amount of critical infrastructure that continues to run in Data centers. If the workloads are large, simple on design and predictable in terms of elasticity & growth, the economics are not in favor of a cloud migration. Hybrid cloud is a solution.

A7: there won't ever be a one-size-fits-all solution for compute. variety to models will continue to persist. all clouds of some sort, but different models: multi-cloud, single-cloud, hybrid, on-prem, etc.

#HybridisHeretoStay

@GaneshJKRam That attitude seems to be growing.

A7: Multi-cloud includes the datacenter! It is an important part of a MC strategy. Not only often from a cost perspective, but HW innovation and best of breed choice. For companies where their infrastructure stack is part of their differentiation--it will always be critical.

A7. I think any Enterprise that has been around for 5+ years knows hybrid & multi-cloud together are the new normal & are here to stay. Edge is a good example of this, so are the on-premise ones (hospital floor, manufacturing floor, onsite maintenance etc.)

There will be a consolidation of data centers for large enterprises in particular - footprints will come down - bit it is not going away.

@codyhosterman The traditional datacenter though is changing. Introducing the cloud operating model for resources will be the future.

@codyhosterman +1, i view it the same way, its part of the overall MC architecture

@codyhosterman - agree. It is changing - but not fully replaceable yet.

agree as well with @codyhosterman, multi-cloud includes all cloud models. its about the arch not the owner/location/size.

This does bring up an interesting operating model choice. Do I start with cloud operational model and homogenize that to DC, or the other way around. Or are these 2 going to be different altogether.

Multi-Cloud means now many big centralized cloud providers, mini-regionalized clouding providers , Edge Cloud Providers and yes many Data Centers.

@GaneshJKRam Completely agree. The key question isn't will there be private data centers. It's what percentage of overall enterprise infrastructure footprint will they represent. Over (a long period of) time, will be small percentage.

@ramsba Cost model is different. So, which direction you come from Data center to Cloud or Cloud to Data center (we have seen people moving in both directions due to Cost , operational cost and also Speed/Agility being the reasons.

Q7: Data centers are not going away yet. Remember many companies that used to be traditional tech companies, are not becoming Cloud Companies e.g. Agriculture, Logistics, etc. and such companies need specialization and distribution that require them to have their own data centers

@codyhosterman I agree. Data center ownership is changing. Now it is not just traditional IT companies, but high value, regulatory heavy industries transforming and need Data Centers to build what they can't get just in time from cloud provdiers