1/ What about biz that want on-prem architecture—what is role of major cloud providers? About platform transitions. fortune.com/2016/09/21/ama…
33 replies and sub-replies as of Sep 26 2016

2/ Every tech transition creates desire for bridge. Use new tech for existing apps, infra. Feels good short term but long term pain. Always.
3/ By defn "cloud" 10x-100x Data Center architecture PLUS massive learning/expertise of sysmgmt, security, scale. Cloud doesn't scale down.
4/ No real cloud implementations are architected like (or feature-similar to) what's been done on-prem. Just different. That's the point.
5/ Cloud also stack of tools/methods at high level can be applied sans cloud, but resulting apps different, eg containers or virtualization.
6/ No shame (none!) in building and improving existing apps/code using tools/tech as they were built. It is efficient and robust.
7/ No shame in integrating w/existing code via separate apps built in new ways via APIs. Consumer Web built via API calls to mainframes.
8/ But if you're building new in new ways, a big mistake is porting code built an old way. You create technical debt to fight from start.
9/ Code is not code. Code is written, intentionally or not, with assumptions for how underlying architecture (data, OS, net) behave.
10/ Mainframe->PC, Character-> GUI, Client/Server->Web, PC->Mobile and more keep showing this same lesson. Best code and approach is new.
11/ Most wanting private cloud should be more about API to existing systems PLUS public cloud, or just ongoing work with existing platform.
12/ When platform shifts happen, best bet to make is "all in" PLUS "respect the existing investment, but keep it distinct". // EOTS
Posit: on-prem infrastructure will sometime soon be akin to running your own bank vaults. Hybrid = a distraction.
You posit-ed the present.
Well, it's more a statement on the mainstream-ness, rather than the truthfulness. Former always lags the latter.
As an enterprise sw vendor, we’re preparing for a hybrid world. For us, it’s about where company data lives.
On-prem investments will have long tail and the data there often will not be forklifted into cloud. Over time, as more tools are...
I get the argument. But what will happen is that as new apps built, that data becomes less critical. Eg Web -> file servers.
As new biz functions go native cloud (CRM or ecomm perfect examples), that’s true. But that’s a long adoption cycle.
But when will banking, insurance, healthcare data be 100% hosted? And legacy data? Hoo boy, I dunno...
Mainframe long tail is great ex. Stand up new apps calling mainframes but don't port mainframe code to client/server.
We’re a 40-year old data integration and analytics company, tho, so our focus is admittedly somewhat biased
… adopted native cloud, their customer data will be kept in cloud. Data agents are the future for these orgs. Tricky stuff.
1000% agree. Wrote this today:
link to this article ?@stevesi
BTW - It’s on-premises…not on-premise Ben :-)
Funny thing I caught it when he sent it to me but been so busy forgot until the screenshot..
I stand by on premise. It's a term of art, not a descriptor. After all, a lot of private servers aren't on premises anyways
So you mean “by reason of“?
For sure last version of on-prem Server should have been ~2012 (same with SQL, Exchange, etc). Using that code holds all back.
In abstract yes, but then IRL revenue crash... versus slower transition. Many apps cannot move. TCO doesn't work, yet
There's a long tail since all of these are effectively "SaaS".