Note: This is a play written from tweets. I’ve combined them occasionally for flow, and polished the occaionsal rough edge caused by the 140-character Twitter limit.
Dramatis Personae
Roger Strukhoff (me, aka @taudir). Executive Director, Tau Institute and Conference Chair of the upcoming @thingsexpo at Cloud Expo in New York. Based in Illinois and Manila.
Rene Buest (@ReneBuest). Gigaom Research Analyst and Chief Analyst, New Age Disruption. Based in Kiel, Germany.
Tim Crawford (@tcrawford). CIO Strategic Advisor, AVOA. Based in the Los Angeles area.
Krishnan Subramanian (@krishnan). (Director of Open Shift Strategy, Red Hat. Based in Seattle.
Tal Klein (@VirtualTal). VP of Unicorns, Adallom. Based in Silicon Valley.
Rodney Rogers (@rjrogers87). CEO, Virtustream. Company based in Maryland. Rodney based in Miami.
George Reese (@georgereese). Senior Distinguished Engineer and Executive Director of Cloud Computing, Dell. Based in Minneapolis.
Prologue – Renee and I
[It started out innocently enough. I asked a simple question about a geeky tweet. Then all heck broke loose.]
[The adventure starts with a tweet from Rene.]
[enter] Renee: Is the PaaS market as we know it dying?”
[Rene was referring to an article that referenced a report from 451 Group, which stated that PaaS some day “will be consolidated into the two other major cloud models, IaaS and SaaS.”]
[enter] Me: Companies come and go, but PaaS remains, right?
[My mild consternation stems from our use of a PaaS program to create back-end forms in a start-up for which I serve as CIO. For us, there are distinct borders between PaaS and IaaS (which we provision separately from one of the major vendors), and PaaS and SaaS (which we use separately from another major vendor). My thinking is I like the distinction, and would hate to see it fade away.]
[Of course, the far more important question is whether our PaaS platform and company will stand the test of time, be acquired, or whatever. I don’t want to be halfway up this creek, then lose my paddle.]
[But just for fun, I thought I’d pose that theoretical question above: the question of whether PaaS will remain or not. Thus, the play begins. Sit back and enjoy.]
Act I – Tim Joins the Party
[The real fun began almost immediately. Tim Crawford quickly weighed in, chiding me…]
[enter] Tim: No, it evolves into v2 of what we know as IaaS & SaaS today.
Me: (sputtering): (But..but..but..) don’t we need to keep the term (PaaS) alive for developers?
Tim: It will take time to evolve; three to five years.The end-state is one with limited middle-ground from a demand standpoint.
Rene (in seeming agreement): Well, the problem with PaaS is its level of control were IaaS gives even developers more liberty using resources.
Tim (nodding head): PaaS helps developers move up the stack. Great for most as they move along the maturity path.
Rene: But from an enterprise POV it makes sense to setup a standardized PaaS for internal developers to deliver resources quickly.
[Why yes, I thought. Even our modest little enterprise benefits from this approach, if for no other reason than all of our developers are not in the same place.]
Tim: As I mentioned in (an) article, Private PaaS has value for some time (editor’s note: italics mine) to enterprises. Eventually that will change though.
Rene (seemingly concluding things in a very civilized way): Enterprises should offer their developers both IaaS and PaaS to satisfy the individual project needs.
Act II – Krish Arrives
[As with all Internet-based conversations dating back to the original flame wars of the 70s and 80s, it was too good to last. A few seconds after Rene’s apparent book-end to the discussion, Krish weighed in.]
[enter] Krish: I disagree.
[beat] [gasping all around]
Krish: A well-architected PaaS can give the abstraction of PaaS with flexibility of IaaS.
Rene: In detail?
Krish: I have scheduled in mid-February about this. Will share once it is done. Crux is that PaaS=Blackbox for Devs to push apps.It can also offer a certain level of flexibility with an open architecture.
Rene: So, it should basically have the same level of control like an IaaS?
Krish: Not exactly the same level of control as IaaS but flexibility needed for most workloads. There could be some workloads that needs infrastructure level controls. A well-architected PaaS offers most of the flexibility as today’s IaaS with an abstraction for Devs
Rene: Interesting, looking forward to read more on that!
Act III – Tim Re-Joins, Hijinx Ensue
Krish: Such platforms helps keep the roles of Devs and Ops separate while making their collaboration seamless.
Tim: Are you referring to general enterprise, corner cases and over what time period? All of these matter.
Krish: Majority of modern workloads that fit the IaaS plate fit PaaS.
Tim: Disagree. Too general.
Krish: Arguments against PaaS are too general, too. Specific cases may or may not fit. But generally, not different. Some PaaS vendors add restrictions that makes it tough for most workloads to fit in. That’s why I specifically used well-architected platforms where restrictions are almost none
Tim: My comments were specific to timeframe, class of apps, etc. Must appreciate current enterprise challenges.
Krish: If you are talking about legacy workloads, even IaaS is a wrong place to host. I am talking about modern workloads in the context of applicability between IaaS and PaaS.
Tim: FTR PaaS=PaaS=PaaS. Very different implementations and useful applicability scenarios. I’m referring to more than Legacy. But Legacy can’t be ignored. It is a major reality for today’s enterprise.
Krish: You could build specific platforms for specific usecases or use a general abstraction that fits most scenarios.
[enter] Tal: Isn’t that «SaaS» then?
Krish: As I said I am talking specifically about modern distributed apps. If it fits IaaS, most of them also fit PaaS
Tim: If by “modern” you mean “greenfield,” that opens up many options. [Takes phone call.]
[enter] Rodney: That may be true for *most* cloud IaaS platforms.. 😉
Krish: 🙂
[beat]
Krish: (But) yeah. Greenfield. My point is that PaaS helps run most of the greenfield apps. I am only pushing back against the notion that PaaS is limited compared to IaaS. Not necessarily but comes close to SaaS.
Rodney: This is what happens when I troll midway through a conversation..
Act IV – Rene Re-Enters
Rene: So aren’t you talking about something like a convenient IaaS an IaaS+?
Krish: I am saying that for greenfield apps PaaS is good enough. IaaS+ handy for web scale in niche cases.
Act V – George
[enter] George: Wrong
Krish: As good as me saying you are wrong.
George: I trust my opinion over yours 🙂
Krish: Well, it means nothing. There is a problem if you don’t trust yours. [Refers to his blog to make a higher-level point.]
Rene: IaaS is also good for greenfield apps but it’s too complicated today. Therefore IaaS+.
Krish: I never said they’re not. I am saying why complicate life when you can have it simple. There are some workloads where IaaS+ might be needed. Otherwise, an abstraction above.
Tim: Wow! Jump on a quick call and now IaaS is thrown under the bus? Never said that.
[To be continued?…]