Private cloud discredited, part 1

Private cloud discredited, part 1

Summary: Microsoft, a most unlikely ally, has published a white paper that shows private cloud solutions are up to 40 times *less* cost-effective than public cloud alternatives for many companies. It vindicates my prediction that private cloud will be discredited by year-end.

SHARE:

Back in January, I made a controversial prediction that private clouds will be discredited by year end. Now, in the eleventh month of the year, the cavalry has arrived to support my prediction, in the form of a white paper published by a most unlikely ally, Microsoft.

Titled simply The Economics of the Cloud (PDF), the document succinctly sets out the economic factors that make the public cloud model an inexorable inevitability, substantiating my long-held views. It deserves a full reading — don't settle for the overview in the authors' blog post announcing it. Here are some headline numbers that should give pause for thought:

  • 80% lower TCO. The combination of large-scale operations, demand pooling and multi-tenancy create enormous economies in public cloud data centers: "a 100,000-server datacenter has an 80% lower total cost of ownership (TCO) compared to a 1,000-server datacenter."
  • 40-fold cost reduction for SMBs. "For organizations with a very small installed base of servers (<100), private clouds are prohibitively expensive compared to public cloud."
  • 10-fold cost reduction for larger enterprises. "For large agencies with an installed base of approximately 1,000 servers, private clouds are feasible but come with a significant cost premium of about 10 times the cost of a public cloud for the same unit of service."

Since I know there's a subset of ZDNet readers who will leap into Talkback to cry wolf on cloud security without bothering to read either the rest of this blog post or even looking at the white paper, here's what it has to say on that particular canard:

"Large commercial cloud providers are often better able to bring deep expertise to bear on this problem than a typical corporate IT department, thus actually making cloud systems more secure and reliable ... Many security experts argue there are no fundamental reasons why public clouds would be less secure; in fact, they are likely to become more secure than on premises due to the intense scrutiny providers must place on security and the deep level of expertise they are developing."

That paragraph alludes to one of the key factors that I've been highlighting in my recent evangelism of the cloud model, the economies of scale for collective scrutiny and innovation. Amazingly, the document reaches its conclusions without adding in the additional economic benefits of this factor, which surely must deliver a knock-out blow to the private cloud concept. The collective feedback and testing from a diversified customer base enhances not only the security of a public cloud infrastructure but also informs and directs its evolution at a far more rapid pace than any private cloud will allow. There's a virtuous cycle here, of course, in that public clouds are already more cost-effective as platforms for innovation, so that there is going to more innovation happening here than on private clouds anyway. That innovation will help to further accelerate the evolution of public clouds, thus amplifying their economic advantage more rapidly and to a greater extent than even Microsoft's strategy team have envisaged.

The document ends by trumpeting Microsoft's own commitment to the cloud, and in particular its work with products such as Office 365, Bing and Azure, which is fair enough. But I have to mention that it also exposes a huge, gaping hole in Microsoft's product line in the SMB market. As I highlighted earlier, the document plainly implies that any SMB would be crazy to factor on-premise or private cloud into their future strategy when these options are going to cost up to 40x more than public cloud alternatives. That finding casts a dark pall over Microsoft Dynamics, a product line that has spent the past decade running away from the cloud (that may sound like a strong verdict, but believe me, I've witnessed the whole, sorry saga). If Microsoft really wants to be 'all-in the cloud' as Steve Ballmer has been saying, then it's going to have to figure out a way to get Dynamics there before its SMB installed base starts deserting it in droves. Acquisition of an existing cloud solution is probably its only viable option.

Some readers may be wondering why I wrote 'part 1' in the title of this post. It's because I fully expect to be able to write a 'part 2' before long. The economic case to discredit private cloud has now been made. The other shoe still waiting to drop is a complete rebuttal of all the arguments over security, reliability and control that are made to justify private cloud initiatives. The dreadful fragility and brittleness of the private cloud model has yet to be fully exposed, although I have done my best to alert people of its existence. I am confident that evidence will surface soon and then we will be able to drive the remaining nails into the coffin of private cloud.

Topics: Cloud, Banking, CXO, Data Centers, Hardware, Security, Storage

Phil Wainewright

About Phil Wainewright

Since 1998, Phil Wainewright has been a thought leader in cloud computing as a blogger, analyst and consultant.

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

46 comments
Log in or register to join the discussion
  • Latency

    Neither the article nor the whitepaper mention user interface latency problem for the public cloud applications. If the data sources are far away, the speed of light sets a performance wall that technology cannot work around. Human productivity is highly sensitive to responsiveness of the application, well beyond the raw time difference in latencies (i.e. human brain, attention, focus, flow,... switch to the slower clocks). Hence, that aspect needs to be factored into the calculations of the comparisons of economies.
    zd-nightlight
    • Well, given that with a public cloud, for most cases, you would be more

      likely to have servers closer to users, you argument look pretty ridiculous. Of course for the specific (and limited) case of all in house users located in one location, you might get a slight advantage on latency. But, seems like as the objections get knocked down one by one, the arguments get more desperate all the time.<br><br>The real thing, is those receiving the salaries that results in the 10x to 40x higher costs do not like this one little bit.
      DonnieBoy
      • RE: Private cloud discredited, part 1

        The assumption "...likely to have servers closer to users" is a wishful thinking. A large data center may have cloud clients around the country, even around world, with roundtrip latencies of public internet that can fluctuate into hundreds of milliseconds or even seconds. The operator productivity (human is the most expensive "component") would drop dramatically as the latencies increase or become erratic. Consider for example that VOIP becomes unusable if the latencies become greater than ~200ms. Humans are very sensitive to latencies especially when response time is erratic (check how much efforts google puts into reducing their latencies).

        That's why the previous incarnations of the 'public cloud' idea, the so-called "thin clients" (and before that 'smart terminals') had lost to PCs with local apps in corporate environments, even though the same economies of scale touted here applied equally well in those cases. Unless you add all the costs, including human productivity, you will only end up with wishful conjectures.
        zd-nightlight
      • Come on, thin client has nothing to do with private cloud vs public cloud.

        The thin client choice is completely separate from where to put your servers. And, big providers have multiple servers around the world, and connect clients to the closest one. They can do it much better than with a private cloud. You are desperate to find reasons for private clouds.
        DonnieBoy
      • Thin client failure is relevant data point

        The relevance of the thin client failure is that it demonstrates:

        a) The same type of economy of scale applies as the public cloud (the same goes to mainframe+smart terminal approach)

        b) They failed because of the latency problem (making human operators less productive)

        c) Unlike public cloud, thin clients didn't have privacy issues (they were usually hosted in the same company), hence the public cloud has more downsides (higher latency, privacy)
        zd-nightlight
      • RE: Private cloud discredited, part 1

        @DonnieBoy And going all in on self-serving bullshit white papers by public cloud vendors is, on the face of it, pretty ridiculous. With public clouds, you're still sharing infrastructure with the rest of the world. With public clouds, you give every script kiddy and professional Russian hacker the challenge of breaking your security. Given Microsoft's record on security, not to mention the Web's record, it's not a bet most company's are willing to place. With HIPAA, European Union, Commonwealth of Massachusetts weighing in on handling of Personally Identifiable Information (PII), and the number of hacks of late, companies that have market-differentiating IP or regulatory burdens, or even reputations that can be seriously damaged by ANY public disclosure of information violations (hacks, thefts, etc.), the ones who make those salaries you talk so blithely about would be negligent if they didn't weigh the pros and cons of public versus private infrastructure.

        Not to mention that, with private infrastructure, you can tune response time, security, and bandwidth to be precisely what you need it to be, instead of hoping and praying that the public infrastructure you share with billions will be reliable, available, scalable and manageable enough to meet your business needs.

        SMBs, sure. Other than that, it isn't just the cost of the service that you are weighing. It's your reputation, your viability, your very existence that you are placing this bet on.

        It's too bad that so many people have such a narrow view of the realities of business and even IT itself, and are so ill-equipped to make those life and death decisions. But if everybody could do it, then we wouldn't be having this debate, would we? And we wouldn't need the suits with the big salaries, who often are better equipped to make these decisions that someone sitting in their cubicle and spouting their personal opinions while watching out of the corner of their eye for their boss so they can hit the boss-key.
        chinese.bookie
    • RE: Private cloud discredited, part 1

      @zd-nightlight - latency is definitely a big issue. the problem can be mitigated somewhat with additional datacenters (and then you need the ability to host a customer's data in multiple datacenters) as well as client-side caching and WAN optimization. and by the way - the speed of light is not a factor. :) latency is related to physical distance but it's not that the electrons can't get there fast enough. i would guess the problem is more often the width of the pipe, inefficient routing, inefficient (or lack of) compression/optimization, not the actual speed of the electrons.
      kayvaan
      • RE: Private cloud discredited, part 1

        Latency IS one of the main issue, but proper use of RIA / Smart Clients / Apps can make this invisible to users and not slow down their work.
        More intelligence at the edge give more rich and responsive user interfaces, consuming less bandwidht and Cloud resources.
        bedin@...
  • TCO is lower for the owner not for the client

    Look at price matrices of the different cloud offering. As for SMB, albeit unmanaged risk for sensitive or crucial data, with no relief possibility in case of failure to meet advertised services, prices are simply so out of scale. Of course you will have to pay for maintaining your server environment, but you can include this is the maintenance package for clients and have a very reasonable price if you have a stable and sensible system...
    s_souche
    • Now that is the most beat-around-the-bush offhand argument I have heard

      Users do NOT see (or know of) any difference between privately hosted applications and those hosted in the cloud, so, you can NOT charge one penny more for privately hotsted applications. YOU, the eager provider get stuck paying for all of your bad decisions!!!!!
      DonnieBoy
      • Did I say that ?

        @DonnieBoy I nevr claimed there was a difference for the user Thought there clearly is ome huge in term of availability but that was not my point). I said there was a legal difference in terms of Liability and responsibility.
        s_souche
      • And, again, most can not afford to have the people 24x7 need to make it

        secure and reliable, and keep from getting sued.
        DonnieBoy
  • RE: Private cloud discredited, part 1

    and if the internet is down, so i my entire office.
    bjs_z
    • For the limited case, that the applications are ONLY for internal users in

      a single location, and you do not need an internet connection for other parts of the application to function correctly, that might be true. But, put your money into two or three different Internet providers, so, that if one fails, you switch to the other. In the end, you are more likely to have down time due the fragility of the private implementation than because the internet is down.
      DonnieBoy
      • RE: Private cloud discredited, part 1

        @DonnieBoy : I see now the advantages of the cloud : instead of managing a lot of separate servers in the enterprise I need now to manage a whole more stack of Internet providers, with different contracts, SLAs and prices. And rewrite my enterprise apps to work in a public cloud scenario (including RIAs, cache etc).

        Compelling indeed. Tell that to a bank with 500+ branches all running legacy time-tested business critical apps and you'll have the "don't call us we'll call you" treatment :-)

        But building a private cloud (with mostly existing hardware) that can run their legacy environment in flexible manner and keeping the existing, working networking infrastructure (also private) it's a whole another story.

        You know, it's about the right tool : choosing the right one for the job at hand. Trying to prematurely proclaim victory in non-existent religious wars (private cloud is dead, only public cloud is the right choice) is immature, at least.

        And the crushing argument ("we've found a Microsoft marketing white paper claiming that own products are the best and we've seen the light") is funny.
        bogdan.farca
    • RE: Private cloud discredited, part 1

      @bjs_z - that assumes a browser-only client scenario. but "cloud" doesn't preclude caching data client-side for offline scenarios.
      kayvaan
    • RE: Private cloud discredited, part 1

      @bjs_z when was the last time your phone line went down? Once upon a time power cuts were commonplace too. Nowadays no-one needs to install a private backup generator...
      vhoong
      • Say what?

        @vhoong Tell that to any medical facility with a patient in the operating room that they do not need a private backup generator....

        Get real....
        linux for me
  • RE: Private cloud discredited, part 1

    As for the Security issue, I did read the white paper. However, some cloud providers are better at security than others and it is difficult to find out which is which. Also so far governments are not to keen on the idea of security on the public cloud. Which means most of the vendors have already put in back doors for departments like Homeland Security. Which also means the back door is available to anyone smart enough to use it.

    Also I am not keen on the idea of completely dumb terminals. Anything that disrupts my local Internet connection has the ability to shut down my business. Unlike other outsourced resources, this state of affairs does not have to exist. Most SMB can't afford to have a billion dollar power plant supplying their electricity but an offline productivity package that syncs back up when the Internet is back is not an unreasonable request.

    There has been little progress on a unified support package. Cloud vendors are pushing clouds and hoping consumers don't "look behind the curtain". Support is expensive and if Google goes down for the entire afternoon, what recourse do their corporate clients have?

    Exit strategies are another pinch point few people have wanted to talk about. Unless cloud computing is somehow a socialistic "can't ever fail and can't ever get sold" business consumers have a valid risk in cloud services. Just because Microsoft doesn't go out of business doesn't mean they won't sell off individual product offerings. What happens to the companies relying on those services. Google decides data storage is really expensive and they have a buyer for their data, what happens to the contracts between companies and Google? What if the company purchasing the data storage services of Google doesn't have the same TOS?

    Just recently Ask.com dumped their blog services. At the last minute another company went in an took it over but still what are the TOS for the new company? If you had an Ask.com blog how does it affect you? Do you have any choices in the matter? Do you have to resort to lawsuits?
    mr1972
    • Wow, great way to pretend that more secure is not more secure. Also, quit

      it with this "if the internet goes down, my business goes down". Get a more reliable Internet connection, or pay TWO providers so you can switch from one to the other in an outage. Finally, most of the applications we are talking about running in the cloud are CUSTOMER facing, and YOUR internet connection is much more likely to go down than your providers internet connection, so this argument actually works against you in most cases.

      And, this is NOT about dumb terminals, you can run the application that connects to the cloud as a locally installed application, or as a web application, regardless of whether you are using private or public cloud. AJAX / HTML5 means that maintaining locally installed applications is getting more stupid every day, especially if you customers are the main people using the application.
      DonnieBoy