Single tenancy, the DEC Rainbow of SaaS

Single tenancy, the DEC Rainbow of SaaS

Summary: If you believe the private cloud vendor spin that simply by including certain features in common you'll capture all the advantages of cloud computing, you're making the same mistake that enterprises made in the 1980s when they invested in DEC Rainbow PCs and IBM PS/2 Microchannel.


Amidst all the content that spewed out of SaaS ERP vendor Workday's Tech Summit this week (disclosure: Workday is a current client, although I was not able to attend its event in person), there was one blog post that really set my teeth on edge. Fellow Enterprise Irregular Josh Greenbaum and I rarely see eye-to-eye on SaaS and his assertion that multi-tenancy is a vendor, not a customer, issue — a supporting tweet called it the 'Intel Inside' of SaaS — elicited this counter-tweet from me: "Yep, & single tenancy's the DEC Rainbow of SaaS." Knowing that a single tweet doesn't carry the same weight as a blog post, I decided to elaborate on the analogy as soon as I had the opportunity, so here we are.

The allusion to the DEC Rainbow will be somewhat obscure to those of you reading this who are younger than the age of, oh, 48 or so. Therefore you'll need to bear with me for a short history lesson. My first job in the computer industry started in November 1985 when I joined one of the UK's leading microcomputer resellers eighteen months after the UK launch of the IBM PC. The dominance of what was then known as the IBM architecture (it was only much later that people realized it was really the Wintel architecture and not IBM's at all) was just starting to become clear and one of my first jobs was to help clear the company's remaining stock of Apple Macs. I had heard of Apple, but there were other, even more mysterious pages in the reseller's printed catalog (this was in the days before the Internet, remember). There was the DG/One laptop and a double-page spread devoted to the DEC Rainbow. What I found curious about this machine was that it had an Intel processor and MS-DOS and it would run all the same leading business software titles as the IBM PC, but you still had to buy DEC Rainbow-specific versions of those titles.

What I later learnt was that DEC, at the time the number two computing vendor after IBM, had taken an approach to the microcomputer revolution that I've since seen repeated in many later generations of technology. The company looked at a new emerging trend and latched on to the elements of it that made sense from the company's own perspective. It saw a new commodity platform emerging and thought it could do a better job of offering the same capabilities with its own proprietary technology (funnily enough, IBM later on attempted the same approach when it tried to divert the PC architecture into a proprietary fork called Microchannel. That effort failed too).

Now we're seeing history repeating itself once again. This time the existing generation of established software vendors look at the new commodity platform of cloud computing and they say to themselves, 'I like this way of doing things, but I can do it better if I do it in my own proprietary way.' So they come out with their single-tenancy, private cloud implementations with all the usual proprietary lock-in that they're used to building into their business models, and they completely miss out on the wider context that differentiates the cloud from earlier models. Customers buy into what they're offering because they trust their existing vendors and they're suspicious of these upstart newcomers that have no established track record. They believe the vendor spin that simply by virtue of including certain features in common (Intel 8086 processors and MS-DOS in the case of the DEC Rainbow, virtualization and automated provisioning in the case of cloud computing), they'll capture all the advantages of the new model. But in truth, they simply don't get it.

So let me spell it out. Multi-tenancy is not the 'Intel inside' of SaaS and cloud computing. Actually 'Intel inside' was a marketing campaign that Intel starting running when it realized that the PC architecture was bigger even than its own microprocessor architecture, and that rivals like AMD had an equally legitimate claim to compete for a slice of the same market. In exactly the same way, multi-tenancy is bigger than any single vendor because it is one of the most defining characteristics of cloud computing.

To say that multi-tenancy is only of interest to vendors and has no relevance to customers is a bit like saying Wintel compatibility is only of interest to PC manufacturers, and so customers should not worry about it. Try telling that to the enterprises that invested and wasted millions of dollars in rolling out DEC Rainbows or IBM PS/2 Microchannel machines in the mid- and late 1980s. If a vendor is selling you a proprietary dead-end that will be obsolete before its time, I'd say that's a factor of huge importance to customers. Don't let anyone tell you any different.

UPDATE [added 03:07am Aug 27th]: Josh has responded to this post, and I've added a comment in reply to his post. Josh raises the intriguing notion that it's possible to have SaaS compatibility without the "SaaS dogma." My point about the DEC Rainbow however was precisely that if you do not understand what the dogma is all about then your claims of compatibility are destined to be ill-founded.

Topics: Data Centers, Cloud, Emerging Tech, Hardware, IBM, Intel, Servers, Virtualization

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.


Log in or register to join the discussion
  • RE: Single tenancy, the DEC Rainbow of SaaS

    Great point Phil. It's also important to note why multi-tenancy matters. From my POV, everybody focused on the economic benefits (shared compute infrastructure) - which is for sure important. But equally important is that all customers are actually running the identical version of the software. In a "private cloud" model, a vendor could theoretically deliver updates to thousands or millions of so-called "private clouds" but that would only work if each and every customer were running identical hardware/software stacks, and each provided the vendor with unfettered access to upgrade the software as they see fit. Not realistic to say the least. And even so, the care and feeding of those systems would still be ridiculously inefficient. And for what gain?
    Dave Girouard
  • RE: Single tenancy, the DEC Rainbow of SaaS

    I'd also like to say that proprietary in the context of dead-end can also happen on open source platforms.... in fact the chance of ending up with a bespoke instance seem to be even higher.

    Multi-tenancy is a given in cloud, vendor longevity is probably more important
  • RE: Single tenancy, the DEC Rainbow of SaaS

    but SaaS/Cloud is also more than just multi-tenancy, isn't it? ideally? you could have a multi-tenant architecture that's still proprietary in some ways, no? so, phil - what would you say are the criteria that define a true "cloud" service? does it have to be multi-tenant AND accesible via an internet protocol?
    • Cloud immersion is absolutely critical

      Good point, @kayvaan: does it have to be multi-tenant AND accesible via an internet protocol?

      As I always say, you can't take the computing out of the cloud and still call it cloud computing. Multi-tenancy only makes sense if the instance is being exposed to a cross-section of customers accessing it over the Internet and constantly evolving in response to their needs, including interaction with other Internet resources.
      phil wainewright
  • Customers still have to swallow the "Trust Me" pill

    Customers have to trust the cloud company's sysadmins. There is no way to encrypt your data such that sysadmins cannot access it. Using Public/Private key pairs for encryption requires a clear text file SOMEWHERE on a computer - and it can be read by a human.
    Roger Ramjet
  • How about the customer point of view


    I don't understand this series of posts/tweets...
    If a SaaS vendor (I'm talking public cloud here) comes-up with a solution to provide a competitive alternative to another known multi-tenant vendor but with a single tenant architecture does it really matter? I mean if the price is right, if the data protection is right, if the elasticity is there why would you care how they manage it (server automation or whatever really)?

    This was the reason I agreed with Josh post in the first instance but I may be missing something in the debate.

    Please enlighten me :)

    Fabrice Cathala
    • Re: How about the customer point of view

      @Fabrice Cathala
      The point is that a multi-tenant environment constantly continues to evolve in response to the needs of its customers whereas a single-tenant environment remains static until its customer decides it's time for an upgrade.

      Now I know that single-tenancy advocates are keen to assert the customer's right to stand still and not benefit from continuous evolution of a shared multi-tenant platform, but is that really in the customer's interest?

      See Dave Girouard's comment at the beginning of the thread for more on this.
      phil wainewright
  • RE: Single tenancy, the DEC Rainbow of SaaS

    (re-post from Josh Greenbaum's blog) A fascinating display of ping-pong! I must admit that early on I was much of a multi-tenant purist - the mathematics is too tantalizing. And if you're just starting out developing a cool app - that's definitely the way to go (that's what we did!). Phil often states the advantages of multi-tenant architectures, and no one can argue with them, it's better all around. But unfortunately, business is not that clean cut. The purist approach of "multi-tenant or die" appears to ignore the impact of the transition for an existing ISV with all its legacy code, infrastructure, and revenue streams to deal with. And let's not forget shareholders and the BoD! There needs to be a middle ground that can provide some of the advantages of the Cloud without re-writing a company's balance sheet overnight. Keep in mind that for a consumer to gain value from a vendors software, that vendor still needs to exist and be financially viable.
    Never has a dogmatic or purist approach to any new technology been relevant over the long haul. (Ask all those folks who thought the 68000 architecture running OS-9 was the "better solution" was! But who cared? Wintel won.) Look over your teller's shoulder at Bank of America and you'll see a DOS client running virtualized on WindowsXP connected to a mainframe...and it's 2010! Is that an optimal solution? Traditional ISV's are vulnerable to upstart SaaS vendors, and a single-tenant approach is a way to strengthen their position and provide added value to their customers. Will it immunize them against these competitors? No. Is it the ideal solution for everyone? No. Will it work? Yes. And that's all we really care about. Having said all that, it's only through these debates that the overall course that we're on as an industry is improved and enhanced. Let the ping-pong continue!
    • RE: Single tenancy, the DEC Rainbow of SaaS

      @sfojames@... I agree, you have to be pragmatic. But realize that going to a virtualized single-instance is a good stopgap. Don't believe the rhetoric from certain quarters that tries to persuade you it's an end destination.
      phil wainewright
      • RE: Single tenancy, the DEC Rainbow of SaaS

        @phil wainewright The only real difference between the two positions may only be timing. I tend to believe that over the long term we'll see an oscillation as different architectures and solutions provide meaningful dislocations to different market segments.
  • Not the DEC Rainbow I Used!

    In the early 1980s, I inherited an ex-management DEC Rainbow. It was rather different from the description in this article and also more interesting.

    Previously, there had been many business programs which ran on the CP/M operating system; lots of "IT experts" claimed that the new-fangled MS DOS would fail because of the lack of useful business program to run on it.

    So, my DEC Rainbow had two different CPUs in the same case: Intel to run MS DOS programs; Zilog Z80 to run CP/M programs.

    Then, Lotus launched 1 2 3 for MS DOS...

    Perhaps DEC, and the people who bought Rainbows, had made the mistake of believing IT Experts?

    At that time there were lots of compatibility problems with non IBM MS DOS PCs; the Rainbow was not alone.

    Also at about that time, our new CEO chose a new business program for his managers: This ran only on Apple Macs so he bought a dozen Apple Macs purely to run that one project management program.

    Good IT is really about satisfying a business need.
  • RE: Single tenancy, the DEC Rainbow of SaaS

    Phil you are right !

    Multi-tenancy is very much a user issue also because even the smallest can be shure the get the same service (SLA) as an enterprise. Moreover a user should love getting the best price / service level knowing that the provider is still making money and he will not go out of business but will grow the service range and appeal to the users.


    P.S. I have a DEC Rainbow (model 100) in my small collection !
  • RE: Single tenancy, the DEC Rainbow of SaaS

    The magic of multi-tenancy is at the end of the day all about scale and complexity reduction. By eliminating so many variables associated with individual single tenant deployments - (you are eliminating the myriad combinations of hardware stacks, OS and infrastructure stacks, security, operations policies and people, etc) - you get vastly reduced complexity, improved cost effectiveness, and operational excellence for both the vendor and the client. <br><br>The key insight is you are literally taking N^N complexity of running software systems and reducing it down to a handful of variables by simplifying and standardizing everything to do with building, running and operating software. As always, this simplification, just like an industrial revolution, means that the output is more reliable, more secure and more cost effective - better in every way. <br><br>I remember at Oracle in the 1990's when we had something like 100 different versions of the Oracle database available for the various operating systems - our porting layer became as much a core competency as the database code itself - and we had literally tens of thousands of different versions and ports of the product out in the market. Now add thousands of different versions of Siebel, PeopleSoft, SAP and middleware running on top, with infinite customizations running on an unknowable combination of hardware it an unmanageable variety of data centers and all being run by IT people with literally no standards or commonality or even consistent skills and training in what they do. <br><br>It's almost remarkable anything worked at all. <br><br>At the end of the day, I think a good way to frame the single tenancy vs. multi-tenancy debate is to consider the differences between cottage manufacturing and industrial production. Clients benefit greatly from consistency, quality, scale and operational excellence.
  • Multi-Tenancy AND Single (logical) instance

    These are all great comments on the importance of multi-tenancy. I wholeheartedly agree with its importance. I want to add a further distinction with the importance of having a single, logical instance. I think this is assumed by all the posters here, but one of the larger ERP vendors is claiming that their newest version, 9, is multi-tenant and can be hosted by the customer, the vendor or a third-party. My question is what good is multi-tenancy if you have isolated instances out in the world?