The Non-Future of Multi-Tenant SaaS CMS

By on June 15, 2012

What’s the future of the traditional SaaS CMS market?  More fundamentally, what does SaaS even mean in relation to CMS anymore?  I’m personally convinced that this model is dead.  Here’s why –

(SaaS = “Software as a Service.”  You rent your CMS and infrastructure from another company.  “CMS in a Cloud,” if you will.)

Traditionally, SaaS has meant two things — one explicit, one implied.

  1. Your system is hosted by someone else who is an expert in managing that system.
  2. Your system is multi-tenant.  You are part of a larger group of organizations using the same system.
I’m going to argue that these two distinctions have gotten really blurry, and saying “we’re a SaaS vendor” doesn’t mean much anymore.  At least, it’s not nearly the differentiator that it was.  The fact is, most CMS vendors can “do SaaS” today, to the same extent that any vendor has ever done SaaS.

First, let’s consider hosting –

Having your stuff hosted by someone else is an advantage to a lot of organizations, and even more so when you consider the warm fuzzy of being hosted by the vendor itself.  However, is this still the domain of traditional SaaS vendors?  Not really.

For Blend and other firms like us, virtualization changed the game considerably, and then Amazon changed it even more with EC2.  Setting up a new environment just means unpacking a pre-existing image.  With EC2, we don’t even have to set-up the hardware.

This has obliterated the hosting advantage of the original SaaS vendors.  Consider this: Blend can give you a dedicated EPiServer installation in minutes.  Ask me for one, then go get a cup of coffee. I’ll have a username and password in your inbox before you get back.

Is this “SaaS”?  No, it’s a virtualized EPiServer install that’s been pre-configured, optimized, then freeze-dried to an Amazon AMI. Blend does the same thing with eZ publish, and you could conceivably do it with any system.  Adding load-balanced instances is fairly simple as well.  You can automatically scale out to multiple instances behind a common load-balancer.

Now, this isn’t new.  People have been doing this for a couple years now.  But, combine pre-configured instances with automatic load-balancing, then combine that with content management, and you’re blurring the line between installed and SaaS considerably.

And if you want vendor expertise, most installed CMS vendors are doing their own hosting these days.  EPiServer has their EverWeb offering, and most other vendors have similar services.

Second, let’s talk about multi-tenancy –

SaaS systems are usually designed to support multiple organizations running completely independent web properties on the same system.  The only reason you see your stuff is because it’s tied to your user account.  You appear to be alone, when you’re actually in a crowded room.

As to the benefit of this, I say…well, nothing, really.  What benefit does this situation actually provide you, the customer? I’m sure there are multiple benefits for the company providing the system — namely that they have fewer systems to upgrade and maintain — but how does that directly benefit you?

It really doesn’t.  There are a couple indirect benefits, sure — since they have one system to maintain, they theoretically offer more stability.  But, this works both ways — some other organization doing something stupid on a CPU you share can have adverse effects for you.

Another cited benefit of multi-tenancy is that the system is being continually enhanced, and you’ll automatically get new capabilities when they’re released.  A valid point, but I maintain that this advantage is negated by the fact that SaaS vendors are always going to limit what you’re allowed to do with your slice of their system.  You are a slave, in effect, to their environment and functionaity.  I’m sure there’s some limited customization, but large-scale integration and development isn’t going to be supported, so if you want new functinality, you have to sit around and wait for it and then hope it’s what you want.

To counter this, some SaaS vendors will set you up with a custom instance, to which you can “color outside the lines,” and do whatever you want.  Wonderful…but now you have a dedicated instance, and how is this different than the simple hosted instance we discussed previously?

And, with that, we’ve come full circle. “SaaS” as a concept is quickly becoming a synonym for “hosted and managed by someone else,” and multi-tenancy is receding as a benefit worth considering.

Most CMS vendors are now offering their version of SaaS.  They’re hosting instances of their own systems, managed by their own technicians, and supported by their own help desk backed by the product engineering team. It’s not multi-tenant, but in a true multi-tenancy situation, that fact was hidden from you anyway, and, again, provides no actual benefit.

(I’d be remiss if I didn’t mention a huge benefit to this set-up.  If you don’t like SaaS, you can almost certainly package up your instance, put it on one of your own servers, and therefore go from SaaS to installed.  Try that with multi-tenancy.)

Licensing gets interesting in these set-ups.  You can purchase a license and bring it, or most vendors will let you either “rent” a license, or amortize the purchase cost of a license over the term of the contract which results in you owning it in the end.  (So how is this different than just having someone host your CMS implementation?  It’s not.  That’s kind of the point.)

Real Story Group talked about this back in February, with a rogues gallery of SaaS vendor updates.

The Real Story Group found that Software-as-a-Service vendors have failed to gain the significant market share that many had predicted over the past decade:
  • Longtime player Clickability was sold last year to Limelight Networks for less than its previous venture investment intake
  • CrownPeak has lost momentum amid struggles to modernize its platform
  • OmniUpdate has continued to grow slowly, but primarily within its core, higher-education customer base
  • None of the major North American SaaS WCXM vendors have managed to gain a foothold in other regions
Read the entire release for some more caveats on running traditionally installed systems in the cloud. (They do make a good point about licensing terms.  Many vendors do not have “true” SaaS pricing — they amortize an installed license over the term of the hosting contract.  This has advantages and disadvantages.)

For smaller implementations, SaaS still works well. I remain a big fan of Squarespace.  Additionally, you can do some neat things on Drupal Gardens and WordPress.com.

But for larger, enterprise installations, a good hosting and support environment backed by a solid SLA is what you really need.  With these in place, pick your system — you can make a SaaS solution out of about anything these days.

###

What Links Here

Comments

  1. After reading your two main points services like WordPress.com don't truly fall into either category. For instance, the shared issue:
    some other organization doing something stupid on a CPU you share
    Doesn't apply because sites on WordPress.com are hosted on hundreds of servers spread out across multiple data centers. If one of those servers runs into something odd and ends up eating all of the available CPU then there are still hundreds of other servers available to repsond to future requests. That also brings up the point of how load balanced would most vendors really take a site that they spun on EC2? Would they be willing to take it to the point of spanning multiple DCs (in the case of EC2 say east and west coast for the US)? If an entire DC goes offline (which happens) will the site still be up? Certainly it would be possible for it to still be up in a ready only mode. Your description of multitenant sounds akin to shared hosting on Dreamhost. In some cases that may be true, but not in all.
  2. Doesn't apply because sites on WordPress.com are hosted on hundreds of servers spread out across multiple data centers. If one of those servers runs into something odd and ends up eating all of the available CPU then there are still hundreds of other servers available to repsond to future requests. That also brings up the point of how load balanced would most vendors really take a site that they spun on EC2? - See more at: http://gadgetopia.com/post/7581#sthash.ITUIkNwG.dpuf

  3. owais93 says:

    If an entire DC goes offline (which happens) will the site still be up? Certainly it would be possible for it to still be up in a ready only mode. Your description Revision Stockholm of multitenant sounds akin to shared hosting on Dreamhost. In some cases that may be true, but not in all

Comments are closed. If you have something you really want to say, email editors@gadgetopia.com and we'll get it added for you.