So, we've hit a little speed bump in our uber social media journey here, and it has to do with finding a hosting model that works.
Sounds like a little thing, but it's turning out to be a big thing.
And, I'm guessing you'll run into this one as well, so you might want to follow this thread.
Why Hosting Is Important
If you're like us, you've got a decent IT group to work with, but they've got their challenges as well. Most of their focus is on internal applications and productivity support.
For things that are visible outside the firewall, they're leaning towards hosted solutions for a variety of reasons. And I agree with their logic -- they need to focus on what's unique and differentiated to EMC's business model, and let other providers do things that are relatively standard.
As we drew the arc of our social media strategy (start inside, get proficient, venture outside), we knew we'd need to find some service providers that offer us hosting services.
But we've hit a snag.
What Hosting Providers Offer You
The term in the industry is "power, pipe and ping". Basically, they'll give you a dedicated server, make sure it's running, has network connectivity, and back it up for you. The rest is up to you.
If you want high availability, that's two servers, thank you.
What Our Software Vendor Offers Us
Jive, our software vendor, has a model where they license per instance. An "instance" refers to a domain of registered users, any modifications to the code. As an example, if you build 6 different sites, you'll need 6 different instances.
What We Need
We've got about 6-12 projects heading outside the firewall in the near future. Using the current model, that would mean between 12-24 dedicated servers, plus 12 dedicated instances of Clearspace.
I don't want to share pricing with you, but you could buy a nice villa on a tropical island for what that would cost each and every year. It's just not going to happen.
Here's Where It's Flawed
We're not running huge communities in this model. Sure, maybe there are lots of named users, but activity will be infrequent at best -- no more than a few people at any one time using the platform.
From a capacity perspective, I could easily run 6-12 communities on a single server, and not ever suffer any performance problems. We know because we're running a large Clearspace instance here.
If all I need is one server (and maybe a second one for failover), why the heck should I have to buy a fleet of idle servers, consuming money and power and generating heat with no productive use?
The software licensing model is flawed as well. I've done my share of software license pricing in my career, and it's a tricky business. The trick is to align the licensing with the value received by the customer. Any model will work, as long as the customer sees more value when they're paying more money.
We were presented two models: a "named user" model and an "unlimited user" model. The "named user" model doesn't work for many of our external communities, because lots of people will show up to check things out, but few will stay and participate.
The "unlimited user" model is pricey, and assumes that we've got a big, honkin' community going. Well, we won't anytime soon.
The Impact
Imagine you wanted to fly across the country.
You called the airline, and they said "well, we can have a plane for you next week, and it'll cost you $100K, oh by the way, feel free to invite 139 of your closest friends on the trip as well, as there's going to be plenty of room".
Not a lot of people would be flying, right?
The Proposed Solution
Since we're technology infrastructure people here, we do a lot of work with VMware. Take any x64 or x86 server application, run it on VMware, and you can run multiple, independent images that don't know about each other. Logically, it looks like multiple servers, each running their own version of the operating system, database, etc. Physically, it's a single server.
It's new to some people, but there are literally hundreds of thousands of instances of this stuff Just Running with no muss and no fuss. Our internal EMC ONE site runs in a VMware instance -- it just works. And we can use the same server for a variety of other things. So far, no interaction whatsoever.
Now, at some point, we may find we need more server resources, but I'd like to have that problem, because then justification will be much, much easier.
How does that apply here? I'd like to rent a single server (plus a backup), run VMware on it, and then load up 6-12 different communities, each with their own database, operating system, security domain, etc. etc. All on one physical server.
I'm willing to bet my badge that it works well the first time with no issues. Unfortunately, there aren't many hosting providers that are comfortable with VMware (yet), and -- of course -- the Jive people have only a faint idea of what I'm talking about.
The second has to do with licensing. We're going to propose an "unlimited user" license to Jive, and we're going to run it on a single, physical CPU, which is one of the licensing models they proposed. However, we're going to partition that server using VMware, and we're going to support as many different communities as the server allows.
Here's Why This Is Important
We all want social media to proliferate, us, Jive, etc.
Sometimes this takes the form of big, honkin' mega-communities. The current hosting models and Jive's licensing model supports this sort of pathway.
Other times, we're looking at a large collection of smaller communities. Sure, some of them may get big some day, but they don't start that way.
And we're going to need a pricing model (hosting and licenses) that supports this evolutionary path.
Otherwise, this party is over and we're going to go in an entirely different direction. I don't want to play that card, but unrealistic pricing just cuts the conversation short.
Like asking customers to pay $100,000 for a short airplane ride -- not a lot of people are going to be flying.
And if you're in the airline business, you may want to give that some thought.
This is a tricky problem. The needs of the buyer are in direct conflict with the software provider. I suspect traditional license software sales, especially with the proliferation of SaaS, are going to change significantly over time. My company spends alot of time debating this very issue. We have lots of ideas about how to price in an innovative way that is consistent with the amount of value the customer gets, but the journey never seems to end. Check out www.groupswim.com under pricing to see how we've approached the problem.
Posted by: Jason | February 04, 2008 at 08:20 PM