Recently there have been a few high-profile examples of small cloud providers turning off support for one or another public clouds. First came AppFog who made the mysterious decision to cut off all support for their PaaS running on Rackspace infrastructure. Only a few days later Xeround announced that they were discontinuing their public cloud database service. It would be easy to parse these two events as simply a case of a small vendor not gaining early traction and shutting down – there’s certainly an element of truth in that perspective. In fact it’s not an uncommon situation to see howls of protest when an entire company, or a particular service a company provides gets shut down – we only need to look at the recent example of Google shutting down Google Reader for proof of that. Another example is OfficeDrop who announced the shuttering of their entire service. As an aside, while the Google Reader news got lots of attention from enraged bloggers, the OfficeDrop was perhaps more impactful in the real world, as ReadWrite pointed out:
The service has been aimed at small- to medium-sized businesses (SMBs), but this experience may leave a poor taste in OfficeDrop users’ mouths. Given the just 16-day notice they received and the hard stop for their files’ existence on the cloud-based service, (a hard stop that falls on a Sunday, which will probably make for a blown weekend for quite a few people), this could put SMBs with potentially limited IT resources in a bind.
Anyway – back to Xeround and AppFog who, I believe, are early victims of a growing theme. I think these two companies are symptomatic of a broader situation that currently exists. It seems to me that the public cloud vendors are running as fast as they can to encourage third parties to launch and host free versions on their partner platforms. Indeed looking at the investment that goes into the marketplaces of arguably the two biggest cloud vendors – AWS with its Partner and Solution Network and Rackspace with its Cloud Tools Marketplace – we can see how importantly these companies regard a growing and vibrant ecosystem.
The reasons for this are obvious – a cloud approach towards infrastructure and IT more generally is enabling organizations to obtain technology stacks that are uniquely tailored to their particular use case – part of this tailoring includes the ability to pick and chose from a vast selection of particular point solutions – want your NoSQL database with an extra dose of scale and integrated with a monitoring solution that has a point of presence in outer Mongolia? There’s a good chance you can find it in these marketplaces (actually I checked, not so much on the Outer Mongolia thing). Organizations genuinely enjoy the benefits of having massively flexible third party integrations.
But there is another side to this mad scramble to get partners signing up to a marketplace, and that is the mismatched size and scale of the two parties here – the marketplace provider and the third party vendor. And I believe Xeround and AppFog are initial victims of this mismatch. While clearly these startup companies have reasonable funding (AppFog raised $10M, Xeround $30M), they can’t afford to keep giving unlimited numbers of customers free accounts indefinitely – eventually the economics need to come into play. If we look at Xeround as an example, they were offering free (as well as paid) service options running on AWS, Rackspace, Heroku and the HP Cloud. While it’s great for a third party to support lots of different public clouds, it’s even greater for those public cloud providers to have a bunch of ad on services on hand – as Lucas Carlson, CEO of AppFog said to me in an email:
Watching unit economics and modeling your startup revenue forecasts are absolutely critical exercises when building public cloud businesses. In many SaaS-based businesses, the per-user costs can be relatively negligible. A public cloud based SaaS service may have significant hidden costs so it is important to track every public cloud service associated with each of your users and make sure the economics work.
I sense that what Carlson isn’t saying is that he feels a little burned by the public cloud vendors who, after scrambling fast to get partners signing up, have kind of ignored the cost implications of that on their more fledgling partners. And ignoring this fact will likely prove damaging to the cloud providers themselves. Already there are some murmurings of concern about the AppFog and Xeround news – organizations relied on these services running on clouds of their choice, when these vendors remove support, it not only calls into question the third party (as my friend Paul Miller points out in this reflective post) but it also casts a broader shadow on cloud ecosystems – if you can’t rely on a part of the stack, the feeling goes, then perhaps it’s not wise to rely on the entire stack. A growth in this sort of concern could prove damaging for the large cloud providers. It seems to me that the “big boys” are thinking a little too short term, they should be looking at the longer term impacts of services going down – it’s a bad look for them.
Carlson touched on this concern when he mentioned the economics that he, and by extension other third party vendors need to think about. He also provided a thinly veiled criticism about the intentions of the public cloud vendors:
That’s why watching unit economics is so important, if you can’t fully track and understand the lifetime value, cost per acquisition and cloud cost per user, then you have no context for understanding how and when economies of scale kick in. For example, if you are going to buy a bunch of reserved instances at AWS to reduce your cost per user, but your cost per acquisition of a paying customer (which may include the amortized cost of free customers) is more than the lifetime value of a paying customer, then effectively you have become a subsidized free marketing tool for cloud vendors who end up being the only ones making a profit because they were watching their unit economics.
A point reiterated by Paul Miller in his post where, talking about Xeround, he says that:
Last year, the company rolled out a freemium offering to entice a far wider set of potential customers. It would appear, though, that any influx of paying customers was insufficient to meet the cost of this rapid expansion across disparate cloud providers.
And therein lies the rub. In an effort to gain critical mass as quickly as possible, cloud startups are rushing headfirst into joining ecosystems that may prove to have bad economics for them. Similarly, in an effort to offer every long-tail opportunity under the son, public cloud vendors are signing partners up left, right and center – with little regard for the economics the partner may face. And caught in the cross fire are customer who have an understandable, if naïve, belief that by being in a partner ecosystem, a product has the economic viability to continue to do so in the mid to long term. Alas this isn’t the case and, as always, caveat emptor applies. At the same time however the public cloud providers have a burden to shoulder here – if they don’t they might see some damaging side effects come back to haunt them.