Database Reference
In-Depth Information
when the data gets larger, figure out how to simplify those systems; rather
than start out with something simple and later rework the models when the
data set size grows. That's really been my tack thus far academically and in the
startup world. I think everyone at P(K) shared that bias as well.
Gutierrez:
As you're building these tools for yourself, is there any chance
you'll go and build another tool company?
Jonas:
I've thought of it. I actually just received funding from DARPA to
fund some of the construction of these tools and to hire some engineers to
specifically build them, which is kind of exciting. But I don't know. The problem
is that it's really hard to do a tools company, especially an open-source tools
company. Tools companies don't fly anymore, especially for these sorts of
things. Of course, the entire big data ecosystem is all built on top of these
tools.
But how large is the space for companies like Cloudera? I'm not really sure.
I think most people who buy things from Cloudera are buying them because
Mike Olson did a good job selling them something and there's a CIO someplace
whose CMO has turned to him and said, “I want to do what Target does. We
need a Hadoop.” So they call up Cloudera and they're like, “I'm buying two
Hadoops.” And then they're like, “Great. We just bought two Hadoops. Now
we have some Hadoops.” And you're just like, “There's no value being created
here.” I think that's how a lot of the most successful tool companies have
managed to get off the ground.
The other thing I really learned at P(K) is that the right thing to do is to not
build a tool company but to build a consultancy based on the tools. Identify
the company, identify the market, and build a consultancy. Later, if that works,
you can then pivot to being a tool company. If you're selling to the enter-
prise, which you should as they're the only people with money, you're never
going to make headway without a substantial services arm. So start with the
services arm first because it's quick revenue, it's nondilutive, and it's great. If
that works and gets traction, then you can go down the standard Silicon Valley
VC trajectory.
I think there are a lot of data-related startups right now that, had they started
by doing that, would be in much better shape. What you don't want to be
doing is burning through VC money just to figure out who your customer
is going to be. It's a painful truth and it's hard work, but it's much better to
approach building a company that way. We had meetings at P(K) around this
question of turning to consulting or continuing to build the platform even
after we had taken VC money. Ultimately we decided to shoot for the moon
and it worked out very well for us. However, having gone through that, I now
think the right thing to do is to start out with a consultancy.
