Carl Ryden is the chief technology officer of
MarginPro, where he manages all product development activities. He is the
author and developer of MarginPro's pricing and profitability management
software. Carl worked with Mitchell Epstein at US Banking Alliance as senior VP
of product development and operations, where he produced a first-generation
net-interest-margin maximizing software. Carl left US Banking Alliance shortly
after it was sold in 2006.
In this interview, we discuss:
-The "no brainer" of the cloud for startups
-The benefits of platform-as-a-service over
-The issue of "lock-in"
-Customer perceptions of one company's SaaS
running on another company's PaaS
-How the cloud is intersecting with risk-averse
industries like banking
-How VCs don't want to fund the construction of
datacenters any more
Could you take a moment to introduce yourself and MarginPro and your experience with cloud
Carl Ryden: I'm
the CTO and founder of MarginPro. We make software for loan pricing and deposit
pricing software that's used by banks of all sizes, all over the world. We
started MarginPro with a blank screen and a blinking cursor in May of 2009 and
turned on our first customer on January first, 2010.
Our application uses Silverlight, talking via Windows
Communication Foundation back to a SQL Server database. We chose that combination
because it's a great fit for what we need to accomplish with banks, which
includes a high level of interactivity and that sort of thing. We built it to
run in a cloud-based environment.
We started with a GoGrid. The reason we needed a really good
data center to run it is because we're providing services to banks and we
needed something that was SAS 70 type II, that was probably secure, and
building that on our own was not an option. It just didn't make sense.
Working with a GoGrid worked great at the time, although it
was a bit of a pain to manage and patch the servers and all that. As soon as Windows
Azure became public, we switched over to it, which took us a couple of weeks of
effort. It was pretty straightforward, and we have had great uptime ever since.
It works great for us.
Robert: There are
lot of articles out there about deciding between the cloud and on premises
systems. Can you talk a little bit about how you guys arrived at your decision?
Carl: As a small
company, the capital and personnel requirements for building our own data
center were prohibitive, particularly since that's not at the core of what we
do. Going to a cloud-based provider was a bit of a no-brainer for us, and it
has allowed us to put our time and capital into making our product better and
marketing it successfully.
Azure ended up being even better than infrastructure-as-a-service.
We were able to get away from the capital cost of the hardware, get a secure
data center and all that, without having to patch and maintain servers as we
would with infrastructure-as-a-service. We really just wanted the platform-as-a-service,
and that has worked out well for us.
didn't start out on Windows Azure;
you started out on infrastructure-as-a-service, and you have alluded to some of
the points that allowed you to make the switch. What costs were you seeing
associated with infrastructure-as-a-service?
Carl: The main
thing is that, since we sell to banks, a security breach could finish us as a
company. It's vital for us, when patches come out for the operating system, SQL
Server, or whatever, not to be worrying about security exposure.
Patching and replacing things is really a painful process,
and it sucked up a lot of our time and energy. Frankly, we couldn't get away
from that fast enough. Now, folks at Microsoft handle the patching and maintain
the operating systems, and we just deploy our solution.
Again, there's more to it than the cost. We have gotten away
from the worry that the business could go kaput because we didn't apply a patch.
We have also ended the painful process of applying the patches, which we
typically had to do over the weekends.
concern that people bring up with platform-as-a-service
is the fear of lock in. How did you view that potential risk, and what
helped you overcome that concern?
always lock in of some sort. It always costs money and it's always painful to
change from wherever you are. You've just got to pick what you get locked into. Azure's like being locked in to luxury
cruise ship with a great buffet.
Maybe I'm locked in, but I like it. Microsoft technologies are just the
best match with what we're doing.
For us, moving to Azure wasn't that hard. Moving back to GoGrid
would cost me money, time, and effort to get into a spot I like less, so
there's clearly no motivation to do so. To me, the idea that there's a
technology decision you could make that doesn't create some form of lock in,
meaning some sort of cost or pain to change, doesn't exist.
often view a platform-as-a-service as really good for new application
development, but not necessarily as strong for having to move existing
applications or legacy applications. Can we talk a little bit about that? In
your case, it sounds like you really had a green-field application.
Carl: Yeah, ours
was new. We started with a blank screen and a blinking cursor, so it was fairly
easy for us to move. In fact, what held us up is that our application was
written and running on .NET 4, and we had to wait for Azure to get upgraded so
we could move. Because we were kind of pushing out on the edge, it was probably
easier for us to move the whole thing, en masse.
My opinion is that it's easy to move parts, and you can pick
different parts that make sense to move to the cloud. And given the level of
interoperability and security, if I had an internal system, I would really
consider moving parts of it to the cloud.
That would particularly be true if I was at the point where I
needed to upgrade a bunch of hardware, to scale it out or anything. I'd move
that piece to the cloud and make it service oriented, because then you kind of
kill two birds with one stone. I've reduced my CAPEX, I've probably reduced my
OPEX, and I've probably improved my level of service, and I've moved to a
service-oriented, more flexible architecture at the same time. So, I think
there's probably an opportunity there for a lot of people.
Robert: One kind
of consistent theme I hear as I talk with customers is this idea of a hybrid
cloud, where I'll keep some of my resources and apps on premises and others in
the cloud, just as long as I have a way of providing seamless access, when I'm
doing things such as Active Directory Federation. But it seems like hybrid
cloud is a logical starting point for organizations that have a lot of legacy
in place. Any thoughts on that?
Carl: To touch
back on lock-in, the fundamental technologies of the cloud, being service-oriented
and that sort of thing, mitigate lock in more than anything else. In many
cases, it's the legacy systems that they are locked into. The cloud is potentially
their path to freedom, so the lock-in concern with the cloud doesn't bother me
We have a partner company who has their own little data
center, where they run all their stuff. And I said, "That's almost a monument
to a bad decision that you had to make at the time," because now they're locked
Starting fresh today, no one would do that. Of course,
people say, "Well, I'm not starting fresh," but other people are, and you live
in a competitive world. You've got to find a way to move to the more flexible,
more agile way of doing things, and the cloud gives you the path to that.
Robert: The early
software-as-a-service providers built their own data centers, or they co-located
some of the equipment with hosters. They could provide a lot of assurances to
customers because they had direct control over the infrastructure. How does
being a SaaS,
running on another company's PaaS, affect conversations with customers?
customers are banks, and banks are risk-averse. When we were selling to them,
if I told them I had my own little data center, or my own colo guy here in
Raleigh or wherever, they would get really scared.
When we were with GoGrid, we'd give them a copy of GoGrid's
SAS 70 and we'd get them comfortable with that, but it took some effort. Now,
when I tell them instead, "I'm running in a Microsoft-run data center" the
conversation is done.
There is a tremendous
amount of brand awareness that I'm selling to business guys. I'm selling to the
chief credit officer, the chief lender of a bank. They might not even know what
cloud is, but they know what Microsoft is, and the weight of that brand is
I just left a meeting with a chief credit officer of a $30
billion bank. When I first met with him, he didn't know what the cloud was. He
actually stopped me and said, "What's this cloud thing you're talking about?" I
said, "We're running in a Microsoft data center in San Antonio." And today in a
meeting, somebody in the room asked a question about where we run. He said, "They
run in a cloud in San Antonio."
Carl: Boy, we've
come a long way in just a couple weeks. That's another reason why moving from
GoGrid and ServePath to Microsoft and Azure was huge for us, because of the brand
recognition among those we sell to.
Again, we want to focus all of our sales and marketing
efforts and all of our delivery methods around loan pricing, deposit pricing,
and our core business. If we end up having a discussion about data centers and
security, we're losing. Now, we don't have to do that anymore, so we can focus
on talking about loan pricing and profitability.
your solution operates in the heavily regulated banking industry, how do you address
customer compliance concerns?
Carl: The key to
our making that work is that if I were to say to a banker, "I'm taking your
account numbers and your customer names and putting them up in the cloud," that
could be tricky. To fulfill our purpose in life, though, we don't need to know
the account numbers. So we do a cryptographic hash, and we can tell to the bank
that the account numbers and other personally identifiable never leaves their bank.
The thing is to figure out which data is subject to regulation
and ask yourself whether you really need access to it. We just need to know
there's a deposit there and have the last four digits of the account number,
and in fact, we can even get by without that in many cases, so we eliminate our
access to it. That's actually a good practice, even if you're doing it on
The problem is that, with on premises systems, people can
tend to get lazy and just move the account numbers all around, which is a bad
idea from a risk-management standpoint. You only want to have personally
identifiable information where you need it, and nowhere else. The cloud just
makes it easier to follow that guidance.
Robert: So Carl,
I know you have some experience with the VC community. Can you talk a little
bit about how
VCs are viewing cloud versus capital infrastructure expenses for startups?
Carl: Well, I
used to be a venture capitalist, and I know a lot of them still. Back in the day,
a lot of the early software-as-a-service guys would build data centers, spend a
bunch of capital, and compete on the data center. Well, now, you can't do that.
You've got to compete on the quality of your application, because the data
center is available to everybody. The data center that I operate in is every
bit as good as the one that Salesforce.com operates in.
VCs look for businesses with large growth potential and
potentially high capital needs, because they want to find that rocket ship they
can ride and put more and more capital into. The fact that using cloud so
dramatically cuts CAPEX changes that relationship.
The amount of capital you need to start up a software-as-a-service
company has gone down dramatically. I can service $10 million of revenue,
paying Microsoft $1,500 a month, and I'm probably guilty of overkill in terms
of the capacity I need, just because it's easy and inexpensive to do.
What it really opens up business opportunities is that if
you have a team of folks with good domain knowledge in a particular area, and
you know how to build quality applications, you're competing on the quality of
your application now, not the quality of a data center. Fundamentally, I like
that a lot better.
It's not as fundable, in some respects, but I think it also
opens up new opportunities from a VC standpoint. A lot of these point-source
applications gather a lot of valuable information, and that information becomes
extremely valuable. So even though they don't soak up as much capital as VCs
might like, the businesses are still very worthy of investment.
Right now, we're pricing about half a billion dollars of
loans through our software each month, and we generate a massive amount of
data. I think there's a tremendous amount of value in being able to collect
data in a massive cloud database and then deliver it out through some of the
delivery mechanisms that Microsoft is putting in place.
Robert: Are you
seeing an uptick in the broad category of financial products?
commercial lending, we do see an uptick, which is great. We see it both on the
operational side of the business, where we see more loans being priced, but we
also see it on the sales side of our business, where there's never been more
interest in our product. People are tired of having money just sitting around. They
want to start putting it to work again.
Robert: What else
do you look forward to in the future of cloud computing?
Carl: From a
technology standpoint, I would say there's nothing. I think platform-as-a-service
is the natural evolution of where things go. There are a lot of systems that
were built not in the cloud, but on big iron back in the day. One of those I
think about is core processing systems for banks.
It used to be that if you were starting up a bank, you bought
an IBM iSeries minicomputer or an AS/400, and then you got one of the core
system providers to put some transactional processing software on that system.
You created a data room with a raised floor and air conditioning, and then you
had a bank core processing system.
That moved dramatically over the last ten years from in-house
to outsourced to data centers. The next logical evolution is to where it's
provided as a service. There's going to be a tremendous opportunity to go into
a bank and tell them, "Instead of putting your capital into your core system,
put your capital into your loan portfolio and lend it out. That's how you make
money, and it will grow with you as you grow as a bank." People will take that
deal all day long.
And I think there's going to be a huge transition, where CAPEX
is going to be owned where it's most efficient to own it. Microsoft has a low
cost of capital, and they can put together and manage enormous data centers
that they sell capacity to guys like me by the drink, who will gladly pay the OPEX
for it, and I can grow with it and scale with it.
From the business standpoint, what excites me about cloud
technology and cloud computing is not so much where the technology's going to
go. I don't know if platform-as-a-service is the final step, but it's close to
the final step in that evolution. Now we're going to start seeing businesses
transform. And how businesses emerge and grow, transform, is going to be around
that tradeoff between CAPEX and OPEX.
The fact that companies are now having to compete on the
strengths of the goods and services they provide rather than their
infrastructure is good for everybody.
Robert: That's a great
place to wrap up. Thanks for talking today.
Carl: Thank you.