Real-time data is a real big problem. That’s the message that startups like
Coral8 are trying to solve.
Coral8 makes a complex event processing (CEP) engine, using a SQL-based
language to make it easier for developers to write more intelligent
What Akamai was doing is this edge The The story we tell to the developer is the same one that The second thing is the scalability. Once you This is how this game is played: The smaller Wall Street, RFID Crystal turned into a There is a lot of This
Though a pretty green field, CEP is seeing traction in financial services,
up with the data streaming in and out of their networks.
Other startups in the field include StreamBase, AptSoft and Aleri, while
major vendors like IBM
have amassed large amounts of the technology needed for
Coral8 CEO Terry Cunningham, who sold
Crystal Decisions to Business Objects
, knows something
about building a company from the bottom up.
Cunningham discussed the emerging market for CEP shortly after his company
Coral Engine 4.0.
Q: What is the genesis of Coral8 and what market forces drew you to
complex event processing?
The company was born by our CTO and founder Mark Tsimelzon who discovered
when he was working at Akamai that there is a very large need for a platform
to build real-time applications.
computing strategy for 15,000 servers all running applications. So that was
where he first saw the problem.
I came from the business intelligence space, and so I spent a lot of time
dealing with historical data — data in data warehouses and reporting and
querying and analysis on that data. What Mark introduced me to was this
notion of real-time data that flows through the pipe that also needs to be
Back then, there were a few obvious verticals that had real-time data, such
as financials. Wall Street was a big consumer of real-time data and couldn’t
ignore it. They spent millions and millions of dollars building custom
applications to deal with the real-time data, and they’ve been doing that for
My concern at the time was that it was just too small of a market. Over the
last three or four years, what we’ve discovered is that the amount of real-time data flowing through the enterprise is actually quite massive.
reason for Coral8 was that there is so much real-time data but no
infrastructure to deal with it.
Q: What does Coral8 Engine 4.0 do?
It processes real-time data, and we provide a standard interface called SQL
to talk to the engine. We handle the plumbing part and you, as the
developer, build the application and talk to the engine through a standard
engine called SQL.
we told to the developer 25 years ago with the advent of the database and
client-server computing, so we’re basically just copying it.
The two main messages we communicate to our customers is the ease of
development, which is SQL, SQL, SQL. If you’re a SQL guy, you’ll pick this
thing up and you’ll be running in an hour. You don’t need system engineers
showing you how to use it.
build your app, you’ve got to know that it will scale to the large
enterprise challenges. Can this thing handle the data and the volume that
we’re talking about? We’ve spent a ton of time and money building this
Q: IBM and Oracle are quietly working on complex event processing products.
If Coral8 and other startups foster enough market interest, they’ll likely
increase their efforts.
Absolutely. We’re in discussions now with IBM and Oracle about a CEP
language, so that at least we can come together as a group and agree on
syntax and format. We fully expect them to be in the market some time in the
future. This is a major space and they cannot ignore it.
The game we play is to launch early, launch often and get this technology
out there and as fast as we can into the hands of people that will influence
its success, which is the developer.
Q: Could you find yourself selling another company in a couple of years?
Obviously, this market will consolidate as the customer demand takes off.
Keep in mind that for the Oracles and IBMs of the world, $20 million, $40
million, $100 million in revenue for them doesn’t really matter. It has to
be big enough for them to care.
companies build early market share and validate the market. When the IBMs
and Oracles decide they have to own it, they start looking at buying
Q: How big a market is this going to be?
When we look at the market, we look at it in two phases. The first is early
adopters and early need.
analysis, are what we consider to be the obvious market opportunities that
need this technology today. That is a healthy but small market. I can’t put
a number on it, but when I look back and do a comparison and ask is this
bigger or smaller than what I started at Crystal, my view is this is a
bigger opportunity than what I started at Crystal.
$350 million software company, all on the basis of client-server databases
and reporting, so there is obviously enough market today to keep the
StreamBase’s, Coral8’s healthy, profitable and growing.
headroom for a lot of these small startups to make a living in this space
Where it gets exciting, an inflection point, is the second phase, where
software companies that are thinking about and starting to re-architect their
applications, like ERP [enterprise resource planning] vendors that are
considering a whole new architecture around event-driven technology.
will require ERP
with events as a major part. When that happens, this becomes a
What Akamai was doing is this edge
The story we tell to the developer is the same one that
The second thing is the scalability. Once you
This is how this game is played: The smaller
Wall Street, RFID
Crystal turned into a
There is a lot of