You hear database, you think Oracle. What does MarkLogic bring to the market that is different?
Our niche is our focus on unstructured data. The relational database was invented to provide flexible queries to structured data, which it does very well. But you couldn’t get good analytics and couldn’t do good ad hoc queries against it. They weren’t thinking of unstructured data – PDFs didn’t exist, web content, blogs, Tweets didn’t exist.
I’ve been working with databases for a long time and know how data people think, because I am one. They often think unstructured data is an inconvenience and they will try and jam information in to [a relational database] for which it was never designed and is suboptimal. That’s why we focused on media companies as our first industry, unstructured information is the product for them.
Enterprise IT is much more used to having content be a second class citizen where it’s shoved in to a CMS or SharePoint and indexed with a search engine. Our niche is that we are arguably the only database company that treats unstructured data as a first class citizen.
And that’s becoming more important to companies in these days of blogs, Twitter and so on.
Absolutely. For example: people will run a query against a site and then subscribe to that query as an RSS feed. That requires the dynamic generation of an RSS feed for a user, which is easy to do in MarkLogic because RSS is an XML format and all the content is in XML. Tweets for example are very complex XML structures. Blogs, Tweets, web pages, all this social content, those are hard problems in traditional technologies because you keep the document on the file system, index it with a search engine and add meta data to it and if you wanted to run a query across that you’d need to add another layer on top. So you end up with information spread all over the place and joins are the Achilles heel of databases and they don’t scale. Maybe they worked in a 1990 or 2000-era data warehouse but now we’re going to petascale, joins just don’t work.
So if you’ve gone down that route do think there is something wrong with Oracle’s approach?
There’s nothing wrong as such. You can use a relational database for unstructured data but it’s not what it was designed for. You can use a dime as a screwdriver, it’s just not a very good one.
Ours is a schema-less database; you don’t need to tell us in advance what the schema is. Traditionally with relational databases you have a table with name, age, date of birth, employee number and so on… tell me what the rows will look like and we’ll blast the info in. But what do you do if you have data that doesn’t fit in those rows? What column do I put it in in an Oracle database? Relational databases don’t do well for those kind of applications, or meta data, which contains a lot of text fields. We put it all in one place and we’ll give you standard indexing and full search.
Content from our partners
Is that the angle do you take when going in to businesses?
Oracle is a large company; you have to be very careful how you engage them as a competitor, a lot of companies end up dead fighting with Oracle. When I was at Business Objects [the BI company acquired by SAP for $6.8bn in 2007] we built a billion dollar business almost hiding in plain sight by focusing on building a great BI tool. The way you beat Oracle is focus and excellence in the small areas, which in our case means unstructured data in the media/publishing and financial services industries.
That’s a very narrow focus – doesn’t that limit how far you can go as a company?
No. It’s all about ends and means. If all we wanted to do was build publishing apps we’d be nothing more than a publishing technology supplier. The fun part of running the business is to know how long to focus on certain markets before you rush to be horizontal.
To stay on the Oracle theme for just a little longer, you wrote a while ago that Oracle is making the same mistakes as CA did in the past. Is that something you stand by?
I believe Oracle is executing a CA-like strategy. In the late 80s CA would buy "living dead" software companies for 1X or maybe 2X revenue. It would jack up the maintenance and kind of hold the customers hostage, and that’s effectively what Oracle is doing. Oracle is to the client-server generation what CA was to the mainframe generation. They are definitely the lead consolidator.
It seems to be working for them, though?
It is and it isn’t. Last time I looked the operating margin was somewhere in the 45% range so they’re delivering good profitability. The issue with having an acquisition-driven strategy is they’re addictive because of the inorganic growth tailwind you get for the first year or so. The consequences tend to be that innovation suffers. You can’t buy the number of companies Oracle have and integrate it all while doing fundamental innovation in databases. I would say customers suffer as well. They’ll pay because they have to but eventually there will be a counterforce.
Let’s talk about customers then. Who are we likely to have heard of here?
Oxford University Press, Yell and Elsevier are big names, and we generate about 10% of our revenue here.
Are you profitable?
Intermittently, we run at roughly break even. Our strategy at the moment is growth, we’ve grown significantly over the last few years.
