Last week, an estimated 70,000 people arrived in San Francisco for Oracle OpenWorld. In the course of five days there were 2000+ sessions, 3000+ speakers, and about 800 exhibitors and demos. There were a lot of talks focused on the cloud, flash storage, engineered systems, analytics, and of course big data. MarkLogic had a popular booth, and our VP of Engineering, David Gorbet, was also interviewed on theCube: https://www.youtube.com/watch?v=IeDrVKgl55c.
So, how did we stand out in the crowd? Two ways:
1: Data integration is really hard and MarkLogic has a solution that relational databases do not.
2: Everyone has now heard of NoSQL and MarkLogic is the only NoSQL database ready for the enterprise.
Data Integration Is Really Hard
In the dozens of conversations I had throughout the week—whether talking to tech executives, DBAs, or developers—everyone spoke about the agony of integrating data. It didn’t matter what industry or geography, whether the company was large or small, successful or unsuccessful. They all faced the challenge of managing and using data of different types. And they were coming to us for a solution.
The problem is that people want to use all of their data to make decisions and build new applications and new products. But, data often sits across multiple systems in multiple formats that change constantly, and thus, bringing it all together is really hard. Traditional relational databases are ill-equipped to handle this problem.
We have long said that 80 percent of data is unstructured and you need a NoSQL database to handle the difficulties with that unstructured data. But, the truth is that handling that 20 percent of structured data is also really really difficult.
Combining data from relational databases is challenging because you still have differing schemas, still have to do ETL, and still end up having to make trade-offs in application development. You end up creating lots of metadata to explain rows, storing large columns of data in rows, and doing recursive joining. All of this increases complexity and development time while hurting performance.
In addition to variety, companies now have to deal with larger volumes of data. Relational databases can be configured to handle large volumes, but not easily or cheaply. You need a bigger box, and some new big data software coupled with Hadoop.
Flash storage was a huge focus in many talks at the conference and was a recommended solution to processing large volumes of data faster. Unfortunately, the performance gains from flash storage are also really expensive.
But, a bigger box and a bunch of flash storage still doesn’t solve the fundamental challenges with data integration. In the end, relational databases have no good answer to the challenge of data integration.
When it comes to handling heterogeneous data of different types and from different sources, MarkLogic is the best solution. MarkLogic is schema-agnostic so that you can load data as-is and not worry about differing, changing schemas. MarkLogic also has sophisticated, highly compressed indexes that enable you to search your data and write complex queries. These and other aspects have enabled organizations to solve their data integration challenges with MarkLogic.
The example we talked about at OpenWorld was healthcare.gov, the US Federal Marketplace and Data Services Hub for The Affordable Care Act. MarkLogic is the database at the heart of healthcare.gov, and it helped sign-up over 8,000,000 for healthcare insurance. In addition to managing data from thousands of public and private insurers, the system also brought together immigration data, IRS (Internal Revenue Service) data, credit data, and more in order to verify information and determine eligibility. This project was completed in only 18 months—an impossible task using a relational approach.
Everyone Has Now Heard of NoSQL
Even Larry Ellison was talking about NoSQL in his keynote at Oracle OpenWorld.
We used to spend a lot of time explaining what NoSQL is and why it’s important. We just don’t have to do that anymore. Most people I talked to knew NoSQL – and were in the midst of comparing various NoSQL technologies, trying to address challenges they’ve had with newer open source options, deciding which use case to tackle first, or how to promulgate the technology through their organization.
MarkLogic stands out from many other NoSQL options because it has the enterprise features that organizations need to run mission-critical applications, and it has newer features such as semantics that are on the cutting-edge of database technology.
Many organizations that have been using Oracle, IBM, and Microsoft products have gotten used to having features such as backup and recovery, ACID transactions, and government-grade security. Most NoSQL options do not have those features. MarkLogic does. That’s why we were telling people at the conference that MarkLogic is “NoSQL. No Compromises.”
MarkLogic also makes databases exciting again. Rather than just offering some gradual improvements, MarkLogic is innovating with new features that no one else has. While the idea of semantics is quite new for a lot of people, I know we’ll be asked about it more and more because it’s just another part of the puzzle in handling complex data challenges. Semantics is already helping organizations store and query billions of facts and relationships so that they can gain more context for their data and build more intelligence into their applications. In addition to semantics, MarkLogic also has other cool features such as real-time alerting and bitemporal (stay tuned for more on bitemporal in MarkLogic 8).
After telling someone at the conference about MarkLogic for the first time, he said, “Oh I get it, MarkLogic was NoSQL before NoSQL was cool.” I laughed because that’s been part of our marketing messages for years, but it’s even truer now. MarkLogic is indeed the market leader among NoSQL databases, and is at the forefront of a paradigm shift as people adopt NoSQL technology as a way to transform their businesses. Just like it took a change in mindset to adopt SQL back in the 1980’s, NoSQL is a new way of handling data.