June 18, 2013 by Matt Asay | Comments
Despite the day-to-day ups and downs of individual relational database (RDBMS) vendors, there is a clear industry-wide shift to NoSQL databases, and particularly MongoDB. Over the past six months, over 100 organizations dropped RDBMSs for MongoDB. These paying MongoDB customers include tier-1 investment banks, global telcos and several government agencies. Were we to include companies switching from RDBMSs to open-source MongoDB without a formal MongoDB customer relationship, the universe of RDBMS defections would balloon further.
These customers are deserting legacy relational database technology at an ever-accelerating pace because it simply doesn’t fit how modern developers work.
It's not that RDBMS has suddenly become bad technology. It's not. The problem is that RDBMSs are fighting gravity, trying to remain relevant in the modern era with legacy technology, an effort that will almost certainly fail over time.
Not that relational database vendors can do much about this. While RDBMS technology was foundational for the last few decades of computing, it is not particularly relevant for the next few decades of computing.
Small wonder, then, that Cowen & Co. analyst Peter Goldmacher highlights a shift in developer requirements for recent earnings pressure on a variety of vendors:
While management pointed to sales execution as the primary issue, we have a hard time believing that almost all the legacy software names are suffering from poor sales execution at the same time. We believe the primary issue is a fundamental shift in the technology landscape away from legacy systems towards a new breed of better products at a lower cost for Data Management, Apps and in other areas. Virtually every emerging software trend is having a deflationary impact on spend.
That is, unless you're one of those vendors, like MongoDB, in which case the "fundamental shift in the technology landscape" is having a very positive effect on our sales and overall business. In fact, as mentioned above, over the last six months, over 100 customers moved off RDBMSs to MongoDB, far and away the world's most popular NoSQL database.
One such customer, MetLife, tried unsuccessfully for two years to integrate 70 different databases into one consolidated, coherent view of its customers using RDBMS technology. With MongoDB, MetLife had a proof of concept up and running in just two weeks and was in production within 90 days. Another, Carfax, migrated its 11 billion record database to MongoDB to gain development agility and "position [themselves] for future growth." The list of such RDBMS-to-MongoDB converts is long, diverse and growing.
MongoDB is driving many of these trends that shift interest away from RDBMSs to NoSQL databases like MongoDB, including:
Some vendors has tried to catch up with these trends, either releasing their own NoSQL databases or bolting NoSQL-like features onto RDBMS technology. Neither approach has really worked, and MongoDB is on pace to be the world's fourth most popular database by year's end, displacing all but Oracle, MySQL and Microsoft SQL Server.
In fact, MongoDB is the only NoSQL database to crack the top-10, which suggests that it’s not simply NoSQL that is rising, but a particular NoSQL database: MongoDB’s document database. MongoDB is a general-purpose database that fits a broad array of use cases without sacrificing performance.
It's not surprising, therefore, that IBM recently threw its considerable weight behind the MongoDB industry standard. Nor is IBM alone. Informatica and other leading technology vendors have announced support for MongoDB, generally the only NoSQL technology they support. Over time we'll see a few NoSQL technologies become significant threats to old-school RDBMSs, but today MongoDB is the RDBMS' most significant long-term competitor.
A short history lesson may help to understand why relational database technology doesn't fit modern applications, and how the industry has shifted in favor of MongoDB and NoSQL.
Consider the fact that RDBMS technology has been on the earth longer than many of today's developers. Born in the early 1970s, Structured Query Language (SQL) was designed to overcome deficiencies in IBM's IMS and the later IDS, two databases that delivered high performance but forced developers to worry about both query design and schema design upfront and hence made it hard to change anything mid-stream.
SQL decoupled query design from schema design, enabling developers to focus solely on schema design with confidence that they could later query their data as they wanted. It was an exceptional feat, and has carried the industry for some 30 years since.
Unfortunately, RDBMS technology's biggest strength (fixed schema design) has become its greatest weakness. At Craigslist, for example, a simple need to add a column to an RDBMS took three months to complete as the system needed to plow through terabytes of data, adding the column row by row. The classifieds leader was forced to update its application once per quarter. Craigslist switched to MongoDB and such changes take minutes, not months.
In modern application development, which depends so heavily on agile, iterative development, RDBMSs no longer tightly fit the needs of a swelling universe of applications.
When what we built were general ledgers and payroll systems, RDBMS worked great. But today, we’re building applications that are mobile, social, involve high volume data feeds, incorporate predictive analytics, etc. These modern applications? They don’t fit RDBMS.
MongoDB is empowering the next generation of applications: post-transactional applications that rely on bigger data sets that move much faster than an RDBMS can handle. Yes, there remains a relatively small sphere of applications currently unsuitable for MongoDB (including applications with a heavy emphasis on complex transactions or where the relational model is a tight fit), but the big needs going forward like search, log analysis, media repositories, recommendation engines, high-frequency trading, etc.? Those functions that really help a company innovate and grow revenue? They’re best done with MongoDB.
This isn't to suggest that MongoDB is perfect. But the range of applications for which MongoDB is actually a much better fit than even the best RDBMSs is large and growing. Over time, we expect NoSQL databases like MongoDB to become the industry's default databases, just as The Guardian now has a "MongoDB First" policy.
This shift is a force that no RDBMS vendor can effectively fight using legacy technology over time.