July 31, 2014 by MongoDB | Comments
Even small organizations struggle with change. But imagine that you have 103 million retail customers, roughly 1700 retail locations to serve them, and $81 billion in revenues at stake. Change necessarily comes hard to a company of that scale and reach.
But change is precisely what Verizon Wireless increasingly enables using MongoDB.
In an organization the size of Verizon Wireless, the business needs are constantly growing and changing, as Shivinder Singh, Senior Systems Architect at Verizon Wireless, told an audience at MongoDB World 2014. These forces push Verizon Wireless to explore new and innovative ways to process manage its data as it seeks to drive greater customer value for its customers.
One of those "new and innovative ways" is MongoDB, which helps Verizon Wireless get greater value from its data while simultaneously accelerating time-to-market and improving its asset utilization.
As the company looks to augment its existing technologies, however, there's always a fair amount of trepidation, not to mention the ever-looming question: why can't we just do this with the technologies we already own and/or know?
Data is changing. The world of relational databases at times doesn't fit the new world of unstructured or semi-structured data.
Traditional technologies which at times would require a dedicated resource weeks to setup a environment could be achieved fairly quick with MongoDB. In a certain case, with MongoDB Verizon Wireless was "able to do that in two hours."
Even so, Verizon Wireless discovered that one of the biggest challenges in moving to MongoDB was to "unlearn" RDBMS concepts and change the mindset to embrace new MongoDB and NoSQL concepts.
But we're getting ahead of ourselves here. How did Verizon Wireless start using MongoDB?
Verizon Wireless opted to start small with MongoDB, though it did try before it bought, one of the cardinal virtues of open source. (More on that below.) The company decided to augment its employee portal, a business critical application that is "basically the homepage of anyone who works for Verizon."
The existing portal was good, but Verizon Wireless wanted to build in new functionality to capture social feeds from Twitter and Facebook and display it specific to that user.
Not so easy for a relational database.
Originally the development team put MongoDB through its paces, first running a proof of concept and then rolling it out. They didn't have anyone dedicated to supporting it, however, so the development team asked Singh's team to support it.
To bring himself up-to-speed with MongoDB, Singh took the route that over 200,000 other people have taken: MongoDB's free online training.
As he describes it, within two days he was at a level that he could comfortably manage MongoDB. Within just two weeks he had re-architected Verizon Wireless' entire development set-up to be in a replicated cluster versus a standalone cluster.
He then proceeded to test and break the cluster, recover it, test the recovery, test failover capabilities and more.
But Singh wasn't done yet.
Going with a new technology can be risky, but choosing a new technology vendor to support is perhaps even more so. To minimize that risk, Singh decided to put MongoDB - the company - to the test. So Singh did what any other conscientious would-be buyer would do:
He faked his death.
Well, not his death, per se, but the death of his server (along with the secondary data center, just to make things doubly interesting). Of course MongoDB would quickly respond to a marquee customer like Verizon Wireless, however, so he also faked his identity, using an @yahoo.com email address.
In other words, MongoDB's support team got a call from some no-name person with a generic email address claiming "my-server-is-down-the-world-is-on-fire-someone-help-me-NOW!"
Within "a short period of time" MongoDB had assembled its engineers to resolve the issue and get Verizon Wireless back on track. Only then did the MongoDB team learn the real identity of Singh and win the deal.
Looking forward, Verizon Wireless has already started a new proof of concept for an online log management system. Not surprisingly, Verizon has "some huge servers, some huge clusters, and all of them generate a huge amount of log data."
Given Verizon Wireless' data volumes, it also is looking for ways to pair MongoDB with Hadoop to leverage the strengths of both together. The company has been evaluating the MongoDB Connector for Hadoop.
As Verizon Wireless moves forward, Singh notes that MongoDB is appropriate for "quite a lot" of its new use cases, and is therefore being evaluated for these new use cases alongside its traditional RDBMSes. That's a big change for a Fortune 50 enterprise, but Singh believes it's necessary to help the company grow and evolve to meet customer needs..
To view all of Singh's slides:
To watch the video, please click here.