In October 2012, Third Iris was intent on expanding its video surveillance as a service (VSaaS) business, but realized that its proprietary database could not perform at the level needed.
"We were making a bigger investment in growing the business, and so we needed an architecture that was scalable," said Chris Grillone, President, Third Iris. "The problem was the scalability and reliability of our proprietary database, which wasn't really designed for what we were doing."
"We had a product that customers really liked, and we felt that we could hardly keep up with the demand with our existing solution," said Third Iris Chief Architect Robert Harvey. "Essentially, we saw that with the addition of a relatively small number of new cameras we would have reached a threshold, and the system probably would have fallen right on its face."
Third Iris provides cloud-based video surveillance to enterprise clients. "For our solution, all you need is the IT camera," said Grillone. "You don't need any extra hardware or software installed somewhere. On our cameras the video is processed, encrypted, and stored temporarily."
"Our customers include very large enterprises," added Grillone, "multinational companies with cameras all over the world. We have cameras on almost every continent right now."
"Our timeline essentially is built with our metadata," explained Grillone, "so that the user can accurately and easily search through a lot of video and find an event that they're interested in."
"We need to have a very accurate, scalable database, because we are sending tons of metadata to the cloud," said Grillone. That was what Third Iris lacked, prior to its deployment of Cassandra built by DataStax.
Legacy solution: unable to scale and store metadata
"The problem," said Harvey, "was that we had a mechanism for storing metadata and videos for our camera surveillance product that had proven to not scale."
"We spent countless hours," added Harvey, "juggling things around just to keep the existing system going without impacting the customers too much. We were absolutely feeling the pain of the scale wall that we had run into with our old system."
"Being a small company and not having resources to dedicate to dozens of engineers to manage database systems," explained Harvey, "we were trying to store metadata in the most efficient way we could, while at the same time keep our existing customers at least at the same service-level."
"We wanted to be with an architecture that was easily scalable, easily maintained, and allowed us to migrate easily," said Grillone, "because we were looking from going from our proprietary infrastructure up to AWS. And within AWS we wanted to be able to easily scale and then deploy over multiple zones."
DataStax had the support and management tools
"We began looking at NoSQL solutions," said Harvey, "having decided that's the direction that we wanted to go right from the start. After investigating the leading technologies at the time, we figured out that Cassandra was a very good fit for us."
During the evaluations, said Grillone, "I got feedback from the team. We did not want to go completely with open source, because because part of the team would have to learn open source management of a Cassandra implementation. That would have taken many more months to get to where we are now, and we were growing."
DataStax training and support got Grillone's attention. "We really wanted to leverage the training that DataStax provided us for their OpsCenter. So that's why we went with the Enterprise version and have the support, because it is really important to our business to make sure our databases are correct."
"Because we had no knowledge at all with anything to do with NoSQL, we needed a lot of help quickly," said Harvey. "Rather than focusing on a generic training, [DataStax] was very targeted, saying, 'we see your problem, and we think this is the quickest way to solve it'."
Grillone added that "we went with DataStax OpsCenter because it would allow us to easily migrate data to multiple zones, and it is easy to back up and maintain."
Third Iris began its deployment of Cassandra built by DataStax in December 2012. "We were up and running in Q1 of 2013," said Grillone.
Results of DataStax Enterprise
Training enables speed of implementation
"DataStax Enterprise allowed us to go from a less reliable, harder-to-manage architecture," said Grillone, "and allowed us to migrate quickly to the cloud, with the training and the support we needed."
"Having all that support," added Grillone, "it didn't require people to become experts and take the time to figure out Cassandra."
Scalability supports growth
"Cassandra has proven to be very scalable for us," said Grillone. "And since then we've grown a lot."
"We are not limited by scale anymore," said Harvey. "Scaling decisions are secondary to strategic decisions, things like deciding where to put servers in the world, rather than what limitations each location has for us."
"We have greater availability of our cloud service," added Grillone. "We no longer have to worry about scaling servers and our database as our business grows."
Dramatic increase in cloud service reliability
"Our cloud service reliability increased dramatically," said Grillone. "It has been extremely easy to scale within AWS, as we get a rapid increase in customers."
"We have a lot more reliability in the system," said Harvey. "We've got the redundancy we need now, that we really did not have before."
Growth and management across AWS zones
"We've had very rapid growth with very large customers," said Grillone, "and we are easily able to remotely manage all of our nodes within AWS. We know that when the time is right, we will be able to scale across multiple AWS zones."
DataStax and future goals
"When we decide to go to AWS EC2 virtual private cloud," said Grillone, "DataStax OpsCenter should allow us to easily do that without any downtime."
"Within the organization," said Harvey, "I would say that we just don't have to think about the databases, we don't have to plan excessively around them, like before. We can focus more on application development."
Brian will do client work for AtTask.
Brian Taylor is a contributing writer for TechRepublic. He covers the tech trends, solutions, risks, and research that IT leaders need to know about, from startups to the enterprise. Technology is creating a new world, and he loves to report on it.