Cloud platforms enable enterprise companies to begin their big data journey much faster than on-premises systems because of centralization of control and administration, massive geographic reach, and because of the elasticity of Infrastructure-as-a-Service (IaaS) that allows you to create just the right amount of computing power on the fly, according to Ashish Thusoo, co-founder and CEO of Qubole and former head of Facebook’s big data initiatives, speaking at the Apache Big Data conference in May.
Those characteristics allow companies just joining the big data revolution to quickly get to the “holy grail:” a system where anyone can make use of the data, across departments and regardless of their job title, Thusoo said.
“In this world where data and analysis has become a competitive advantage across different industries, it’s very important that companies think about systems they put in place in order to make that data accessible to their users,” Thusoo said. “If you put [self-service] systems in place, the results are transformative. We saw that at Facebook, and I’ve seen it in the context of Qubole. That’s really the holy grail.”
Thusoo told the conference crowd in Vancouver that companies have had trouble implementing big data projects because of four major reasons:
-
They have a rigid data infrastructure currently installed and it’s hard to change those legacy systems,
-
The software they’re currently running for business intelligence isn’t adaptive,
-
There is a skills gap in the organization where many people don’t actually know how to gain access to the data or pull insights from it, and
-
New systems are very difficult to build from scratch.0
Each of those problems point to a cloud-based solution, Thusoo said.
“Is there a mechanism where enterprises can get to this vision faster, in a much less risky manner, and in a manner which is much more failure proof?” Thusoo said. “To me, the answer there is really the cloud.”
Cloud can allow a single network administrator to support hundreds of users and open the doors for new roles to gain access to the data, and implement a single point of control on who sees exactly what data set. Because cloud data centers are built around the world, a cloud-based system has no limitations on geographic reach, Thusoo said.
And cloud-based systems allow you to shape your infrastructure to the job you need done, not the other way around.
“You can create infrastructure to fit your applications, as opposed to the other way around, which is generally the norm,” Thusoo said. “This ability to create dynamic infrastructure itself allows us to think of a problem in a very different way: it allows us to completely automate the orchestration of the infrastructure.”
Watch the complete presentation:
