Cloudera’s business relies on its ability to deliver best-of-breed support, services, management, and tooling surrounding its open source core. Since its inception in 2008, the company has led the big data market by driving the adoption of Apache Hadoop and related technologies.
This message has been made clear through Cloudera’s ongoing Voice of the Customer (VoC) survey program which measures clients’ satisfaction with Cloudera and its ability to resolve support challenges. TIme and again, Cloudera’s VoC survey results reveal two key messages.
"Our top priority when interacting with Cloudera is that you can solve our problems quickly."
“We love Cloudera’s Proactive Support.”
This feedback has driven Cloudera Support’s mantra: “know the customer.” To expedite resolution of support tickets and to help Proactive Support address potential issues before they occur, Klein’s team built an enterprise data hub (EDH). This EDH—known internally as the Customer Support Interface (CSI)—ingests data in many forms, from many sources, to provide a comprehensive view of each client. CSI is accessible through different tools that suit a variety of users and their needs. The result? Cloudera’s time to resolution on customer support tickets has been reduced by 35%.
Cloudera has found success by providing the highest quality technical support available in the industry, and has hired many of the best, most experienced Hadoop developers and engineers to do so. With Hadoop’s evolution from a niche technology used by web and media companies to a core component of the Fortune 100’s data management infrastructure, Cloudera needed to find a way to scale the level of support it could deliver. Its customer base was quickly growing and those customers relied heavily on Cloudera for support.
To expedite support case resolution, Cloudera Support built its Customer Support Interface, CSI, on top of an internal HBase cluster. Cloudera Support rolled out an early version of CSI on HBase, with a custom-built user interface (UI), and received positive feedback. But there were still areas for improvement. CSI’s ability to thrive required two key capabilities.
Data ingest, visualization, and real-time analysis of large-scale log files: The system needed to give users the ability to quickly parse through diagnostic data, often from many different machines, and to correlate events that happened within the files at similar times.
Combination of log files with other data sets in a consumable form: CSI would be much more effective if the diagnostic data could be combined with support case metrics and content from external sources, in a format lending itself to exploration.
In evaluating technologies that could deliver these capabilities and plug into their HBase- powered CSI, the team realized they didn’t need to look far to find the best solution. “Both Cloudera Search and Cloudera Impala were superior tools,” said Klein. “We would’ve chosen them even if we didn’t work at Cloudera.”
Impact: Time-to-Resolution Reduced by 35%
CSI has become one of the single biggest factors that has allowed us to improve our service by expediting time-to-resolution,” said Klein. “We have data suggesting that in cases where CSI was leveraged, we’ve seen about an 35% drop in time-to-resolution.”
By centralizing all customer data in Cloudera’s EDH, the company’s proactive support organization can run predictive analytics that prevent customer issues before they happen. Further, COEs (customer operations engineers ) can provide guidance on tools to use or best practices for deploying new use cases on Cloudera, based on similar use cases in production at other accounts.
“The original purpose of our team was to enhance the support of each individual case or customer, but because we’ve collected all of this data, we have the ability to look over all of it and say, ‘What kind of hardware are most customers running?” said Alan Jackoway, a Cloudera developer. “If we get a problem that is identified as a bug, we have the ability to say, ‘These five other customers have the same configuration that we found to be bad in this support case. Let’s reach out to them and try to solve their problem before it actually runs out in production.’”
- Time-to-resolution on support tickets reduced by 35%
- Predictive analytics empower Proactive Support to prevent problems before they happen
Big Data Scale
- 60-80 TB on CDH, growing to 100s
- 28,000+ customer nodes observed
- 150+ total active users; 30-50 users every day