With high-velocity data coming in from a wide variety of sources, implementing a Big Data stack makes sense but does it solve all your data problems? Having all your data on Hadoop, without being able to access it through a BI (Business Intelligence) platform is like keeping your data in cold storage until it loses its sheen.
Storing rich historical data on Hadoop is the first step, but once you have it all in one place, it becomes necessary to make it accessible to all the stakeholders who need it.
Moved to Hadoop? Now what?
To get insights into your business, you need instant access to Hadoop data so that you can generate reports and get answers to your questions. Most organizations find it challenging when users try to connect traditional BI and analytical tools directly to Hadoop, as they face significant performance issues.
When the size of data gets massive, and the complexity of reports increase, most BI tools fail on response times. Wait times get larger, ranging from a few minutes to a few days and business users can’t wait for that long. As a result, data cannot be used for problem-solving or better decision making.
Traditional OLAP and its problems
Organizations have tried several approaches to solve the problems associated with the speed and scale of Big Data. When they attempt to use traditional OLAP (Online Analytical Processing) solutions on Big Data, they fail as these conventional tools cannot deal with the volume, cardinality, dimensions, and the variety of Big Data.
Another common approach is to pull data from Hadoop to an external data mart and then perform analysis, but OLAP-in-memory causes latency and brings in limitations on the amount of data that can be processed.
Thus, it becomes necessary to use a solution that moves analytical capabilities to data instead of trying to move data to the analytical environment.
OLAP on Hadoop comes to the rescue
OLAP on Hadoop solves the problems of Big Data analytics without the need to move data out of the Hadoop platform. Multi-dimensional OLAP cubes are created directly on Hadoop, and these cubes provide instant response to all queries enabling quick analysis on massive amounts of data and variety of metrics.
Kyvos offers its innovative OLAP on Hadoop technology to pre-aggregate huge volumes of data across multiple dimensions and build highly optimized cubes that are stored in a distributed manner across the Hadoop infrastructure. These multi-dimensional cubes store data in the granular form supporting deep analysis across any dimension with interactive response times.
The magnitude of cubes that Kyvos builds is much higher than traditional OLAP tools or OLAP-in-memory solutions, making the solution infinitely scalable. Once created, the OLAP cubes can be processed incrementally, making it easy to ingest new data. Users can use their existing BI tools to connect to these OLAP cubes and conduct instant, interactive analysis on their Big Data.
Case in Point: OLAP on Hadoop at Bell Canada
The OLAP on Hadoop offering from Kyvos is being used by several fortune companies to enable instant BI on Hadoop. At the Strata Data Conference this year, Kyvos’ customer, Bell Canada, presented how they increased the scale of BI exponentially with OLAP on Big Data. Bell Canada, one of the largest telecommunications companies in Canada, offers Internet, Wireless, TV, Home Phone, and other services to its customers and relies heavily on data to make accurate business decisions and deliver better services.
With thousands of reports of varying types and frequencies, used by over 10,000 employees across the organization for different purposes, they were facing issues in giving interactive access to huge volumes of data that was being generated continuously. Their existing BI tools, MicroStrategy and Tableau, were struggling to deliver performance or deal with the increasing scale of data.
They implemented Kyvos’ OLAP on Hadoop solution to meet the reporting needs and performance expectations of their business users. Kyvos helped them increase the scale of their BI by delivering instant response times and very high performance on almost any size of data and surpassed their business SLAs of 3-5 seconds response time with sub-second query responses.
Want to try OLAP on Hadoop?
If you want to get more details on how OLAP on Hadoop can be used to increase the scale of BI, download the Whitepaper on BI on Big Data Trends.