The acquisitions of Looker by Google for $2.6 billion dollars and Tableau by Salesforce for $15.7 billion highlight just how important Business Intelligence (BI) is in the modern enterprise. Even so, organizations struggle to deliver BI in a way that satisfies both IT and the business.
A recent Yellowbrick webcast discusses how Yellowbrick enables enterprises to reap the full benefits of self-service analytics. Read on for a summary of the webcast.
Business Intelligence imperatives
A recent Forbes article included results of a Dresner Advisory Services study on technologies and initiatives strategic to business intelligence. About 25% of respondents listed “end-user self-service” as critical and over 60% as critical to very important.
The webcast provides the following five guidelines for enterprises to achieve successful self-service analytics:
- Develop and deploy a data handbook with empathy
- Foster a common understanding of the role of data warehouses, databases, and data lakes
- Recognize you’ll likely need more than one BI tool
- Plan for consolidation and workload management
- Understand analytic personas, plan for the enterprise
Develop and deploy a data handbook with empathy
For enterprises to reap the benefits of self-service analytics, they should have a data handbook and data stewards, users should be educated in how to properly use analytics and everyone in the company should be on the same page.
A recent Airbnb initiative illustrates the effectiveness of such programs. Airbnb gave teams that their data scientists support a “Data U” Intensive course for boosting data literacy. This program resulted in a 30% increase in daily SQL user activity, as users were able to now perform their own analytics to hopefully make better business decisions. In addition, Airbnb saw a 50% reduction in ad hoc requests made to data scientists, likely freeing time for the data scientists to perform more sophisticated analytics of their own.
Foster a common understanding of the role of data warehouses, databases, and data lakes
Enterprises maintain many different data stores, including databases, data warehouses, and data lakes. In general, databases are ideal for transactional applications, data warehouses are ideal for delivering analytics for large numbers of users, and data lakes are good for ingesting and storing large volumes of unstructured data.
When possible, enterprises should use a data warehouse for their analytics needs. Running analytics on a database can slow performance for applications and users accessing the database. Data lakes typically cannot deliver the performance needed for fast analytics, support large numbers of users, or provide the analytics capabilities offered by a SQL-based data warehouse.
Recognize you’ll likely need more than one BI tool
There are many Business Intelligence applications on the market that meet many different business needs. For example MicroStrategy excels at delivering real-time insight to users, while Tableau excels at providing a rich desktop experience. As enterprises extend analytics to more and more users, they should not be surprised to find they need more than one.
Plan for consolidation and workload management
To deliver self-services analytics, enterprises need to serve more users and more analytics simultaneously. Self-service analytics programs should include the following:
- A consistent approach across the organization achieved through the use of a data handbook and training.
- A platform that can consolidate systems to ensure data consistency and make both data and infrastructure simpler to manage
- A platform with sufficient performance to serve all users and groups at peak load times
- Workload management to ensure that business critical jobs are not held up by less important ones
Understand analytic personas, plan for the enterprise
Enterprises must make seek balance between meeting the needs of the individual users and creating a scalable consistent approach to analytics that can scale across the business. The table below illustrates two common different analytic personas.
|Ad hoc queries||Regular reports|
Freeform by business user
Unique to user
Business critical info and KPIs
Often received passively on a periodic basis
Expected to look the same
The optimal solution will address common needs of personas and then productize common data exploration tasks to include query tuning for performance, security, and other SLAs that can then be deployed globally.
Picking the right data warehouse for your organization
While there is no shortage of data warehouse options on the market, Yellowbrick has been designed from the ground up to meet enterprise needs, including the need for self-service analytics to many users on a compact footprint. It includes the following features:
- Always on and available
- Ad hoc SQL queries that do not impact operational workloads
- Correct answers on any schema
- Easily scales to petabytes of capacity in a compact footprint
- Fast performance for simultaneous, mixed workloads, including real-time inserts, batch jobs, and interactive applications
- Support for thousands of concurrent users to add analysts and discover new optimizations
Yellowbrick looks just like PostgreSQL to applications making it easy for enterprises to deploy, integrate with applications, and users to run. Some highlights include:
- Real-time feeds Ingest IoT or OLTP data at 100,000s of rows per second
- Periodic bulk loading up to petabytes, over time.
- Load and transform using enterprises existing ETL tools, including intensive pushdown ELT.
- Interactive applications. Yellowbrick serves short queries in under 100 milliseconds
- Powerful BI analytics. Respond to ad hoc complex BI queries in seconds
- Business critical reporting with workload management to ensure fast, prioritized response
- Flexible deployment in your datacenter in a 6U on-premises system or any cloud.
Workload management demonstration
A demonstration of Yellowbrick workload management features shows how Yellowbrick customers will not impact business critical workloads. The demonstration used JMeter to simulate a mixed Telecommunications workload with random think times and diverse random queries (ETL, SAS, Tableau, R, and applications) all coming from many different groups (ETL heavy transformations, data scientists, marketing, CEO dashboard, and callcenter customer lookups).
The demonstration ran over a thousand queries from 402 concurrent users in 10 minutes. 95% of the queries met an aggressive SLA of a responding in just a couple of hundred milliseconds.
Solving for enterprise analytics
No other analytics solution on the market meets all enterprise needs, including support for large data sets, high performance, ease of use, and low cost.
Contact Yellowbrick to learn more about how Yellowbrick can help you achieve your self-service analytics goals.
You can view the 21-minute on-demand webcast here.