Production deployments continue to drive rapid pace of innovation
When we announced AtScale 5.0 you may recall that I was quite excited about the rich set of analytical capabilities included in the release, including multi-fact support and an improved design experience for complex models. These capabilities have been well received by our customers, and have helped them to put more use cases into production using the AtScale Intelligence Platform.
With the 5.5 release we’ve continued to provide new capabilities in working towards a modern BI platform for big data and we have also included a number of “production-ready” enhancements for our existing customers. Read on to learn about the key themes of AtScale 5.5!
Robust Analytic Use Cases
One of the key catalysts for AtScale’s rapid adoption among the world’s largest enterprises is our ability to support the complex business use cases and analytical models that these customers demand. Some examples of these use cases across multiple verticals are highlighted in table 1.
Table 1. BI on Big Data: More than just fast queries
With 5.5, we’ve expanded our support for these use cases by enhancing our ability to handle complex multi-fact models (for example, check out our blog on the “Per Member per Month” calculation). This release includes improved unrelated cell handling to support multi-level metrics for use cases like sales planning. Now, customers are able to provide guidance on which fact tables should be used to provide the raw records used for measure evaluation dependent on the level for which the measure is being evaluated. In cases where a simple roll-up of lower level values is not desirable, this unlocks a number of use cases for AtScale customers.
The video below shows how this powerful new feature helps with sales planning and evaluation.
Production Ready High-Availability (HA) Deployment
AtScale’s customers depend on our ability to always be available to satisfy queries from customer-facing applications that are powered on the back end by AtScale. Additionally, AtScale is used for internal use cases that are a core part of daily business operations. As a result ,it’s imperative for the AtScale Query Service to be deployed in a clustered mode to ensure no down time in case of an individual server or service failure.
In 5.5, it is now possible to deploy all AtScale services in a “hot-hot” deployment mode, as shown in Figure 1.
Figure 1. AtScale Clustered Configuration for High Availability
With AtScale Configured in High Availability Mode:
- Queries can be distributed (via the load balancer) to any AtScale node at any time.
- Query graphs are consistently maintained across the Engine service to ensure consistency of results.
- Metadata services are in sync across all AtScale nodes via PostGres.
- Watcher services ensure that SQL engines (Spark and Hive) are up and running.
Prediction-Defined Aggregates (a.k.a Adaptive Cache 2.0)
Based off of insights from a number of production deployments, we have been able to develop a series of algorithms that are able to anticipate certain query patterns a priori, and create aggregates to support them even before a single end-user query is executed against an AtScale virtual cube. We call these Prediction-Defined Aggregates.
At the core of Prediction Defined Aggregates is a statistics system that is constantly evaluating statistics - row counts, attribute cardinality, join quality - of underlying data sets. Once statistics are available, this information is fed into a series of algorithms that are able to predict the potential value of creating aggregate tables to satisfy anticipated query patterns. To see Prediction Defined aggregates in action you can check out the video below.
As a result of Prediction-Defined Aggregates , end-users experience faster out of the box cube performance, less manual training, and increased aggregate hit rates.
Time to Get Started!
As you can tell— I am very excited about this release. With AtScale, our customers can immediately derive value from the new features above, and will continue to benefit from the industry’s richest business interface for big data:
- Multi-dimensional Calculation Engine providing the industry’s most scalable computation engine for modeling the toughest business processes.
- Performance Optimization Engine powered by machine learning to automatically optimize query performance for “speed of thought” analysis.
- Data Abstraction Layer enabling access to relational and Big Data data sources on-premise and in the Cloud.
- Enterprise-grade Security, Governance and Metadata Management for providing safe, trustworthy and consistent views of any data for any BI tool or custom application.