Acceldata Architecture

Event Processing & Data Collection

image30

     

Data is generated at different stages of the application execution. Data access jobs are creating data at various mapper and reducer stages, nodes are producing metrics for I/O and networking status, in addition to JMX and other sources of data. Acceldata  Kafka based event pipeline has the ability to stream such unrelated data sources. 

Deployment Architecture

image31

     

Acceldata is recommended to be deployed in the edge nodes which have access to the cluster nodes and services. Acceldata streaming node requires bidirectional access to the cluster services and nodes for data collection. Limited data, dependent upon the streaming configurations is collected in the TSDB over Kafka stream and over HTTP. 

Flexible Dashboards

image32


Acceldata's flexible dashoarding allows various data metrics to be correlated and plotted together, which enables Data Operations to understand the potential problems. Such queries are 

Intelligence

     

Heuristics 


Acceldata produces heuristics about query execution repeatedly to produce insights in areas where optimizations can be applied. List of heuristics contain default configurations, which are easily extensible by system administrators. 


Root Cause Analysis 


Acceldata collects data from various sources, and then applied unsupervised Machine Learning Algorithm to identify the root cause of the application. This processing is performed on the Application Node, to reduce the load on the TSDB instance 


Anomaly Detection & Time Series Forecasting 


     

Rules can be created to identify various kinds of anomalies in terms of capacity, memory, concurrency and throughput values. List of such anomalies contain default configurations, which are extensible by system administrators. 

See what AccelData #DataOps can do