Platform Spotlight

This video explains how Pepperdata APM enables you to manage and achieve optimal hardware performance in your multi-tenant distributed computing environment.

Platform Spotlight enables operators to quickly understand performance impacts and receive recommendations on how to better optimize the big data infrastructure. Platform Spotlight enables you to quickly diagnose Hadoop performance issues up to 90% faster while making resource decisions based on user priorities and needs.

360° Platform View

Pepperdata cluster performance monitoring provides relevant real-time and historical information about your cluster including system demand, abusive users, and wasteful applications, as well as queue and container sizes. Drill down or zoom out to analyze any big data app to understand its performance in the context of the entire multi-tenant cluster.

• Complete instrumentation of your big data ecosystem – Monitor any process including Kafka, Impala, HBase, Hive, MapReduce, Tez, Spark, IBM BigSQL, LLAP and more.

• Quantify the impact that big data applications have on your cluster – Identify the impact that applications have on cluster-wide resources such as CPU, disk, memory, network, NameNode, and HDFS.

• Quantify the impact that your Hadoop cluster has on big data Hadoop and Spark applications – Identify which application slowdowns were caused by disk, network or node failures/congestion.

• Attribute accurate costs to cluster users and business units – Pepperdata chargeback provides you with detailed utilization of every cluster resource, which when combined with your cost model, generates the most accurate reports of which users and business units are the most costly.

Real-Time Platform Tuning

Platform Spotlight real-time platform tuning allows you to automatically improve the performance and efficiency of your Hadoop and Spark infrastructure to automatically recover wasted cluster resources, maximize your existing infrastructure investment, and autoscale big data cluster resources for peak efficiency.

• Maximize your existing big data infrastructure investment – Increase hardware utilization and eliminate or reduce millions of dollars in hardware costs.

• Autoscale Hadoop cluster resources for peak efficiency – Rightsize resource allocation based on actual real-time capacity

• Works with existing YARN scheduler – Be confident that all your YARN capabilities will remain intact.

• Identify applications reading and writing small files and improve HDFS performance – Improve system scalability by determining which applications and users are reading and writing small files, impacting the overall performance of your HDFS environment.

Peppedata Platform Spotlight real-time tuning automatically improves the performance and efficiency of your infrastructure to recover wasted cluster resources, maximize your existing infrastructure investment, and autoscale big data cluster resources for peak efficiency. With Capacity Optimizer enabled, some customers see up to a 50% throughput improvement.

Platform Spotlight recommendations enable operators to achieve optimal application and cluster performance on big data, multi-tenant platforms.

Platform Recommendations

Big Data operators are challenged with configuring and sizing critical resources running on multi-tenant clusters with mixed workloads. Pepperdata provides actionable reporting and recommendations to rightsize containers, queues and other resources.

• Size big data queues appropriately based on criticality – get recommendations for queue sizing

• Size containers for optimal resource consumption – Get recommendations for optimal container sizes for workloads

• Accurately forecast resource needs – identify growth trends across resource groups

Alerting

Application Spotlight Enables you to create and receives alerts about events that interfere with application performance.

• Identify cluster bottlenecks

• Use Pepperdata’s rich real-time data to identify resource bottlenecks, including CPU, memory, and IO

• Identify application bottlenecks

Benefits

Operators

• Understand how jobs use resources

• Understand why jobs are running slowly

• Understand performance of different job phases

• Compare job runs over time

Developers

• Generate application-specific recommendations to improve application performance

• Highlight applications that need attention

• Automatically identify bottlenecks, and alert on duration, failure conditions, and resource usage

• Search all applications running on the cluster, compare current and previous runs, and visualize Spark applications and its stages for easy root cause failure analysis and performance tuning.

Enterprise Organizations

• Report on capacity trends

• Get accurate chargeback reporting

• Increase productivity