Distro compute

Introduce

WorkloadsDistributed compute in a nutshell (where many nuts > few nuts)

Workloads

Distributed Compute Developer

Big Compute Big Penguin Big Data

BIG ANYTHINGHello All Worlds

Big Co$t?

That’d be like Microsoft, right?

Microsoft Research Genomics

InspectDistributed compute in a nutshell (with many little nuts)

Inspect

HPCHead Node

Broker Nodes

Compute Nodes

Allows on-premises

And hybrid option

Compare Architectures

Big DataName Node

Data Nodes

Allows cloud or on-premises no hybrid option

Hadoop

HPC

All distributed compute works on the basis of taking a large JOB and breaking it to many smaller TASKS which are then parallelised

Develop

Deploy

ExamplesHow do you take your compute?

OUCHLessons learned from getting too close to the coalface

A broken cluster is no place to be diagnosing

Scalability < Elasticity

Hybrid HPC is next to useless

95% through a petabyte is a bad place to find a bug

Thank you!