at Your Fingertips, in the Azure Cloud An Azure/UberCloud Sales - - PowerPoint PPT Presentation

Engineering Simulations at Your Fingertips, in the Azure Cloud

An Azure/UberCloud Sales Guide

For Use by Microsoft Sales

Wolfgang.Gentzsch@TheUberCloud.com

Benefits of HPC* as a Service

Right-Sized Infrastructure

1

End User in Control

3

Enables New Use Cases

4

Faster Time to Solution

2 flexible - on-demand - user-friendly *) HPC = High Performance Computing

HPC as a Service

Success Criteria

Success Hardware Expertise Software

Microsoft & UberCloud have deep experience in CAE Applications and the Cloud UberCloud’s software packaging technology makes your CAE software available on Azure Microsoft’s focus is building the High Performance Enterprise Cloud Platform – Azure

For Use by Microsoft Sales

Customer-phasing fully automated self-service Azure/UberCloud CAE Cloud Platform

Azure UberCloud Solution

Azure High Performance Enterprise Cloud Platform H-Series N-Series

InfiniBand & GPU UberCloud’s software packaging technology makes CAE software available on Azure Microsoft’s focus is the High Performance Enterprise Cloud Platform – Azure

UberCloud Containers

For Use by Microsoft Sales

Key Benefits

Azure UberCloud: Works Like a Desktop, only Much Faster

1

Full desktop user interface for ease of use

2

Pre and post-processing on Azure

3

Flexible cluster sizing

4

Flexible licensing

• Desktop like user experience
• Ease of uses eliminate training needs
• Software is fully installed and ready to use,

including pre/post processing, meshing, solvers, Intel MPI for parallel processing

• Cluster sizes are flexible and can be scaled up

and down based on analysis requirements

• Shared storage can be sized based on needs
• GPU, Infiniband and SSD, support is available
• Hourly licensing (PoD) is supported.

Usability, Flexibility, Performance

For Use by Microsoft Sales

For Use by Microsoft Sales

For Azure sales:

How do I find out if my Microsoft customer needs or already uses HPC?

 Do you run engineering simulations? Just on workstations or also on HPC servers?  Which technical computing software do you currently use for your product simulations?  How many product design and simulation engineers do you have?  Do you have the in-house expertise to stand up and manage an HPC system?  What would be the impact to your business if you ran many simulations at once?  Or much finer geometries, or more detailed physics? If you got several ”Yes”, then ask: “Can we inform you about Azure Cloud? Are you interested in a demo?”

Contact your accounts in manufacturing, get in touch with your direct contact there, and ask the following questions:

 Is your company developing and selling products?  Does your company design and develop these products itself, in-house?  Do you have an R&D department (or an engineer) doing the product design (e.g. with CAD)? If you got 3 “Yes“, then ask your direct contact: “Please introduce me to the head of this R&D department.“

After you have identified the R&D department head or the design engineer, get in touch and ask the following questions:

For Use by Microsoft Sales

First High-Level Classification: HAVE’s vs HAVE NOT’s

• Owning an HPC cluster supported by an

HPC support team which

• HPC is part of enterprise IT services, or

part of the research / engineering team

• Many user can share this HPC system
• HPC system has a scheduler (Grid

Engine, PBS, LSF, Slurm)

• Schedulers can often automatically ‘burst’

into the cloud

• HAVE NOT simulation engineers have just

a desktop workstation

• Not owning nor access to an in-house

shared HPC cluster

• No HPC middleware tools, e.g. schedulers
• No central HPC support organization, no

HPC budget

• HAVE NOTS are not necessarily smaller
• rganizations (org size is not a criteria)
• Engineers have to manually burst to the

cloud

The HAVE‘s The HAVE NOT‘s All simulation engineers have a desktop workstation. Some have also access to HPC servers, but many others have not.

For Use by Microsoft Sales

Asking the Right Questions to HAVE’s and HAVE NOT’s

You are talking to a HAVE customer if he answers YES to these questions:

• Do you have access to a shared in-house

HPC server with several compute nodes?

• Do you have a centralized HPC support

team?

• Do you use scheduling / middleware tools

to manage your cluster & jobs?

• Have you automated HPC runs with most

workloads batch oriented? You are talking to a HAVE NOT customer if he answers YES to these questions:

• Your simulation engineers purchase or

manage their own workstations?

• Do individual departments make their own

computing purchase decisions?

• Are most of your users using simulation

software interactively, just on their desktop, through a graphical user interface?

All simulation engineers have a desktop workstation. Some have also access to HPC servers, but many others have not.

For Use by Microsoft Sales

Azure Value Proposition for HAVE’s and HAVE NOT’s

• Burst to the Azure cloud for additional

compute capacity

• Improved Total Cost of Ownership (TCO)
• Implement a Green Agenda
• Enhanced Security

Microsoft partners provide, in addition:

• HPC Expertise
• Proof of Concept and ROI
• Production-ready
• Expert engineering, Cloud, HPC Support
• Lower the barrier to HPC
• Increased collaboration
• Leverage flexibility in HPC platforms,

compute models and cost Microsoft partners provide, in addition:

• No need to learn anything new
• HPC Expertise
• Proof of Concept
• Production-ready
• Expert engineering, Cloud, HPC Support

The HAVE‘s The HAVE NOT‘s All simulation engineers have a desktop workstation. Some have also access to HPC servers, but many others have not.

Azure HPC Customer Profile Form

Company name, company size, department, group, contact person and position How many experts in the R&D department / group are doing simulations? Which major application software packages are used in your group / department? Software A: name, type and number of licenses (seat, floating, cloud) Software B: name, type and number of licenses (seat, floating, cloud) Software C: name, type and number of licenses (seat, floating, cloud) In-house developed software: name, type Which computing resources are you using (workstations, server, cloud)? Which are your major simulation challenges (computer is too slow, not enough memory, too many simulations (many parameter studies, design optimization), long wait queues in your server, etc...)? What would you like to improve (more software licenses, more compute power, bursting in the cloud for higher demands, . . .)? How can Azure/UberCloud help? After you have found out that your account is performing engineering simulations, and you have identified the right contact within this group please fill in the following HPC customer profile form and send it to the Azure Black Belt group email alias.

Additional Documentation for Azure HPC Customers

• Big Compute: HPC & Batch – Large-scale cloud computing power on demand:

https://azure.microsoft.com/en-us/solutions/big-compute/

• Engineering Simulation on Azure: https://simulation.azure.com/
• UberCloud-Azure integrated on-demand CAE offers on the Azure Marketplace:

https://azuremarketplace.microsoft.com/en-us/marketplace/apps?search=ubercloud&page=1

• Azure UberCloud case studies:

ABB: https://simulation.azure.com/casestudies/Team-182-ABB-UC-Final.pdf CFD Support: https://customers.microsoft.com/en-us/story/expanding-customers-computational-horizons HSR: https://www.hsr.ch/fileadmin/user_upload/iet.hsr.ch/Downloads/Team_189_Henrik_Wind_Turbine_final.pdf CAE Technology: https://customers.microsoft.com/en-us/story/cfd-setup-and-computational-time-cut-50-percent

• 80 case studies from CAE & Life Sciences, in UberCloud’s annual Compendium of Case Studies:

Compendium 2013, Compendium 2014, Compendium 2015, Compendium 2016.

Thank You

Wolfgang.Gentzsch@TheUberCloud.com

For Use by Microsoft Sales