HPC Cloud Friendliness Tool

This tool is designed to help engineers, IT professionals, HPC users and buyers, and business executives understand the extent to which their workloads are amenable to operation on a public cloud. The tool is broken down into a series of questions that delve into the specific characteristics of a particular workload, allowing a user to assign a numeric score for each, rated on a scale of 0 to 10

For each characteristic, a value of 10 means it would be better suited to a public cloud, 5 mean no clear advantage on either platform, and a 0 would be better suited to an on-premise system. Each characteristic includes a short description of what a 10, 5, and 0 value entails.

The purpose of this tool is not to promote either option, but rather to provide the user with an un-biased assessment of where their application could be best operationally suited. A high score at the end of the assessment means that the application is friendly to the public clouds, albeit with on-premise systems still being a viable option. Likewise, an overall application score that leans towards the on-premise option, does not mean the application cannot go to the public clouds, but rather that a public clouds may not be the best suited option for that specific type of workload today.

After you have filled in answers for all 13 characteristics, you will be presented with your score in a printable scorecard.

HPC Cloud Friendliness Tool

This tool is designed to help determine the extent to which your applications are cloud friendly

First Name
Last Name
Email
Company Name
Job Title (Optional)
Country
Industry
What is the name of the application that you are evaluating:

HPC Cloud Friendliness Tool

This tool is designed to help determine the extent to which your applications are cloud friendly

Attribute: Application Peaks

a . Is your application fairly consistent with peaks less than 20% (0)?

b . Does your application have some significant peaks, but only on a sporadic basis (5)?

c . Does your application have major peaks that are 50% or more (10)?

Attribute: Workload variability

a . Is your application generally consistent in terms of processor counts, memory, storage, and data transfer requirements (0)?

b . Is your application variability in terms of processor counts, memory, storage, and data transfer requirements large but only for some small set of jobs (5)?

c . Is your application highly variable in terms of processor counts, memory, storage, and data transfer requirements (10)?

Attribute: Embarrassingly parallel

a . Is interprocessor communication a key performance determinant (0)?

b . Is your application a mix of tightly and loosely coupled elements (5)?

c . Is the application embarrassingly parallel or has many highly independent parts (10)?

Attribute: Application job size

a . Is the application large, e.g. each application uses 100s or 1,000 plus of nodes, and/or requires the memory space across 100s or more nodes as measured by CPU counts and memory size (0)?

b . Is your application run as a mix of small and large jobs (5)?

c . Is the application small, e.g. can run on a single node, using the memory of one node or just a few nodes (10)?

Attribute: Data set size

a . Is your data in the petabyte range (0)?

b . Is you data in the 100’s of gigabyte range (5)?

c . Is your data in the gigabyte range (10)?

Attribute: Data security

a . Do you have strong on-prem data requirements due to data protection regulations or security/privacy compliance issues (0)?

b . Do you have a limited number of jobs that require on-prem hosting due to data protection regulations or security/privacy compliance issues (5)?

c . Do you have few or no on-prem data protection regulations or compliance requirements or do clouds offer a sufficient solution for your legal and data security needs for this application (10)?

Attribute: Data movement

a . Do you have large data sets with high data movement across nodes during processing (0)?

b . Do you have a mix of large and small data sets with varied data movement requirements (5) ?

c . Do you have smaller data sets or larger data sets with lower amounts of overall data movement during processing (10)?

Attribute: Special hardware

a . Do you already have the required special hardware that this application needs on-prem (0)?

b . Do you need a mix of special and general-purpose hardware to support your application (5) ?

c . Do you need access to special hardware (e.g. GPUs, new CPU types, special interconnects) that you do not currently have on-prem (10)?

Attribute: Processor speed

a . Does this application need specialized, high speed, high-end SKU processors (0)?

b . Does this application need a mix of specialized and standard processors (5)?

c . Does this application work okay with standard processors (10)?

Attribute: Interconnects

a . Does your application need specialized high bandwidth, low latency interconnects (0)?

b . Does your application require a mix specialized and standard interconnects (5)?

c . Does your application work well on standard interconnects (10)?

Attribute: Software Stack

a . Does your application require custom or special software stacks or separately purchased expensive software (0)?

b . Does your application call for both special and standard software stacks (5)?

c . Does your application use standard software or a range of preset software stacks or software that can be purchased on a pay-as-go basis (10)?

Attribute: Virtualization

a . Does your application run significantly slower in a virtualized system (0)?

b . Does your application run effectually on either visualized or bare-metal configurations (5)?

c . Does your application run with little penalty over bare metal when virtualized or are there benefits of running in a virtualized environment that help this application (10)?

Attribute: CAPEX & OPEX

a . Is reducing operating expenses (OPEX) an important budget constraint for this application (0)?

b . Is reducing both operating expense and initial large investments the most important budget constraint (5)?

c . Is reducing initial large investments (CAPEX) an important budget constraint (10)?

Attribute: Application move

A . Have you already moved this application to a public cloud?

Yes
No

Question 1 of 13

Hyperion Research HPC Cloud Friendliness Assessment Tool Results

Application name:
Date:
The overall HPC cloud friendliness score for your application is:
Detailed HPC Cloud Friendliness Scores:
1 . Application Peaks:
2 . Workload variability:
3 . Embarrassingly parallel:
4 . Application job size:
5 . Data set size:
6 . Data security:
7 . Data movement:
8 . Special hardware:
9 . Processor speed:
10 . Interconnects:
11 . Software Stack:
12 . Virtualization:
13 . CAPEX & OPEX:
14 . Application move:

If you have any questions, please email anorton@hyperionres.com