Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Pattern Replication for Full Software Stack Workload Test Exercisers

IP.com Disclosure Number: IPCOM000247689D
Publication Date: 2016-Sep-27
Document File: 5 page(s) / 232K

Publishing Venue

The IP.com Prior Art Database

Abstract

A method to determine, encode, validate, and deploy hardware usage patterns via low level test exercisers that are equivalent to a full software stack, e.g. for emerging workloads like cloud, analytics, mobile, social, and security is disclosed.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 34% of the total text.

Page 01 of 5

Pattern Replication for Full Software Stack Workload Test Exercisers

Disclosed is a method to determine, encode, validate, and deploy hardware usage patterns via low level test exercisers that are equivalent to a full software stack, e.g. for emerging workloads like cloud, analytics, mobile, social, and security. New age

workload patterns are used to test modern computer system hardware. The workload patterns are deconstructed to their basic components and then reconstructed into low level hardware exercisers to easily test modern workloads like cloud, analytics, mobile, and social.

The IT industry is going through a huge transformation and due to this transformation, there are many new use cases for computer servers. Newer applications are also creating new test challenges for computer server platform testing. Server system test professionals have to find new efficient ways to test servers for new age workloads. These new age workloads are typically presenting 4 to 5 layers deep software before touching hardware and that includes OS, database, middleware and multiple layer of high level SW. The disclosed method identifies new age workload patterns and leverages low level hardware exercisers to mimic the exact same behavior. Usage patterns of computer server units like compute, storage and networking are monitored, parameterized, and the results are captured for a duration. On a typical social platform, video sharing and viewing is a common feature. Videos hosted on social platform will

be of varying sizes. So a storage server that is hosting videos will move variety length files. The videos will typically be cached and then burst out to users via a network. This is similar to some popular video post going viral on a social platform, what does it

do to the platform is what we would want a program to mimic exactly. The number of video hits keeps increasing exponentially once it is uploaded on social platform. The idea here is to increase the number of image reads in exponential manner. There is an element of both storage and networking subsystems of a platform in running this

workload.

While a workload is running a monitoring tool would capture the following:

Number of threads driving storage & network operations.


1.

Range on storage seeing accessing operations.


2.

Random/sequential storage operations control count


3.

Storage and networking data transfer sizes


4.

A low level storage and networking exerciser would need capabilities to control the

same list of parameters. A low level exerciser creates threads that perform write

operations on a given range of storage along with network receive operations. This will depict the uploading of video. Later exercisers will be doing mostly read operations targeting a particular range of storage. Exercisers would increase the number of threads performing reads to mimic viral behavior. A networking exerciser would

perform send operations in tandem with the same parameters. Overall, the storag...