Big Data Architecture

USP BOA

(BigData Oriented Architecture) :
A Software Appliance dedicated to a Big Data Processing based on the Unicage Development Method.

OUR BIG DATA ARCHITECTURE

Feature Hadoop Appliance Devices USP Boa Description
OSS Based X X Runs under any UNIX-based OS (Linux/Free BSD)
Inexpensive HW X X Runs on entry-class servers
Scale Out Architecture X X Can be scaled out similar to Hadoop
Can increase processing nodes dynamically based on load
High Speed Performance X X X High-speed commands (usp Tukubai) and distributed processing
Can be used for front-end processing for legacy DB apps and Hadoop
Maintenance and Support X X Can be supported by anyone savvy in OS kernel technology
Stability X X Only uses the features of UNIX that have been stable for 40 years

USB BOA PERFORMANCE

Proceed Task Speed
1. Select (para-grep) Select all records starting with the text “123” from among 1 billion records 3 secs.
2. Sort (clust-qsort) Sort 1 billion random in ascending order 97 secs.
3. Sum (clust-sm2) Sum key fields in 1 billion random records 35 secs.
4. Mathematical Operations (clust-awk) Perform mathematical calculations between fields on 1 billion records 22 secs.
(clust-lcalc) Perform precision floating-point operations 67 secs.
5. Join (clust-join1) Perform a join operation on 1 billion 37 secs.
6. Complicated Operations (clust-shell) Distribute 1 billion records by key block units and perform several calculations (keys sumup, average, round literal) 17 secs.
7. Big Data Select (apli-select) Perform a matching select on 10,000 transactions (join and exclude) from among 10 billion records 4.5 secs.
8. Big Data Update (Add & Change, Delente, Sum) Apli-update Apli-delete Apli-sumup Update (add & change), delete and sum 10,000 transactions from among 10 billion records 5.5 secs
9. Big Data Search (apli-search) Search account holder data based on Rank, Gender, Geographical Region, Age Group, Length of Membership and Minimum Average Score among 10 billion records 1.2 secs.

USP BOA DEMO ENVIRONMENT

scheme

USP BOA DEMO ENVIRONMENT

hen large amounts of data processing are required in the shell script, we use the commands in the “clust” and “para” families.

scheme