Page 213 - Building Big Data Applications

Use cases from industry vendors

Kinetica

USPS DEPLOYS KINETICA TO OPTIMIZE ITS
BUSINESS OPERATIONS

THE BUSINESS
largest logistic entity in the country, moving more individual items in four hours than the combination of UPS, FedEx,
and DHL move all year.

and storage platforms, and the adoption of Big Data techniques to improve the ability to respond to market dynamics,
enhance competitive practices, and accelerate the delivery process.
processing, tracking, and delivering mail on a limited budget.

THE CHALLENGE the use of graphical processing units (GPUs).
Kinetica is helping the USPS turn their data
So how does an organization that makes
into knowledge, enabling them to improve
daily deliveries to more than 154 million
addresses using several hundred thousand
For the first time in history, USPS is
able to see their entire mobile workforce
based on visual near-real-time data? The
United States Postal Service (USPS) turned in real time.
to Kinetica and its GPU-accelerated database KINETICA SOLUTION
-
The USPS runs Kinetica on a large cluster
FIGURE 1: A REGION SHOWING A HEAT-MAP overspending. composed of 150 to 200 nodes. Each node
OF DELIVERY POINTS WITH COLLECTION BOX Postal customers these days expect so- consists of a single X86 blade server from
ICONS REPRESENTING MAIL-DROP. phisticated services like just-in-time supplies, Hewlett-Packard Enterprise, half a terabyte
to a terabyte of RAM, and up to two NVIIDIA
tracking and delivery updates, and dynamic
Tesla K80 GPUs. The system went live in
shipment routing. The USPS knew they need-
2014, and was bolstered with a high availabil-
ed to improve their end-to-end business
process performance while reducing costs ity redundancy in November of that year.
With those 200,000+ USPS devices
entire workforce of USPS carriers with a de- emitting location once every minute, that
vice that emits its exact geographic location amounts to more than a quarter billion
every minute. Armed with this location data, events captured and analyzed daily, with sev-
the service aims to improve various aspects eral times that amount available in a trailing
of its massive operation, including improving window.
USPS’ parallel cluster is able to serve up
To fully optimize data usage, the USPS to 15,000 simultaneous sessions, providing
is rapidly replacing its traditional legacy the service’s managers and analysts with the
systems with revolutionary high-performance capability to instantly analyze their areas of
computing (HPC) solutions. An in-memory responsibility via dashboards and to query it
as if it were a relational database.
when that technology proved too costly and And Kinetica has been running since with
complex, the USPS looked to KInetica and 5 9’s uptime.
Used with permission from Kinetica, http://www.kinetica.com.