The New York Stock Exchange (NYSE), which trades shares
in Europe, the US and Asia, is using off-the-shelf technology to
cut the time taken to access business-critical data on its network
from as long as 26 hours to just over two minutes.
NYSE, which runs the Paris stock exchange Euronext, has replaced
an Oracle IO relational database with a datawarehousing appliance
from Netezza, allowing it to conduct rapid searches of 650
terrabytes of data.
The move forms part of a wider
trend for financial services firms to move away from complex legacy
technology towards off-the-shelf products, which gives
businesses greater flexibility to meet changing market demands.
The NYSE technology - which was installed in December -
collects, stores and manages trading data, collected by its New
York exchange, its Paris operation Euronext and its Asian online
exchange Archipelago.
The new appliances combine a database search engine and storage
in the same unit and use multiple processors in parallel to find
the data quickly.
Brian Clark, chief architect at NYSE, said, "It was painful to
put data in and get it out with the previous system."
The previous system had to send large quanities of data from
storage devices to the database over the network, and trawl through
large amounts of irrelevant information to complete searches.
Clark said the appliances, which carry out server, storage and
database management activities, will pay for themselves on the
storage benefits alone. "The financial analysis we did was just on
storage volume and we were able to get a return on investment in
one year," he said.
He said the Oracle system used up too much storage capacity
because IT staff had to build indexes that were used to help search
for information, which meant some data was repeated.
Ralph Silva, analyst at TowerGroup, said if the exchange can
source data faster it can cut the costs associated with answering
questions from the regulators. "If you have good data [management]
you can answer the regulator's questions in seconds without the
need to put three or four people on the job," he said.
Silva said that NYSE will also be able to sell more data. "The
big market for data currently is from the algorithm writers who use
it to test algorithms against [historical marker conditions] to see
if they work."