INTERNET OF THINGS DATA
improve how they respond and identify
anomalies, and in-memory provides a
solution.
“We want to revolutionise data
access and processing and Grid Gain
is able to offer solutions that are
now being used by global enterprises
across a variety of sectors including:
nancial, software, ecommerce, retail,
online business services, healthcare,
telecom and others. However, I would
contend that it’s still very early days in
the adoption of this technology,” says
Ivanov.
Financial services
One market where in-memory is being
used is in nancial markets.
London-based Finastra is using
the company’s solutions to enable
nancial institutions to become more
resilient, more ef cient and more
competitive.
The Grid Gain solution is able to
connect data stores (SQL, NoSQL,
and Apache Hadoop) with cloudscale
applications and enables both
massive data throughput and ultralow
latencies across a scalable,
distributed cluster of commodity
servers.
“Our In-Memory Data Fabric
solution is a comprehensive,
enterprise-grade in-memory computing
platform for high volume transactions,
real-time analytics and hybrid
transactional/analytical processing,”
explains Ivanov.
“The main advantage of in-memory
is its performance. Reading and
writing data that is purely in memory
is much faster than data that’s stored
on a disk or on a ash drive.”
Finastra provides nancial
services software to retail banks and
is involved in transaction banking,
lending, and treasury and capital
markets.
Its decision to use in-memory
was driven by the need to ensure
that customer demands for realtime
services could be met and that
it would be able to satisfy evolving
compliance and reporting regulations.
Finastra implemented a Java-based
IT architecture to support the use
of data lakes instead of traditional
databases and in order to handle the
caching of data from the data lake and
distributing the cached data across a
network cluster for parallel processing,
turned to Grid Gain and in-memory
computing.
“In the context of increasing
needs for real time calculation and for
risk regulations, Grid Gain provided
a scalable and highly performant
platform for pricing and risk analytics
through its ef cient processing of
massive amounts of data,” explained
Benoît Riquet, Director Product
Management for Fusion Risk and
Pricing.
The Grid Gain in-memory computing
platform, based on Apache Ignite,
enables high performance transactions
that run up to 1,000x faster than diskbased
approaches.
It is able to provide high speed
transactions, real-time streaming
and fast analytics in a single,
comprehensive data access and
processing layer that is able to work
with any common RDBMS, NoSQL or
Hadoop database.
Grid Gain was also able to offer
ACID – Atomicity, Consistency,
Isolation and Durability -
transaction guarantees and it was
ANSI SQL-99 compliant.
The solution is able to
power existing as well as new
applications in a distributed,
massively parallel architecture on
affordable, commodity hardware,
which can then be scaled by adding
more nodes to the Grid Gain cluster
Prior to implementing Grid Gain,
Finastra relied on batch processing
of large amounts of transaction data,
which resulted in long delays before
the data was available for querying.
Now it is able, in real-time, to
process and compute with low
latency by storing the data in memory
and parallelising processing across
multiple machines in the cluster.
A typical Finastra cluster consists
of commodity servers relying on Xeon
processors with 256GB of RAM.
The GridGain platform,
FusionFabric.cloud, was able to
integrate Finastra’s trading systems
with cloud-based components to
offer a business-wide, cross-silo
approach to handling OTC derivatives,
exchange-traded derivatives, in ation,
xed income, FX/MM, hybrids, and
structured products from trading
through to accounting as well as
provide computational elasticity and
real-time risk with full valuation.
“All of which gives a massive
competitive advantage when pricing
complex strategies, or handling big
swings in volumes,” explains Ivanov.
With regulatory requirements
continuing to evolve, banks are
under pressure to improve data
management and performance.
In-memory solutions enable
companies like Finastra to also
document and store trade and
market data in memory and run
the calculations in parallel,
to dramatically reduce
processing time.
While the advantages of inmemory
are signi cant it’s not
without its problems and operational
costs can still be seen by some as
its biggest downside, but while it
might not be suitable for all cases
there is an increasingly strong
business case to use in-memory
technology.
30 10 September 2019 www.newelectronics.co.uk
/www.newelectronics.co.uk