Powering Analytics, AI, and ML in Financial Services

JPMC electronic trading services operates in 60 colos globally that are private network clouds, all interconnected with essentially a private network of clients and exchanges. For greater insights, JPMC is creating data lakes that gather more data to mine for more correlations. It is also using machine learning to look for anomalies in millions of lines of log files. All of this requires low latency, where time is measured in single digit nanoseconds. When JPMC switched to Pure Storage FlashBlade®, it saw a double-digit increase in time sensitive KDB+ data analytics performance. And with FlashBlade, the firm can combine production and development in the same chassis for more efficient capacity utilization.

Data is doubling and tripling every single year, and to run AI/ML workloads, that data needs to be stored. In addition, to ensure it is hitting performance goals and surviving potential single points of failure, JPMC needs to inspect every transaction and build in redundancy. At the same time, JPMC has very sparse real estate and it's expensive, so the firm needs simplicity and the ability to easily scale without taking up massive rack space.

No one at JPMC needs a PhD to manage Pure’s storage. Its speed of trades meet customer expectations, and smaller footprint allows JPMC to expand to meet data growth. Efficient storage reduces rack space needed for 1.2PB to five rack units, with 15% to 20% uptick in performance for time sensitive KDB+ data analytics and support for data lakes used for discovery and priming algorithms.