High Performance Market Data Analytics

Introduction

Rapidly detecting degradation in the quality of data feeds is a considerable challenge for any market participants who use or transport real-time market feeds. Market data transport technology is primarily based around the use of multicast, which does not have any error correcting mechanisms at the network layer. This means that any packets containing key trade data that have been lost cannot be detected until they are passed through a feed handler system at the end user site.

This creates several challenges:

• Many feed handlers use feed arbitration (between A and B feeds) to autocorrect message drops, so application teams may neither be aware of problems nor quickly pass-on details to their network operations teams.

• Application teams may be aware of the problem, but the tools they use may not be available to the network operations teams who are responsible for diagnosing and resolving the issue.

• Detecting a problem at the feed handler does not tell you where the problem occurred. Was it a problem with the Exchange or Market Data Vendor’s ticker plant? Was it a problem in their internal network? Was there a problem in the third-party network carrier or extranet provider used to transfer the market data? Was there a problem with one or more of the upstream production switches? Or, was the problem in the end user’s internal network?

All these questions lead to slow decision-making, lengthy problem identification and remediation times that can be measured in days and not minutes. Much of the technology used to monitor this infrastructure today is either not up to the task or involves expensive data capture, storage, and analytics technologies that have remained the same for twenty years. These technologies are expensive and consume valuable power and real estate within expensive colocations and data centers.

Highlights:

• High Performance Multicast Gap Detection provides real-time gap detection from 4096 simultaneous multicast channels at 160 Gbps throughput to 16k channels at 1.6 Tbs 

• Advanced Latency Analytics measures nanosecond latency and jitter from an exchange to remote sites, among WAN colo sites and DCs, and within the same data center

• High Resolution Microburst Monitoring detects traffic spikes with 15.2 µs sampling at line rate • Time Stamp Insertion accurately (sub 200 ns accuracy to the external time reference) time stamps all packets and metadata at line rate using PTP time synchronization

• Channel Health monitors channel performance and alerts to anomalies, simplifying feed troubleshooting

• VLAN tag insertion and identification simplifies management and troubleshooting for data traversal paths across network segments

• Exchange Feeds are pre-programmed for easy set up and configuration, updated regularly

• Super Reliable with FPGAs, not dependent on error prone CPU-based expensive servers

• Ixia Fabric Controller (IFC) enables single pane-of-glass management of multiple TradeVisions or Vision-series NPBs

• Visual Dashboards and End-of-Day Reports simplify issue identification and remediation

• Supports both fan-in and fan-out to increase 10 / 40 port density

• TradeStream streaming metadata enables real-time high-resolution statistics and market