The most mature and effective co-location strategy employs a network-rich data center. In such installations, financial services firms are able to access dozens of carriers and extranets, leverage global market data and order flow sources and add substantial space, power, cooling and headroom.
The network-rich data center also helps manage overall cost and improve throughput, typically driving down latency through improved access and connectivity. Six-sigma uptime (99.999%) supports business resiliency, as well.
Comparing distributed and hub (co-location) models
To manage the growing complexity of the financial services ecosystem, many firms have pursued a model in which specialist services are distributed across a wide variety of network connections. In this model exchanges, data vendors, clearing firms, financial extranets, brokerdealers and others have operated as nodes on a far-flung network.
This approach is effective in supporting specialization within a firm, and network utility and system value of a financial network has increased as users and providers are added to the distributed ecosystem. Unfortunately, the nature of this global marketplace means that the need for high-volume, reliable interconnections have increased exponentially.
The distributed model fails as the number of endpoints increases. The distributed model has also struggled to maintain the capacity required to support algorithmic trading, among other demands. The result: bandwidth requirements are increasing rapidly, and infrastructure upgrades are trending toward continuous.
By comparison, a centralized financial ecosystem leverages economies of scope and scale for all participants. By co-locating within a network-rich data center, exchanges, data vendors, clearing firms, financial extranets, broker-dealers and others can dramatically reduce the number of circuits required to support multiple counterparties.
This approach helps financial services reduce and better manage costs while also improving throughput. Participants in a network-rich data center pay for one data connection into the data center, and counterparty transactions take place within the data center.
The centralized or hub model also improves throughput. The efficient, network-rich data center configuration reduces latency to its physical minimum. The shared internal connections support significant growth, so individual user bandwidth can increase without incurring any cost increases or penalties. The value of a shared centralized financial ecosystem is also seen in availability. As noted earlier, 99.999% uptime is the standard for these configurations.
Unquestionably, the financial industry will continue to scale, and its technology must increasingly do so in new, more predictable and more effective ways. By migrating from a distributed architecture to one that capitalizes on the “law of the hub,” financial services firms will be able to:
- Use dense interconnection points to reduce costs for circuits;
- Leverage economies of scale for all participants in the centralized network;
- Simplify connection needs by housing participants in a limited number of geographic locations; and
- Simplify network management by unifying connection standards in an environment in which throughput and uptime are critical measures.
