Celoxica FPGA Ticker Tackles OPRA Feed Volume, Server Footprint

OPRA, which provides consolidated market data for 12 US options exchanges, is set for a substantial growth in message volumes later this year when the ISE Mercury and BATS EDGX Options markets go live.

In the past, to handle the sheer volume of data OPRA creates, firms were “throwing servers, infrastructure—and in turn, a lot of cost—at addressing the problem… and more and more servers have been put in place to handle the growing volume,” says Lee Staines, global head of sales at Celoxica. “In this climate, that’s not really viable anymore…. Today, firms are having to have multiple servers all listening to OPRA and just consuming parts of it across those multiple servers,” he says, adding that CTP can consume the OPRA feed on a single FPGA-accelerated server, compared to tens of traditionally configured servers.

“That’s an order of magnitude that allows firms to control costs pretty much immediately, but also to become future-proof because those volumes can be handled well into the future,” Staines adds. 

CTP uses FPGA cards to consume and decode the OPRA data directly into the servers in an efficient manner, Staines says. “We can consume all that market data and then distribute it over multicast to consuming applications very quickly, in low single-digit microseconds,” he says. “It’s also straightforward for consuming applications to use the market data, and also to filter the required market data  if needed to only those instruments that they want to consume…. Different consuming applications may need different sub-sets of data, and they can do that through the subscription functionality that we provide. They can request [only] the data that’s important to them and their trading applications.”

Staines says the challenges of handling the OPRA feed are reflective of broader industry challenges of dealing with growing volumes of data, where FPGA cards can also provide a balance between performance and helping firms reduce their costs and infrastructure overheads. “We’re excited about CTP for OPRA, but also across the other asset classes we support, because it takes us from the more specific trading use cases that we may have been seen to address—such as single-server trading strategy, market data and execution requirements—into the more enterprise-wide market data challenges, of which OPRA is a perfect example, and using the technology to be able to address those.”