Latency Measurement: Impact of the FIX IPL Standard
TS-Associates’ Henry Young, Co-chair of the FIX IPL Working Group, discusses the anticipated impact of the new FIX Inter Party Latency (FIX IPL) standard.
The FIX Inter Party Latency (FIX IPL) standard, version 1.0, will hit the streets shortly after this issue has gone to press. Now that all the hard work of designing, formulating and testing the standard has been completed, thoughts turn naturally to issues of adoption and impact on the market for latency monitoring solutions. But let’s first revisit the motivation for FIX IPL.
The 1.0 release of FIX IPL is designed to achieve two things:
• The standardisation of where latency is measured, and
• Interoperability between latency monitoring solutions.
The first point enables latency statistics published by different firms to be compared more meaningfully – ‘apples to apples’ style. The second point enables the latency monitoring solutions operated by different firms to be interconnected or ‘peered’, so that inter party latency can be measured without requiring each firm to operate a latency monitoring solution supplied by the same vendor.
The next point to consider is the likely adoption of such an interoperability standard. What’s in it for financial market participants, and for latency monitoring solution vendors? As participation in the FIX IPL Working Group has demonstrated, both constituents see advantages in such a standard existing and being widely adopted. This will make inter party latency monitoring easier, it will reduce costs for participants, lower the barriers to entry for new solution vendors, thereby creating a larger, broader and faster growing market for latency monitoring solutions. This will be to everybody’s advantage. Those who follow the proprietary route and fail to adopt the standard will be left out in the cold. We expect support for FIX IPL to become a standard tick box item in latency monitoring RFPs.
FIX IPL Architecture
The FIX IPL architecture has been designed to support latency monitoring for price and order flows, both FIX and non-FIX, in a wide variety of situations. This schematic shows a simple case that demonstrates the advantages of the FIX IPL interoperability standard. It shows an order flow between two trading parties A and B. Thesecould be either a buy-side client and a broker, or a broker and an exchange. The flow is monitored in each party’s domain using a FIX IPL Source, which observes each message in the flow, extracts some content from each message for unique identification purposes and associates a time stamp with each message observation. A FIX IPL Source transmits a series of observation messages, each of which contains the unique identification content from the observed message and an observation time stamp.
The FIX IPL messages generated by each FIX IPL Source are then brought together by a FIX IPL Sink, which could be operated by either of parties A or B, or by an entirely different party C, as shown in the schematic. The FIX IPL Sink then correlates the observation messages from each of the FIX IPL Sources, matching up the message observations at A and B, and calculating latency by subtracting the observation time stamps. The resulting per-message or per-order latency metrics can then be aggregated into time interval statistics describing the latency properties of the order flow.
There is, however, a more subtle aspect to what the FIX IPL standard will achieve. Let’s consider for a moment the components of a latency monitoring solution (see FIX IPL Architecture). In order to monitor latency, the minimum requirement is for two FIX IPL Sources to generate FIX IPL messages, and for one FIX IPL Sink to correlate the FIX IPL messages and thereby calculate latency. Standardisation of the communications between these components means not only that the two FIX IPL Sources may be of differing origins, but the FIX IPL Sink may be of a third origin. The FIX IPL standard therefore achieves not only horizontal decoupling between FIX IPL Sources, but also vertical decoupling between FIX IPL Sources and FIX IPL Sinks.
The implication of this vertical decoupling is that we can now decompose latency monitoring solutions into two layers – the generation of FIX IPL messages and the correlation/analysis of FIX IPL messages. This opens up the market to vendors who may wish to embed FIX IPL sources in their trading infrastructure components with no concern for correlation/analysis. Meanwhile, other vendors may wish to focus exclusively on correlation/ analysis of FIX IPL messages without needing to be concerned with their generation. The disciplines required for each layer are quite different. The former requires the use of specialist hardware techniques for precision time stamping and clock synchronisation. The latter is a purely software-based task with no dependency on specialist hardware.
In conclusion, the FIX IPL standard will herald an opening up and broadening of the market for latency monitoring tools. We expect to see an increasing number of solutions vendors offering more specialised products that will be interoperable. New entrants will probably focus on one of the two layers facilitated by the vertical decoupling between FIX IPL Sources and Sinks. FIX IPL message generation will become an embedded feature of trading system components, and specialist instrumentation solutions. FIX IPL message correlation/analysis will be open to a wider range of monitoring solution vendors.Some industry participants will probably feel that reaching the version 1.0 release point is the end of the story. Actually, it is only the beginning. Unleashed onto the open market as a non-proprietary standard that is free to all to use, the future adoption, use and evolution of the FIX IPL standard is impossible to anticipate with any certainty. But I think one thing is certain… it’s going to be a lot of fun watching how things play out over the coming years.
FIX Inter Party Latency (FIX IPL) Standard
The FIX IPL Working Group is part of the non-profit, global, FPL organisation and has over 190 members encompassing exchanges, brokers, funds, service providers and latency monitoring solution providers. The working group has brought together leading industry experts to standardise two specific aspects of latency measurement.
Firstly, the working group has developed a taxonomy of standardised measurement points.This is vitally important, so that when different organisations publish latency statistics that comply with the standard, financial market participants will know the data can be compared on a like-for-like basis. The FIX IPL taxonomy is shown in the schematic.
Secondly, the working group has developed a latency data interoperability protocol. This has been designed to enable interoperability between latency monitoring solutions, thus freeing the industry from unacceptable bilateral constraints on purchasing decisions. A widely adopted latency measurement interoperability protocol will, as with FIX, break down the barriers to the uptake of standardscompliant latency monitoring solutions.