Low latency messaging use is expanding beyond traditional algorithmic trading applications to address wider business requirements. Meanwhile, issues such as reliability, manageability and cost are becoming more visible. IntelligentTradingTechnology.com discussed low-latency messaging directions – and messaging directions in general – with Craig Betts, CEO of Solace Systems.
Q: You recently announced a deal with the London Stock Exchange. Can you tell us what that’s about?
A: The London Stock Exchange selected our technology for their downstream systems, which in exchange parlance means the applications and databases that receive information from the trading platform, such as billing, risk, and surveillance. These can be thought of as the middle and back office of the exchange.
Q: You’ve also recently deployed your appliances for market data distribution at a couple of Canadian exchanges. That’s another new application for your appliances. How easy is it to configure your appliances to take on these different kinds of messaging appliances?
A: We’re installed at a number of exchanges and other liquidity providers for a wide range of use cases, and a few have gone public with their deployment to demonstrate their technological leadership. LSE uses us for downstream systems as we just discussed, TMX Group and CNSX use us for market data distribution, and Liquidnet has standardized on our hardware for all their messaging.
It’s easy for us to address a wide range of use cases because we’ve built our appliances in such a way that companies can customise them to meet their own unique functional and capacity requirements. Developers use our unified API to specify what kind of messaging to use for each interaction, and we offer best-in-breed performance in getting it done. Best of all, our appliance is turnkey so you don’t have to be a rocket scientist to configure and operate a Solace messaging network – it’s more like running a network of IP switches or routers than dealing with the complexities of distributed messaging software.
Q: So you are tackling a different facet of low latency messaging, beyond the algorithmic trading applications where it began. How are the functional requirements similar, and different? And how low does it need to be?
A: Market data infrastructure for algorithmic trading continues to be where people are pushing the envelope in terms of lowest latencies, but we’ve seen the requirement for performance in the small numbers of microseconds moving into applications like smart order routing and pre-trade risk management.
As our recent customer announcements highlight, exchanges have also had to up the ante to compete with alternative sources of liquidity. Just a couple years ago, exchanges thought about performance in terms of many milliseconds, but today end-to-end trade matching takes less than a millisecond. Within that, messaging is in the tens of microseconds.
Exchanges differ most from algorithmic traders in the mission-critical nature of their business. If an algorithmic trading engine doesn’t like market conditions, it can exit open trades and sit on the sidelines. An exchange doesn’t have that luxury – it has to be fast and available 100% of the time. And where algorithmic traders continually seek market advantages and inefficiencies, it is incumbent on exchanges to ensure fairness across all market participants.
Q: And what about the requirement for low latency – is it still top of the checklist?
A: The vast majority of messaging requirements in financial services are not low latency at all, at least not in the way it is defined in the media or used in application planning. Most mid-office and back-office applications operate in the milliseconds without issue, and any interactions that include a long-distance WAN link will always take many milliseconds. For these applications, customers care much more about traditional IT issues like reliability, manageability and cost.
While low latency systems have the most sex appeal in the financial services market, it’s important to remember they represent a small fraction of the IT budget in most capital markets and banking firms.
Q: Do you see a general trend among firms wanting to implement an enterprise messaging bus, and why is coming from the specialist low-latency world a better solution than deploying the likes of MQ?
A: If you talk to the business, they want everything to be the best. Usually that means fastest or most reliable and all other factors are less important. If you talk to IT, of course they care about those things, but they also care about consolidation, cost and manageability. As such, IT is almost always the driver for an enterprise-wide messaging platform. And the bigger the company, the more out of control their operations have become from years of trying to satisfy the business with incompatible technology deployments. It’s not uncommon for financial firms to have thousands of servers running five or ten different kinds of messaging they’ve collected over the years. It really is a huge problem for operations.
Messaging appliances like ours have addressed the issue from both ends of the spectrum: we deliver best-of-breed performance for the business, and we bring cost and complexity way down for IT. The same way that large scale IP routers allowed consolidation of TCP, Appletalk, SNA and DECnet traffic into one network, messaging appliances are helping companies consolidate their application traffic.
As to your second question, MQ and low-latency messaging are apples and oranges. You simply can’t achieve MQ-grade reliability and manageability at enterprise scale with a low-latency architecture, and you can’t achieve very low latency with an approach optimised for MQ-style queuing.
We deliver our low latency messaging and guaranteed messaging capabilities as an integrated solution in a single chassis, but we do so with dedicated hardware and firmware that tackles these very different jobs in very different ways. There are many kinds of messaging for a reason – because the functional and operational needs are radically different from application to application.
Q: What’s going to be the focus on Solace – business and technology – in 2011?
A: From a technology perspective, we have innovation coming in many dimensions: we are continually improving our solution’s throughput and latency, we will expand the breadth of our offering to address new requirements, and for the lowest latency applications we’re making the middleware layer more network aware. We’ll also continue to unveil select alliances with companies that offer complementary technologies and services.
On the business side, we’ve approximately doubled our revenues for four straight years, and are looking for high growth to continue in 2011. In financial services, that means deepening existing relationships and continuing to deliver results for more investment banks, retail banks, exchanges, financial information providers, and hedge funds. Beyond finance, our efforts towards horizontal growth have been going well with recent wins in government, telecom, transportation and gaming. These markets don’t need microsecond latency, but they value reliability, simplicity and lower costs, all of which our appliances deliver in spades.