Few things are transforming software, business, and nearly every aspect of our lives like cloud computing and software-as-a-service (SaaS). It’s easier than ever to build and deploy new applications, and the result is a plethora of new companies vying for our attention across nearly every domain. The cloud revolutionizes data collection and analysis and empowers SaaS, making it simple to distribute software to anyone, at any time, in nearly any location.
It’s no surprise, then, that the cloud is finding its way into trading systems as well. TT was an early adopter of cloud technologies in the trading space. We began talking about our next-generation trading platform in early 2014, but a year and a half later, and particularly with new entrants using similar terms, I think it’s important to ask: what does it mean to be a “cloud trading platform,” is there a difference between SaaS and “cloud,” and where does TT fit into the mix?
New technologies such as HTML5/Javascript make the next-generation TT platform accessible to users via the internet, mobile device or desktop client.
Whereas historically, trading systems were built and deployed in roughly the same manner, today there are a variety of approaches for building these systems. These varied approaches may not look very different in marketing materials since they all boast performance, ease of use and access to APIs, but understanding the differences is critical when making a decision about how to best allocate investment in trading infrastructure.
- First, there is the classic deployed trading system, which is still the norm in the industry. You allocate hardware in a data center. You deploy software to it. Every client application needs some kind of dedicated connectivity. Market data is handled in its raw form, and order routing is done as close to the matching engine as possible. These systems are costly to maintain but generally give you the ability to optimize every part of the system for performance.
- On the opposite end of the spectrum is the new “cloud trading platform.” These systems leverage web services accessible over the internet typically via a thin client (often web browsers). They may or may not be deployed literally on a cloud provider’s infrastructure since sometimes “cloud” just means “available via the public internet.” They simplify accessibility and will often boast that they are high-performance, but in reality, they don’t source raw market data and order-routing feeds and are therefore significantly slower than their classic, large-infrastructure predecessors.
At TT, we find both alternatives to be lacking. The cost to maintain your own infrastructure with a classic deployed trading system is too great, and as a technology vendor, having many discrete on-premises deployments makes it a challenge ensuring all of our end users have access to the latest markets and features we support in a timely fashion. Conversely, the compromise in performance, security, reliability and flexibility in “cloud trading platforms” is equally unacceptable. While the internet will continue to fundamentally improve the accessibility of our markets, it shouldn’t be the only connectivity option. So, we’ve decided to find a way to get the best of both worlds–to make no compromises.
We know that SaaS is the only way to truly scale a software business. At the same time, we know that many trading strategies are latency-sensitive and require that the user have control and customization that is not traditionally part of a SaaS offering. We know that the cloud also is not a fad, and yet these same latency sensitivities that pose a challenge to a SaaS platform make the cloud a less viable tool for system development and deployment.
TT is a hybrid solution that doesn’t make you compromise performance for accessibility, cost for customizability, control for scalability.
First and foremost, TT is a professional trading platform that delivers best-in-class performance. Our vanilla, out-of-the-box offering with no infrastructure charge gives you double-digit microsecond turnaround on complex trading strategies with integrated pre-trade risk controls that allow both a broker and a trader to get what they need. The back end is C and C++ on Linux and is colocated with exchange matching engines and market data feeds. We don’t rely on someone else to aggregate market data, nor do we have to route orders through someone else’s FIX infrastructure.
At the same time, by embracing new technologies (in particular HTML5/JavaScript), we are able to deliver the most accessible professional trading system ever built. When you’re done reading here, go straight to trade.tt, click “Create Account / Log In” and create your demo account. You can be trading in less than a minute. Alternatively, you can go to the App Store or Play Store and complete the same process to be trading from your mobile device. In mid-November, our desktop client also will be available, giving you a third option for how to access the platform. None of these require complex packaging or installation from your (or your broker’s) corporate IT department, and they give you the freedom and flexibility to interact with the trading system on your terms from wherever you are.
Never before has a trading system with the back-end power of TT been this easy to access.
So where does the cloud come into play in TT? Our approach is to not be married to a single technology stack but to always strive to choose the right tool for the task at hand. Our legacy is as a C++ shop, yet today we leverage C, C++, Scala, Java, JavaScript, Python, Objective-C and Lua. Where latency is essential, we are using C and C++ on Linux with industry best practices for reducing latency. Where scalability in the cloud is needed, we are using Scala, Java and Python. The front end leverages JavaScript in the browser and desktop, Objective-C on iOS and Java on Android. And Lua is used for inline scripting for a subset of applications where customizability is critical. The point is clearly that we look to use a variety of tools, allowing us to maximize the value in every part of the platform.
Specifically, the cloud is used in TT for a handful of functions, none of which are involved in latency-sensitive activity.
We leverage Cassandra and Scala in AWS for the “forever audit trail” and a global order book where every execution-related message for every user for all time is encrypted, stored and retrieved. This implementation leverages the best of what the cloud offers in terms of scalability and cost. We also use the cloud to store UI preferences and settings, the security master database and historical market data going back more than 10 years as well as to host the front-end HTML5 application.
None of these applications benefit from low-latency technologies, hardware acceleration or colocation. This hybrid cloud-colo implementation allows us to better scale and reduce the costs our users incur to use TT.
If all TT had done was take a standard trading back end and put a Javascript application in front of it, then we could say this was software-as-a-service, but it wouldn’t be transformational for the industry. What makes TT so unique is that along with this zero-overhead, zero-maintenance (you never have to install an upgrade!), high-performance back end, accessed easily from browser, desktop and mobile, there will soon be capabilities for customizing the platform to get the kind of performance, protection and scalability every professional trader and broker needs. Specifically, in Q1, we will begin rolling out support for:
- Dedicated/private network connectivity, removing the internet from the equation
- Dedicated servers, allowing brokers to eliminate the risks of multi-tenant/cross-contamination
- Custom deployment of colocated services, allowing for “nirvana” configurations and providing best-in-class algo/Autospreader® performance
We believe that cloud-based solutions are asking you to compromise performance for cost and accessibility. Deployed platforms force you to do the opposite.
TT asks you to stop compromising and start leveraging a platform designed to give you the maximum in performance, accessibility and cost.