Just before Bats Trading's announcement today that it's expanding its hosting space in Savvis' Weehawken data center (called NJ2X), we sat down with Bats COO Chris Isaacson to hear about how the alternative exchange is growing — it's poised to launch a U.S. Equity Options Exchange and second U.S. Equities Exchange in early 2010 — and how he's planning to meet these increased demands on the IT side. Bats Exchange is the third largest equity exchange in the U.S.
WS&T: Will this Savvis arrangement help Bats grow its own presence in New York, or is it more geared toward your New York clients to let them be close to your matching engines?
Isaacson: It's purely for Bats' own needs, we have never sold colocation space to others, nor do we have any plans to do so. It's to handle the organic growth on the existing exchange in the U.S, as well as the planned launch of the Bats options exchange and second equity exchange in the first quarter of next year. We also plan to expand in Europe, there's organic growth happening there.
WS&T: What will be the difference between the second equity exchange and your existing equity exchange?
Isaacson: The primary difference will be pricing, which is to be determined; we just submitted our rules. But there will be pricing flexibility allowed on a second book. It will also give us a chance to try out new order types and be a grounds for us to try new things.
WS&T: With the SEC watching.
Isaacson: Absolutely. This is all well in front of the SEC, they know exactly what we're doing.
WS&T: Would you say Savvis is like a right arm to you in growing your IT infrastructure, or are they just a piece of it?
Isaacson: I would say that Savvis is a primary piece, they offer proximity hosting for us, both here in the U.S. in our primary site and in the U.K. They're a very important partner for us. But we're controlling our own infrastructure within the data center and ramping up accordingly to meet the strategic objectives planned for the first half of next year and to handle the continual growth of our membership base, which is currently at 435 members and growing every week. As members come on, new connections are added and the network continues to get bigger and bigger.
WS&T: Can you give me a sense of the size of your operation in Weehawken?
Isaacson: I can't disclose the actual footprint or any other terms of the deal. But we're multiplying our space there, we're not simply adding to it. Building an options exchange, handling the messaging rates both inbound and from our members as well as outbound requires significant hardware capacity.
WS&T: Can you tell me about any of the technology components you use, types of servers, routers, feed handlers?
Isaacson: I can understand why you would want to ask that question, but that's all proprietary information, our secret sauce and why we believe we're the fastest in the world.
WS&T: Do you build a lot of it yourself, such as your message handling software?
Isaacson: We don't outsource any of our software, so...
WS&T: And you don't buy packages off the shelf?
WS&T: Do you have any key technology partners other than Savvis?
Isaacson: Certainly, we deal with a whole slew of technology hardware vendors, the usual suspects on the Street. I won't disclose their names. We keep that tightly held as do most exchanges who own their own technology.
WS&T: Except the NYSE, which puts out press releases about the technology it uses every other day.
Isaacson: They have a different strategy. They're actually trying to sell software and technology, where we're using technology to run an exchange, to run order flow.
WS&T: Is Bats expanding into other countries besides the U.S. and the U.K.?
Isaacson: Our primary focus is in the U.S. and Europe. We've been live in Europe for about a year and we've captured 4% market share in Europe, as well as about 10% market share in the U.S. We plan to garner market share quickly in the options industry here next year. [In the announcement today, Bats said it will expand its footprint within the Savvis Docklands data center in London in preparation for continued growth of Bats Europe.] These are the focus of our attention right now. That's not to say we won't go into other markets as time and regulation allow. But we still are a small, nimble firm, less than 70 people in U.S., less than 100 people globally, but we're developing a footprint in the two largest capital markets in the world.
WS&T: Can you tell me what your goals are for your IT infrastructure, such as latency, throughput or scalability?
Isaacson: Number one, it has to be the most reliable in the U.S. and in Europe, the most competitive capital markets in the world. It's a very fragmented market and there are many choices, so if you're not reliable people won't use you. We're proud of our history of reliability. Our members appreciate that. And then speed, from the very beginning we built a platform that could handle incredibly high throughput at very low latency, even during the heaviest loads, especially when the Fed announces there's some major market move, which is when traders need performance the most. We believe the most important thing is performance under extreme load. And capacity needs to be very high. We've tested our system to 600,000 messages per second. Throughput is in our infrastructure, our architecture. It's simply a function of how much hardware you want to buy. We have projects under way to decrease latency and further improve throughput.
WS&T: Are you to some extent at the mercy of the local telecommunications providers in the New York area for getting information to and from the customers there?
Isaacson: It depends how the customers connect to us. Some are colocated within the Savvis facility and some are not. That's a similar story for the other exchanges. Our members are dependent on the carrier they use, the access method they use to get to BATS, we don't dictate that, we're carrier neutral. But once the order hits our walls, we have complete control over that and we give equal access and performance to every member. We put our latency statistics up on our website.
WS&T: Some people say there's no way to get reliable measurements of latency because there are so many points in a system where a bottleneck can occur, and even if you have multiple monitors set up, each computer tends to have a slightly differently calibrated clock so the time stamps can be mismatched.
Isaacson: We measure latency from the time an order hits our outside network until the response to that order, an order acknowledgment, leaves our walls. So from door to door through our entire network, software, hardware infrastructure and all the way out, the entry and exit is measured on the same device. Obviously, getting from the customer to us and from us to customer depends on their connectivity method to us. We believe we've been as transparent as possible in this regard. I don't see any abating of the desire for speed in the market.