David Weiss
SVP, Head of Connectivity
Intercontinental Exchange|NYSE/NYSE ARCA
11 Wall St. NY, NY 10005
desk-212-656-6974
cell-203-722-1263
Twitter= @daveweiss68
|
Question Answer |
Email Text |
|
Mon, Apr 7, 2014 at 9:14 AM, From Dave Weiss:
I saw your linked in message that you don’t support banning co-location. Do you support keeping it the way it is or modifying it in some way ? |
|
| Mon, Apr 7, 2014 at 9:17 AM, From Eric Hunsader I support exchanges using the SIP for routing - as per Reg NMS. I'm not sure how/when/why that was allowed to change. Specifically, this Nasdaq filing and subsequent SEC rubber stamping: http://www.nanex.net/Research/RegNMSRecinded/RegNMSRecinded.html |
|
|
Tue, Apr 8, 2014 at 9:07 AM, From Dave Weiss: I looked at your tweet this AM. Your idea of a fix for this would be to send the original exchange timestamp, correct ? right now the only timestamp is from the SIP, we don’t send a timestamp from the exchange to the SIP. For 603c purpose we do ensure we send to the SIP before we send to direct feed. We have a surveillance tool that monitors it. | |
|
Q1 |
Tue, Apr 8, 2014 at 9:45 AM, From Eric Hunsader Do you monitor when it leaves your outbound port, or when it arrives at the SIP? |
|
A1 |
Tue, Apr 8, 2014 at 9:46 AM, From Dave Weiss When it leaves our market data processor. |
|
Q2 |
Tue, Apr 8, 2014 at 9:48 AM, From Eric Hunsader What is the distance between the market data processor and the SIP? |
|
A2 |
Tue Apr 8, 2014 at 12:03 PM, From Dave Weiss UQ/UT is in carteret |
|
Q3 |
Tue, Apr 8, 2014 at 12:07 PM, From Eric Hunsader Yes, I may have mentioned earlier that NYSE quote in NYSE stocks (TAPE A) aren't going to have the same issue as Nasdaq Quotes in the same stocks. Which is what the data shows. My understanding is that within your facility, co-located computers all sit behind a 1000 foot coil - does the SIP for Tape A also sit behind a 1000 foot coil? |
|
A3 |
Tue, Apr 8, 2014 at 12:11 PM, From Dave Weiss Yes. |
|
Q4 Q5 |
Tue, Apr 8, 2014 at 12:16 PM, From Eric Hunsader TCP or UDP? 1 GBps or 40 Gbps? |
|
A4 |
Tue, Apr 8, 2014 at 12:25 PM, From Dave Weiss TCP input, multicast output |
|
(repeat) Q5 Q6 |
Tue, Apr 8, 2014 at 12:28 PM, From Eric Hunsader 40 GBps or 1 GBPs? Do all inbound exchanges use the same speed/protocols? |
|
(not) a 5 A6 |
Tue, Apr 8, 2014 at 12:29 PM, From Dave
Weiss Getting the answer on 1 vs 40. All incoming use same speed and protocol. |
| Q7 | Thu, April 10, 2014 at 9:43 AM, From Eric Hunsader Subject: Another SIP Question Is the TCP option "disable Nagle" part of the SIP input specification, or is the default to use TCP as is? Thanks! |
|
A7 |
Thu, April 10, 2014 at 9:46 AM, From Dave
Weiss Its default. Can you point me to the 603c page that describes sending data to SIP vs. Direct ? Remember, we do send to SIP before we send to private and we are required to provide proof to the SEC that we do. |
|
Q8 |
Thu, April 10, 2014 at 10:14 AM, From Eric
Hunsader How does a NY-ARCA quote on a Nasdaq listed stock get to the SIP at Carteret 40 miles away, at before or at the same time as direct feed data gets to a colocated user 1000 feet away? |
|
(not) a8 |
Thu, April 10, 2014 at 10:31 AM, From Dave
Weiss Proximity is a constant. If we didn’t colo and didn’t have direct feeds someone would still get NMS feed first. My point is, what is the fix. Again, I am not pushing an agenda, I, like you, am thinking this through. |
|
Q9 Q10 Q11 |
Thu, April 10, 2014 at 10:57 AM, From Eric
Hunsader I'm not sure what you mean by "proximity is a constant". Does the exchange own the direct feed cables all the way to the subscriber, or just to the outbound port? How did the 1000 foot coiling practice come about - was that SEC mandated? Does Nasdaq and other exchanges also use a 1000 foot coil for direct feed subscribers? |
| avoided 10 A9 A11 |
Thu, April 10, 2014 at 10:59 AM, From Dave
Weiss Outbound port. We don’t own the line to the subscriber. What I meant by that is someone will always be closer regardless of who they are. Not sure if Nasdaq does the cables that way |
|
Q12 |
Thu, April 10, 2014 at 11:19 AM, From
Eric Hunsader (sent twice) If the subscriber's responsibility begins at the outbound port, what's to prevent them from using a shorter cable? |
| Thu, April 10, 2014 at 11:19 AM, From Dave
Weiss I am researching this |