Hello,

I am new to IQFeed. I have written my own software in .Net to interface with IQConnect via TCPIP. I added a Level1 feed to my current trading infrastructure in a limited fashion. I quickly ran into problems. Although I don’t have direct evidence to support this, I suspect I received an unsolicited “S,SERVER CONNECTED”. By unsolicited, I mean that I did not send the “S,CONNECT” right before I think I received the connected message.

I haven’t found documentation related to unexpected disconnects from NTD servers. Unfortunately, I am unable to use the “pull the plug” method to test the disconnect behavior of IQConnect. My infrastructure runs in the cloud.

I’m looking for any insight into the following.

1. How does IQConnect decide when a connection to DTN servers has failed? It there a timeout on “heartbeats”? (say my internet connection is down)

2. When a connection is thought to have failed, will IQConnect try indefinitely to reconnect?

3. If the feed reconnects, will current watches be maintained? Will I start seeing Level1 data arrive without resubscribing?

4. On a derivative feed, will missing interval bars be replayed? ( I think not because I believe IQConnect computes these locally from tick data, but I could be wrong)

5. Is there anyway to get IQConnectLog to append instead of overwrite?

---

MAC

More...

6. Is there a recommended maximum duration I can run IQConnect without a restart? Is 24/5 ok? or should I restart daily?

---

MAC

I was able to block inbound traffic from DTN (IQFeed servers) on my cloud instance. I have some answers to my questions.

1. IQConnect will determine there is a connection problem if nothing is received for 5 seconds. I guess this is why they send frequent (1s) TimeStamps.

2. After 10 minutes IQConnect was still trying to reconnect. Although this is not indefinite, it indicates IQConnect will keep trying to reconnect.

3. I was not able to fully test since markets are closed due to the holiday but IQConnect still shows the correct number of watched symbols. I will need to test on Monday.

4. TBD on Monday

---

MAC

1. How does IQConnect decide when a connection to DTN servers has failed? It there a timeout on “heartbeats”? (say my internet connection is down)

You are correct. It tests for 5s gap in data.

2. When a connection is thought to have failed, will IQConnect try indefinitely to reconnect?

Yes, it will try indefinitely.

3. If the feed reconnects, will current watches be maintained? Will I start seeing Level1 data arrive without resubscribing?

Yes,

4. On a derivative feed, will missing interval bars be replayed? ( I think not because I believe IQConnect computes these locally from tick data, but I could be wrong)

No. The system, after retrieving the history, switches over to streaming data and builds it as you mentioned. But, that would add a lot of value to the product I think. I'll write it up for a future enhancement request.

5. Is there anyway to get IQConnectLog to append instead of overwrite? Not currently. We are looking at options to improve our logging currently. Hopefully, we can make this configurable in future versions.

6. Is there a recommended maximum duration I can run IQConnect without a restart? Is 24/5 ok? or should I restart daily?

Some people leave IQFeed up for weeks at a time, but the other day I saw a 17GB log file, so I would say you might want to at least have a script that either copies the log into another file or deletes it every once in awhile so the log will remain useful.

Thanks for info. I have another question.

Can you give any information about the last know service interruption IQFeed had?

When it occurred?

How long it lasted?

Which exchanges were affected?

In general, how often do you see service interruptions which are not the fault of the exchange?

---

MAC

I actually don't recall the last full outage we had, I started 11 years ago, so I asked if anyone remembered one and someone said it was 16 yrs ago that we had our last outage where we we had 0 data in, 0 data out. But, that was before we had our own ticker plant, so it doesn't have any relevance to today.

We have always said we maintain a 99.9% up time, but I am sure we exceed that, even at an individual exchange or server level. Of course, there could be street maintenance somewhere that cuts a fiber line, but that would impact only that given exchange and only affect one server center as major exchanges have multiple feed inputs and we would just redirect users to good servers during such an event.

We have times where we roll out updates over the weekend which could result in us having to rollback. That might happen a couple times a year. but we stagger our releases, so all systems are not affected by any one update. We also have engineered many of our systems so that they can be updated without ever bringing anything down or losing data, so an event like this may be invisible to you.

So yes, things happen, but I would say this is one area we really excel at and will continue to focus on as the product evolves. Sorry for the late reply, but hopefully this helps answer the questions.

Tim