Over the last few days (probably since the weekend upgrade) I have noticed that the 156 farm has been performing badly. Basically the number of updates received is well below expected. If I force a reconnect to the 148 farm and watch the same symbols the rate of updates increases dramatically. I have a attached a chart showing bandwidth usage when connected to each of the farms.

The problem does not appear to be network related as it occurs on multiple client machines in different countries and basic route traces show no problems (~55ms ping times to both farms).

Dennis

Attachment didn't seem to work so here's a link.

http://www.anywhich.com/156v148.jpg

Hello Dennis,

I am not showing similar results although it is probably due to not knowing exactly what to watch.

Do you have any more information that might help us figure this out? For example, did you notice a pattern to which data you weren't receiving? Do you have any more data you can share with us concerning this? I know that you watch NYSE and NASDAQ on separate machines, which machine was the bandwidth chart from? I will setup some tests for tomorrow to watch this more closely. I'm sure I have a copy fo the symbols you are watching from one of your previous reports but if you want to email me (to developer support) a current list, it would be appreciated.

I don't think I'm going to be of much help diagnosing this problem. I will send you our two watch lists via email.

It happens to both of our machines and we monitor the data rate on both. If the number of updates gets too low then we automatically reconnect to a new data server. Our software ensures that we alternate data farms when we reconnect For the example I posted our server was watching 1300 NYSE symbols and was connected to the 66.112.156.225 server. The data rate dropped to 11 updates/sec and we reconnected to the 66.112.148.114 server.

It doesn't appear to be symbol specific. We seem to get updates for all of our symbols (I say 'seem' because I am only sure about the symbols we've purchased which is typically 20-30 per day).

Dennis

We have located one possible source of this problem and it will be resolved over the weekend.

I will continue testing on monday.

I did some additional testing today and after the weekend update, I am no longer seeing a discrepancy between the two server farms.

Steve,

I have switched to using the 148 farm exclusively so I can't confirm if my observed performance issues are still there. I will switch back to using both farms for Tuesday and report back.

Sorry Steve, but I'm still seeing the same behaviour.

Dennis

I want to verify what I am seeing in this new screenshot.

You were connected to 148 initially
You reconnected to 156 and after the initial burst of summary and fundamental data, the messages "settled out" with a much lower rate than you had been experiencing on the 148 farm.
At this point, you reconnected back to the 148.

Correct?

Dennis, once again today I am not seeing the same results as you.

I ran the same symbols (your 1300 NYSE list) on 2 separate machines today using the current release version of IQFeed (4.2.1.4) and had the same overall output from both server farms.

At this point, I believe there has to be something in the route you are taking to the 156 server farm that is causing problems.

I checked on your connection to the 148 farm today and (as you might expect) everything seemed normal. I would like to do the same for you on the 156 server farm but that will require some collaboration between us. If at all possible, please contact me in chat sometime over the next couple days when you have a few minutes to test this.

Sorry Steve, I should have explained the chart.

I was actually initially connected to the 156 farm (very left of the chart)and everything was going okay. The data then became very choppy, sometimes falling to a very low rate and then spiking upwards (middle of the chart). I then forced a new connection twice (it chose another 156 server initially) and connected to a 148 server. The data became steady again (right of chart).

After I posted that chart we had some minor trouble with the 148 farm where data would stop completely for a few seconds. Unfortunately this triggered a reconnect to the 156 farm which failed completely! (Hundreds of watches failed, data feed almost non-existent). This happened several times throughout the day and the 156 farm failed in this way every time from then on.

Once again, network traces showed no problems with the network to either farm and using a different ISP didn't help either.

Steve,

I'm not trading on Friday so I'll get in touch (via chat) sometime in the morning after market open.

Dennis

We believe that our internet provider has identified and resolved this issue. Please let me know if these symptoms persist.