Synchronization of Servers

[Gavishti]

Is there a current estimate (that you can share) of when you will complete the synchronization of the data servers? Just ran four tests on the same efs on the same time interval for the same day and and got four different results. Not a lot, but not the same.

Thanks.

[AndyS]

Hi Gavishti,

We are about 80% complete on the roll out of the synchronization to the server farm. We expect to complete the roll out by early next week.

If you're still experimenting problems then I believe it's due to you being connected to one that is not yet upgraded. You could try logging out then back in to reconnect to a different server and hopefully, one that's been upgraded. Beyond that, next week, you should be good to go.

Regards,

[Gavishti]

Thanks

[ScottJ]

Update:

The server synchronization process is now complete . All servers within our farm will now synchronize to a single time source each minute of the day. We believe this will improve the consistency of data pulled from different tick servers and on different PC's.

Please keep in mind that your PC clock can drift and since you do cache data locally for active charts, slight differences could still appear from time to time when you refresh the data from our servers.

Thank you all for your patience while we completed this lengthy project.

Thanks.

[MartinWood]

slightly changing tick volume for forex

Hi All,

Just been doing some data testing with forex data. I have been re-importing EUR A0-FX data to check if any of the values change after re-connecting to eSignal and then disconnecting. See the table below:

Symbol EUR A0-FX
Timezone GMT (London)
Interval 180min

Date/Time Close High Low Volume (number of ticks)
26/02/2004 15:00 1.2445 1.2468 1.2413 15,150
26/02/2004 15:00 1.2445 1.2468 1.2413 15,187
26/02/2004 15:00 1.2445 1.2468 1.2413 15,187
26/02/2004 15:00 1.2445 1.2468 1.2413 15,187
26/02/2004 15:00 1.2445 1.2468 1.2413 15,187
26/02/2004 15:00 1.2445 1.2468 1.2413 15,428
26/02/2004 15:00 1.2445 1.2468 1.2413 15,491
26/02/2004 15:00 1.2445 1.2468 1.2413 15,491
26/02/2004 15:00 1.2445 1.2468 1.2413 15,428
26/02/2004 15:00 1.2445 1.2468 1.2413 15,428
26/02/2004 15:00 1.2445 1.2468 1.2413 15,238
26/02/2004 15:00 1.2445 1.2468 1.2413 15,323
26/02/2004 15:00 1.2445 1.2468 1.2413 15,491
26/02/2004 15:00 1.2445 1.2468 1.2413 15,187
26/02/2004 15:00 1.2445 1.2468 1.2413 15,487
26/02/2004 15:00 1.2445 1.2468 1.2413 15,454
26/02/2004 15:00 1.2445 1.2468 1.2413 15,187
26/02/2004 15:00 1.2445 1.2468 1.2413 15,454
26/02/2004 15:00 1.2445 1.2468 1.2413 15,491


As you can see the close, high and low don't change - this is excellent. Also the tick volume doesn't really change that much, however I was wondering if they should show the same value every time.

If not then I would like to know if taking the maximum value recorded would be the best value for scientific purposes. I am assuming that the server can't count a tick that doesn't exist, therefore more information is captured.

Cheers,

Martin

[ScottJ]

Although the synchronization is complete, it doesn't mean that all retrievals will be guaranteed equal. There are still some subtle timing issues ( including possible drift on your pc clock ) that could move ticks from one time period to another. That makes it also possible for a high, low or last to be potentially off if those values were associated with one of the ticks moved to another time frame. So far, we're seeing vast improvement with the new synch'ing.

Without dumping each tick on each server and directly comparing each individual difference, I can't make an assumption that the retrievals with more ticks should be considered preferred or not. The differences might simply be associated with what I described above.

Thanks.