InteractiveBrokers charts delay in new Beta  [SOLVED]

Questions about MultiCharts and user contributed studies.
Nick
Posts: 496
Joined: 04 Aug 2006
Has thanked: 4 times
Been thanked: 24 times

Postby Nick » 19 Jul 2007

Technically there are several areas which can cause MC to be slower than other programs. At one time or other these have all been problems however one or even all may or may not be fixed now. To be honest i don't pay as much attention as I used to. The key ones are -

1) Incorrectly determining that bars are needed when they are already in the cache. (it should not be necessary to go to 'local mode' MC should know what data it has got and what it needs regardless of mode and without going online). To be honest this seems better than it was.

2) Sub optimal algorithm to request bars. This needs to be as fast as possible without violating IB's pacing algorithm. I know it doesn't help that IB does not publish documentation. I think perhaps this could do with improvement.

3) Processing and storing data - hard for a user to tell. At one time MC processed pretty small 'chunks' of data at a time. You really want to request the maximum bars that won't cause a pacing violation in one hit. I reported this (requesting just 1 or 2 bars at a time) and it was improved not sure if there is room for more.

4) Retrieving data from the data base, processing it, and rasterising it on the screen. Sometimes this seems pretty slow. Opening a new 5 minute chart and requested 20 days of history from the cache (as I already had a 1 minute chart that was up and running). As the chart was new no indicators or anything to calculate this can take between 2 and 10 seconds. Now I know a couple of seconds might not seem much but it is whole orders of magnitude slower than it should be. I know that displaying data in chunks will help with online retrieval but it may well mask a more fundamental problem that exists here. If you are doing testing and code profiling you might want to look at retrieving and displaying data from disc. Some software (Tickquest spring to mind) processes tens of thousands of bars in 'no time at all'.

2&4 look like obvious candidates for improvement.

Cheers.
Nick.

shcheuk
Posts: 8
Joined: 21 Jun 2007

Postby shcheuk » 19 Jul 2007

shcheuk,
1. Could you send us your workspaces?
2. Could you send us your log file from QuoteManager?

We are working on gradual data loading, but want to see why your problem is so severe.
The problem is not the way u download the data. It is u combined ploting data and downloading data together. That mean I cannot download the data without ploting it.

Here is the siutation:

1 stock have 20000 ticks (including trade/bid/ask) per days, my data provider can backfill up to 60 days, and I have 100 stocks, future, I have interested in 20000X60X100=120,000,000 data! I have to plot 120,000,000 data if I want to store them into the harddisk.if each stock 1,200,000 data~ 70Mb It will be 70MbX100=7GB, and I only got 2GB DDR Ram.How can I plot 7GB data into a chart? No matter how good your gradual data loading method is, there will be some errors, since the system nearly hang becuase of lack of memory.

If u enable data downloading without ploting, most user will have no problem on downloading or data ploting. Here is the reason:

Although most of the user have interest in ~50-100 stocks&future, we don't need them at the same time or at the same chart. We just look at it one by one. Looking at it 2-3minutes and then go to another one. We just need MC to plot ~2 symbols X 70Mb =140Mb Data in one time.

Let say 1 symbol is 70Mb, since I cannot download all the 100 symbols data in advance because of the 70MbX100 ploting Limit, we have to download it one by one at the time when we plot it. Everytime we look at one of the stocks. It then start download 70Mb data from the internet. Even if I plotted it last week, It will start download 70MbX7days/60days=8Mb data from the internet. no matter how fast your gradual data system is, it will take at least 10 minutes to finish and then we analyze it for 2-3 minutes, click another symbol, wait another 10 minutes, :evil: Everyone then accuse ur quote maanger blaming it for its extra low speed on downloading data. Simply because then cannot see the chart in a minute.

Let said I can download the data without opening a chart. I will instruct my programme to download 1000symbolsX60daysX20000ticks data ~ 70Gb first and store them into the harddisk. Then I will schdule it to download ~ 1GB new data everynight. Every time I pick one symbol, it will load the data from the harddisk up to yesterday and then download today new data in the internet.

Your problem is if I pick one symbol, it may not be the one I picked yesterday. It will start download the data begining from the date u opened the chart last time to the current day. It may be two weeks or even one months since u last open. Then It will be 14 times or even 30 times slower.

Don't develop a scanner before u can pre-download at least 100 stock data for your scan or the scanner will stuck at the middle.
Last edited by shcheuk on 19 Jul 2007, edited 1 time in total.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 19 Jul 2007

Billco,
Your test is correct and we reproduced the same issue. It is limitation of IB. You can make 60 requests during 10 minutes. 60 requests x 1440 seconds (we request 1 second data). So what you see it is inevitable. It is possible to use tricks. For instance use 5 seconds instead of 1 second as basis resolution. However it is obvious that charts will be different. We think it is better to use 1 second because it is most realistic. Price doesn't vary greatly over a second. We tried to construct data from 5 second data and it took 12 minutes. See atttached screenshot.
Attachments
343.PNG
(38.25 KiB) Downloaded 4788 times

billco
Posts: 15
Joined: 30 Nov 2006

  [SOLVED]

Postby billco » 20 Jul 2007

Billco,
Your test is correct and we reproduced the same issue. It is limitation of IB. You can make 60 requests during 10 minutes. 60 requests x 1440 seconds (we request 1 second data). So what you see it is inevitable. It is possible to use tricks. For instance use 5 seconds instead of 1 second as basis resolution. However it is obvious that charts will be different. We think it is better to use 1 second because it is most realistic. Price doesn't vary greatly over a second. We tried to construct data from 5 second data and it took 12 minutes. See atttached screenshot.

Thanks for the input. I compared MC with Ensign and QuoteTracker on 100 bars back, and the loading was indeed within a few seconds of each other, with QT being the fastest, then Ensign, and MC being the slowest. All were bare charts based on 500 volume bars. Obviously, there is still work to be done.

That said, there still remains the issue of starting the program each morning. This morning it took 18 minutes from opening MC until all the charts were loaded, 2-4 per workspace, four workspaces total, all 24 hour volume bars. Note these are unchanged workspace/charts that I use daily that usually run from 7:00AM EST to around 6:00PM EST. Both Ensign and QT will load the charts including backfill of missed data within 3-7 seconds of opening the workspace. Therefore, I either do not have something set correctly, as pooled data is obviously not being used, or MC is not designed to automatically pull from that data on start up.

I have no problem subscribing to a fee based data service, but having tried several over the years, I have concluded that the paid for data is no better than what IB provides, not to mention most only provide a few days of tick data. Given the numerous other postings about the slow loading of non-time based charts in MC, I am open for any suggestions you or others may be able provide.

Regards,

billco

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 23 Jul 2007

Bilco,
It is not correct to compare MultiCharts with Ensign, because it is a known fact that Ensign downloads data from DTN IQ servers. They are good and return quotes quite qucikly.
Regarding QuoteTracker. Please make sure it uses 24 hours sessions. It goes without saying that it is impossible to download 39,600 quotes (11 hours of night trading* 3600 seconds per hour) for 3-7 seconds. You can double check it.

I do confirm that MultiCharts works slower than QuoteTracker, but the difference is not significant.

billco
Posts: 15
Joined: 30 Nov 2006

Postby billco » 23 Jul 2007

Bilco,
It is not correct to compare MultiCharts with Ensign, because it is a known fact that Ensign downloads data from DTN IQ servers. They are good and return quotes quite qucikly.
Regarding QuoteTracker. Please make sure it uses 24 hours sessions. It goes without saying that it is impossible to download 39,600 quotes (11 hours of night trading* 3600 seconds per hour) for 3-7 seconds. You can double check it.

I do confirm that MultiCharts works slower than QuoteTracker, but the difference is not significant.

There is an option in Ensign to download IB only up to 20 hours. I used IB data only when making the comparison. Furthermore, I did note that the difference was slight although measurable. All charts were set to 24 hour, 100 bars back. The comparison was fair and honest.

As for MC starting time, this is and was separate from the testing information I gave. QT will indeed load up each morning in mere seconds, 24 hour charts, as will Ensign with global refresh disabled (uses pooled data to give fair comparison). MC takes 30-45 minutes to load using exact same settings. Furthermore, once started and loaded, and then doing a backfill of all charts on either QT or Ensign takes less than 30 seconds. I clearly stated the morning start times in MC were from existing charts/workspaces that pooled daily data.

It's clear those at MC refuse to acknowledge there is an issue with slow load up times. After having seen numerous complaints on this forum as well as others, and having seen the issue firsthand, I decided to chime in as well. My bad for actually thinking you were interested. :roll:

I believe I will take your co-workers advice that she gave another poster when he expressed his views regarding MC's drawing tools. "Don't like it, use Ensign." I'm already there and believe that is where I will stay. :)

I appreciate you taking the time to respond. It was a real eye opener and quite beneficial to say the least. :D

b

duwdu
Posts: 6
Joined: 25 Jul 2007

Postby duwdu » 25 Jul 2007

Bilco,
It is not correct to compare MultiCharts with Ensign, because it is a known fact that Ensign downloads data from DTN IQ servers. They are good and return quotes quite qucikly.
Regarding QuoteTracker. Please make sure it uses 24 hours sessions. It goes without saying that it is impossible to download 39,600 quotes (11 hours of night trading* 3600 seconds per hour) for 3-7 seconds. You can double check it.

I do confirm that MultiCharts works slower than QuoteTracker, but the difference is not significant.

There is an option in Ensign to download IB only up to 20 hours. I used IB data only when making the comparison. Furthermore, I did note that the difference was slight although measurable. All charts were set to 24 hour, 100 bars back. The comparison was fair and honest.

As for MC starting time, this is and was separate from the testing information I gave. QT will indeed load up each morning in mere seconds, 24 hour charts, as will Ensign with global refresh disabled (uses pooled data to give fair comparison). MC takes 30-45 minutes to load using exact same settings. Furthermore, once started and loaded, and then doing a backfill of all charts on either QT or Ensign takes less than 30 seconds. I clearly stated the morning start times in MC were from existing charts/workspaces that pooled daily data.

It's clear those at MC refuse to acknowledge there is an issue with slow load up times. After having seen numerous complaints on this forum as well as others, and having seen the issue firsthand, I decided to chime in as well. My bad for actually thinking you were interested. :roll:

I believe I will take your co-workers advice that she gave another poster when he expressed his views regarding MC's drawing tools. "Don't like it, use Ensign." I'm already there and believe that is where I will stay. :)

I appreciate you taking the time to respond. It was a real eye opener and quite beneficial to say the least. :D

b
Andrew,

The various positions with MC's slowness at loading/displaying IB data as canvassed by billco and shcheuk are very, very correct. Please listen to them and address the issue appropriately, if you can. Do not even ask me to explain because I cannot express the problem half as much as they clearly have IMHO. What I can tell you for a fact is that I am also an experienced TS user, and have been with/purchased MC from about the beginning, and can attest that MC is VERY and FRUSTRATINGLY SLOW loading workspaces with IB data compared to either TS or QT.

This morning, I upgraded to 2.1.928.1050 beta (from 880.530 beta) and the story IS the same - very slow workspace load up/display performance. I am running WinXP sp2 with the latest updates and a P4 with 2GB of memory, and IB is my primary trading medium.

Don't get me wrong, there are many good and unique aspects to MC (thanks to you and the MC Team!) and for me for the time being those outweigh these IB data loading/display annoyances, and so I will stick with MC and your very good pace of development. However, fix these complaints you must -if you can. Good luck!

'Nuff said.

Duwdu

(By the way, I used to be able to log in and contribute to this forum up till a couple of months ago. Today, I was unable to log in - the system would not recognize any combination of my username and password I thought I remembered. So I had to re-register. Funny enough, the system did not say the username was already in use at any time in the immediate past. Is there a way you or any admin could restore my post count? The number is not great, but just want to keep the records right.)

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 25 Jul 2007

Bilco, Duwdu and everybody concerned,
We carefully analyzed how QuoteTracker works with IB and found stunning facts.
Environment: We just installed QT and use all default settings. We requested 1 tick bar chart. All my statements are based on our observations and could be wrong. I suggest to double check them.

First of all, yes it is fast. The most interesting question is how is this possible?

The answer was so unpredictable. They show incorrect data. If you request 1 tick bar chart QT will request minutes. It is obvious that it will be hundreds times qicker.

So they get minute bars and transform them to ticks. Example they get OHLCV= 1500 1700 1300 1600 10,000
and create 4 ticks
P:1500 V:2500
P:1700 V:2500
P:1300 V:2500
P:1600 V:2500
If they get 1 minute bar with O=C=L=H they don't transform it.

It is obvious that it is not tick data, but ultra low quality substitute which couldn't be called ticks, because ticks are first-hand and most accurate information available.
The picture demonstrating my observations is attached.

Taking into consideration the above-mentioned it is not correct to call MultiCharts slow until you have not right benchmark.

All comments are appreciated.
Attachments
qtracker - @ESU7.PNG
(49.2 KiB) Downloaded 4782 times
Last edited by Andrew Kirillov on 25 Jul 2007, edited 2 times in total.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 25 Jul 2007

P.S.
Since lots of people like QT It seems it is more important for users to have the fast data downloading rather than valid data. Do you think this is a way to go?

gregorio123456
Posts: 117
Joined: 08 Nov 2005
Been thanked: 3 times

Postby gregorio123456 » 25 Jul 2007

P.S.
Since lots of people like QT It seems it is more important for users to have the fast data downloading rather than valid data. Do you think this is a way to go?
No Andrew

Thanks

Ze

Guest

Postby Guest » 25 Jul 2007

P.S.
Since lots of people like QT It seems it is more important for users to have the fast data downloading rather than valid data. Do you think this is a way to go?
I think people mention QT for reference's sake. Don't read into messages that are not there. And don't tell people to go use QT if they like QT so much. People have made a commitment to MC. They are pointing out the deficiencies, not nagging. When feedbacks fell on deaf ears, that's when feedbacks became naggings.

duwdu
Posts: 6
Joined: 25 Jul 2007

Postby duwdu » 25 Jul 2007

P.S.
Since lots of people like QT It seems it is more important for users to have the fast data downloading rather than valid data. Do you think this is a way to go?
Absolutely NOT, Andrew. Thanks.

Duwdu

Guest

Postby Guest » 25 Jul 2007

P.S.
Since lots of people like QT It seems it is more important for users to have the fast data downloading rather than valid data. Do you think this is a way to go?
no need to compare it to QT.

Every program has its problems.
QT is not perfect. I can list 10 things that are no good with QT. But that does not relieve MC's duty and responsibility to face the reality.

duwdu
Posts: 6
Joined: 25 Jul 2007

Postby duwdu » 25 Jul 2007

P.S.
Since lots of people like QT It seems it is more important for users to have the fast data downloading rather than valid data. Do you think this is a way to go?
I think people mention QT for reference's sake. Don't read into messages that are not there. And don't tell people to go use QT if they like QT so much. People have made a commitment to MC. They are pointing out the deficiencies, not nagging. When feedbacks fell on deaf ears, that's when feedbacks became naggings.
Please, please, I'll call for caution here... Andrew was not pushing we users towards QT. As I read it, he was just asking whether, in the light of his own discovery, people would want MC to acquire data that way? The obvious answer to Andrew, IMHO, is a plain NO.

The fact still remains though that the eSignal and TS platforms still load/display genuine tick data very quickly. Although it is "their own" data, tssupport created excellent solutions that worked with those data on other platforms before, for example before TS went in-house, so they CAN solve the problem with IB tick data! :wink: :)

duwdu

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 25 Jul 2007

If you say MultiCharts is slow, you should understand what fast is. Until you have a standard you can't state that it is slow. Our facts speak to the contrary…
We are open to change our point of view, but we can’t change reality. If IB works slowly we can’t change anything. All we can do is to change visuazilation.

User avatar
gautama2
Posts: 96
Joined: 10 Jul 2007
Has thanked: 1 time
Been thanked: 1 time

Postby gautama2 » 25 Jul 2007

MC keep on your track.
I don't need fast and wrong workarounds.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 25 Jul 2007

duwdu,
You understand me right.

Guest

Postby Guest » 25 Jul 2007

If you say MultiCharts is slow, you should understand what fast is. Until you have a standard you can't state that it is slow. Our facts speak to the contrary…
We are open to change our point of view, but we can’t change reality. If IB works slowly we can’t change anything. All we can do is to change visuazilation.
that's a correct approach:

When IB is the culprit, let users know that IB is not delivering the requested data.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 26 Jul 2007

We will plot data gradually and show IB status right on the chart in the future versions.

duwdu
Posts: 6
Joined: 25 Jul 2007

Postby duwdu » 26 Jul 2007

We will plot data gradually and show IB status right on the chart in the future versions.
Andrew,

That will be a fair and welcome solution giving the circumstance and your stunning discovery. Thank you. I believe you guys ARE listening.

duwdu

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 26 Jul 2007

Thanks for your understanding.

Guest

Postby Guest » 27 Jul 2007

Slow loading is not always due to IB.

You guys can do a test......
At the end of market close do a reload of all the symbols you keep track of to get the latest updated data from IB.

So by right if you start up MC tommorrow before market open , it should be fast since you do not need to download anything from IB right ?

Wrong !!!!!
Every morning, it still takes me up to 20mins just to load 4 workspaces with 12 charts.....

MC is just dammed slow loading from its local database....

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 27 Jul 2007

Slow loading is not always due to IB.

You guys can do a test......
At the end of market close do a reload of all the symbols you keep track of to get the latest updated data from IB.

So by right if you start up MC tommorrow before market open , it should be fast since you do not need to download anything from IB right ?

It is not right. Please read this thread carefully to understand the issue. You use 24 hours sessions and the program downloads 10 hours of night data.
Wrong !!!!!
Every morning, it still takes me up to 20mins just to load 4 workspaces with 12 charts.....
MC is just dammed slow loading from its local database....
It is incorrect. Use Offline mode to see that it takes seconds to load data from the database.

Guest

Postby Guest » 27 Jul 2007

Slow loading is not always due to IB.

You guys can do a test......
At the end of market close do a reload of all the symbols you keep track of to get the latest updated data from IB.

So by right if you start up MC tommorrow before market open , it should be fast since you do not need to download anything from IB right ?

It is not right. Please read this thread carefully to understand the issue. You use 24 hours sessions and the program downloads 10 hours of night data.
Wrong !!!!!
Every morning, it still takes me up to 20mins just to load 4 workspaces with 12 charts.....
MC is just dammed slow loading from its local database....
It is incorrect. Use Offline mode to see that it takes seconds to load data from the database.
Andrew, you are not offering a realistic example.. since your customers cannot be expected to work with offline charts whilst trading.

You should try the exact scenerio that was reported....
i.e. download the data at end of market close and see if it takes a long time to startup the next day before market open.....

If the data has been loaded the night before.... MC should not take so long to start up regardless of whether its in online or offline mode.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 27 Jul 2007

Andrew, you are not offering a realistic example.. since your customers cannot be expected to work with offline charts whilst trading.

You should try the exact scenerio that was reported....
i.e. download the data at end of market close and see if it takes a long time to startup the next day before market open.....

If the data has been loaded the night before.... MC should not take so long to start up regardless of whether its in online or offline mode.

This is my last response to this question, because I don’t think it is necessary to explain the same thing dozens of times.


Please try to think logically.
1. If you want to make a fair test of database speed, use offline. It will allow you to see the maximum speed the database can archive. I don’t offer you to work offline. My proposal is to see standard speed and eliminate IB servers and the internet factor.

2. The program have to download 10-12 hours of night data. It leads to at least 1 pacing violation error and thus you will have to wait 10 minutes.
In order to fix it, please set regular sessions in MultiCharts and QuoteManager. It will force the program to use the data you already collected and MultiCharts will not download night data from IB.

Guest

Postby Guest » 27 Jul 2007

Andrew, you are not offering a realistic example.. since your customers cannot be expected to work with offline charts whilst trading.

You should try the exact scenerio that was reported....
i.e. download the data at end of market close and see if it takes a long time to startup the next day before market open.....

If the data has been loaded the night before.... MC should not take so long to start up regardless of whether its in online or offline mode.

This is my last response to this question, because I don’t think it is necessary to explain the same thing dozens of times.


Please try to think logically.
1. If you want to make a fair test of database speed, use offline. It will allow you to see the maximum speed the database can archive. I don’t offer you to work offline. My proposal is to see standard speed and eliminate IB servers and the internet factor.

2. The program have to download 10-12 hours of night data. It leads to at least 1 pacing violation error and thus you will have to wait 10 minutes.
In order to fix it, please set regular sessions in MultiCharts and QuoteManager. It will force the program to use the data you already collected and MultiCharts will not download night data from IB.
This is also the last time i am explaining to you since you don't seem to get it....

All the reguired data has been downloaded as of the last market close by forcing a reload on all symbols. There are no pacing violations as there should not be any need to download from IB.
Yet it is taking a long time to startup the next day before market open......

Why confuse the issue by bringing in regular sessions, when i am not refering to any globex data... i am refering to contracts like hong kong, sgx, ose.....

Maybe some users can understand the problem.. will try it and bring it up to you the next time...

Or maybe some of your other colleagues can understand this....

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 27 Jul 2007

Please understand that program seems slow only when it can't download missing data. Sometimes it can think that it should download data even if it shouldn’t.

Even if you use Asian markets where there is no overnight trading, you should check if you have 24 hours sessions set in Format Symbol>Settings.
Please set regular. After that please take a look at QuoteManager->Edit Symbol->Sessions and make sure they reflect actual sessions. If it is correct please see Tools->Exchange->Select Exchange->Edit Exchange->Time Zone.

Guest

Postby Guest » 27 Jul 2007

Please understand that program seems slow only when it can't download missing data. Sometimes it can think that it should download data even if it shouldn’t.

Even if you use Asian markets where there is no overnight trading, you should check if you have 24 hours sessions set in Format Symbol>Settings.
Please set regular. After that please take a look at QuoteManager->Edit Symbol->Sessions and make sure they reflect actual sessions. If it is correct please see Tools->Exchange->Select Exchange->Edit Exchange->Time Zone.
There are all in order. Right timezone, right session settings.

When i check the log for slow startup, it wasn't trying to download any data from IB, as i already mentioned, no pacing violations too.
But startup is just slow.

Did you try the scenerio i describe and the loading is still fast the next day ?

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 27 Jul 2007

Please do the following:
1. Start MC in the morning and wait til the charts are loaded
2. Send me the log from QuoteManager->Event log
3. Send me your workspaces

jek
Posts: 181
Joined: 24 Dec 2006
Has thanked: 1 time
Been thanked: 2 times

Postby jek » 27 Jul 2007

Andrew -

Could you describe the "Save data to local database" option in QuoteManager?

Would that have the effect of caching the historical data and speed up people's startup time?

Thanks.

Guest

Postby Guest » 27 Jul 2007

Please do the following:
1. Start MC in the morning and wait til the charts are loaded
2. Send me the log from QuoteManager->Event log
3. Send me your workspaces
The scenerio is easily replicated and you can test it out yourself.

Unless you are telling me that you've tried the exact scenerio and startup is fast on your machine, then i will send you my logs and workspaces ( which i am reluctant )

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 30 Jul 2007

Andrew -

Could you describe the "Save data to local database" option in QuoteManager?

Would that have the effect of caching the historical data and speed up people's startup time?

Thanks.
This option is intended for TS data only. It slightly improves performance.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 30 Jul 2007

Please do the following:
1. Start MC in the morning and wait til the charts are loaded
2. Send me the log from QuoteManager->Event log
3. Send me your workspaces
The scenerio is easily replicated and you can test it out yourself.

Unless you are telling me that you've tried the exact scenerio and startup is fast on your machine, then i will send you my logs and workspaces ( which i am reluctant )
It is not efficient to reproduce a bug if I’m not aware of your environment. The information you provided is not enough. Please do what I requested.

Nick
Posts: 496
Joined: 04 Aug 2006
Has thanked: 4 times
Been thanked: 24 times

Postby Nick » 31 Jul 2007

P.S.
Since lots of people like QT It seems it is more important for users to have the fast data downloading rather than valid data. Do you think this is a way to go?
Perhaps offering it as a choice either as a new bar type ? Just a thought.

Nick
Posts: 496
Joined: 04 Aug 2006
Has thanked: 4 times
Been thanked: 24 times

Postby Nick » 31 Jul 2007

Please do the following:
1. Start MC in the morning and wait til the charts are loaded
2. Send me the log from QuoteManager->Event log
3. Send me your workspaces
The scenerio is easily replicated and you can test it out yourself.

Unless you are telling me that you've tried the exact scenerio and startup is fast on your machine, then i will send you my logs and workspaces ( which i am reluctant )
It is not efficient to reproduce a bug if I’m not aware of your environment. The information you provided is not enough. Please do what I requested.
I think what the poster is trying to say is this happens pretty much under all circumstances. This is item 1) That I described a dozen or so posts before. It appears that MC is not sure that it has all the data it needs in the cache already.

It should not be necessary to 'tell' MC it has all the data it needs by switching to offline mode. It should be able to look at the data base and quickly see that there is no data missing. (as it retrieves it to put it on the chart). Put another way the speed to load from the database should be hardly any different regardless whether you are in offline mode or not if all the data is there.

Often I do not have all the data I need but I certainly have most. Often I have a bar or two missing from yesterday if I have been lazy and finished early :oops: sometimes I have a bar or two missing from toda if I over sleep :oops:. In this case starting in offline mode is not an option. Similarly if you re-start MC during the day you may miss a bar or two. In this case just getting a small amount of missing data can seem 'slow'. Let me stress offline mode is not an option as you will need a couple of 1 minute bars.

Please note this is just my interpretation of what the poster above is saying. My general observations tend to agree but for now my 'hardcore' testing days of MC are on hold. (In the past I would have examined the database to see what was there and examined the logs to see what was actually requested).

Cheers,
Nick.

Guest

Postby Guest » 31 Jul 2007

Nick, thanks....yup it happens everyday is definitely easily replicable.....

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 31 Jul 2007

Nick,
There is misunderstanding. I don't offer to work "offline". Offline mode allows us to isolate the problem.

If you run MC most of the time you can use Online mode with "I want to download missing data" option unchecked.
In this case the program will take data from the database (almost instantly) and connect to live data.

Anyway I need workspaces to identify the problem.

guest02

Postby guest02 » 31 Jul 2007

if you want to solve the problem yourself, you can diagnose it anyway you like.
if you want mc to solve the problem, get what mc needs.
i don't see what is the point of the discussion.
this is not a philosophy class.

User avatar
Andrew Kirillov
Posts: 1589
Joined: 28 Jul 2005
Has thanked: 2 times
Been thanked: 31 times
Contact:

Postby Andrew Kirillov » 01 Aug 2007

I'm fully agreed with guest02 - we spend valuable time for discussion rather than fixing the issue. Please provide the required material.


Return to “MultiCharts”