I need lots of historical tick data to run some statistical functions.

If I am trying to access tick data via IHistory.ReadFeedData, the ticks not present in local cache will get downloaded, which is too time-consuming to be feasible, but pointing the client to my local tick database with IClient.setCacheDirectory may result in all the data getting wiped, as per to the API documentation.
Provided that I am absolutely comfortable with having a massive amount of cached ticks on my hard-drive (talking multiple years and dozens of instruments), what would be an elegant and maybe robust way to access these from the API preferably as some kind of Java collection?

Any help is welcome, Thank you in advance.

Hi,

Do you need this 'real time'? Or is it sufficient to run those functions on yesterday's data?
If the latter, there is the Historical Data Feed. Maybe it can be used...

Thank you very much, but I do need the data real-time.

How I see it at the moment is IHistory interface being somewhat unsuited for bulk tick-data access, as the data that is already present locally gets compressed and sliced up, making it quite resource-intensive getting those arrays unless it's about a few hundred data points.
It would be by all means nice to provide some option of having historical data cached in an uncompressed manner for those of us being more worried about execution times than hard-drive footprint.

What I am trying to do ATM is using the downloaded tick data to build a custom "history-engine" with java.nio and memory mapped files. Maybe I'll update this thread in case of any considerable progress.