Hi Thomas,

The files being generated are anyways required to be persisted for
other processes to use them.
The data can still be sent over TCP apart from being written to file.
But then that would create performance issues on the writer side since
now the writer needs to do both write over TCP as well as write to
Files.
We need to measure if that write performance impact is bearable.

Instead, we were thinking of keeping the CSV file store in either SSD
or Ramdisk.
That should increase the read/write performance of the CSV files as
compared to writing them on magnetic disks.
Ramdisk will not work if the CSV files are being generated from
different machines.
But attaching a common SSD disk (or disk array) across multiple
machines could provide an alternative.

Do let me know your thoughts and experiences.

Regards,
Vipul.

On Jan 23, 2:53 am, Thomas Mueller <[email protected]>
wrote:
> Hi
>
> > What about the overall design?
> > CSV files will be continuously generated by various processes.
>
> Is it possible to send the data over TCP/IP or HTTP instead? Creating
> and reading intermediate files sounds like a performance problem.
>
> > Do you think H2 should be able to take that load - 24x7 operations
> > with 99.99% availability?
>
> If you need 99.99% availability, I suggest to use some kind of
> failover. I don't think that regular hardware can do 99.99%.
>
> Regards,
> Thomas

-- 
You received this message because you are subscribed to the Google Groups "H2 
Database" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/h2-database?hl=en.

Reply via email to