Ger<br><br>couldn't you use the multi VFL driver? In any case, the nGB/sec rate will not be sustainable by any hard drive today in the market right? Even if you buffer in SSDs still the disk write speed is your bottleneck rather than hdf5...<br>
<br>HTH<br><br>-- dimitris<br><br><div class="gmail_quote">2010/1/15 Ger van Diepen <span dir="ltr"><<a href="mailto:diepen@astron.nl">diepen@astron.nl</a>></span><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
We have an instrument generating several GBytes/second.<br>
We would like to dump these data as fast as possible to disk (of course<br>
in a parallel way), preferably with O_DIRECT to bypass the OS file cache<br>
and to avoid seeks to maintain meta data.<br>
Yet, it would be nice if these data can be seen as an HDF5 dataset.<br>
<br>
My question is if it possible to do something like this. Either that<br>
HDF5 has the option to write data in this way or through some mechanism<br>
that HDF5 can treat such an external file as HDF5 data, possibly by<br>
dynamically loading a shared library that knows how to interpret the<br>
data.<br>
<br>
I'm asking this because we have the impression that something like an<br>
extendible dataset does not give us the performance/robustness we need.<br>
In case of a crash we would like to lose as few data as possible, which<br>
means that a regular flush has to be done which might kill performance.<br>
I guess that without a flush the indices are only written at the end, so<br>
the data cannot be found in case of a crash.<br>
<br>
Cheers,<br>
Ger van Diepen<br>
<br>
<br>
_______________________________________________<br>
Hdf-forum is for HDF software users discussion.<br>
<a href="mailto:Hdf-forum@hdfgroup.org">Hdf-forum@hdfgroup.org</a><br>
<a href="http://mail.hdfgroup.org/mailman/listinfo/hdf-forum_hdfgroup.org" target="_blank">http://mail.hdfgroup.org/mailman/listinfo/hdf-forum_hdfgroup.org</a><br>
</blockquote></div><br>