Hi Eugen,<br><br><div><span class="gmail_quote">On 4/21/06, <b class="gmail_sendername">Eugen Leitl</b> <<a href="mailto:eugen@leitl.org">eugen@leitl.org</a>> wrote:</span><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
> I'm trying to use DRBD on *CompactFlash* media. This kind of media is<br><br>Current flash is not optimized for write-intensive applications.<br>You're risking flash failure, soon.</blockquote><div><br>Well, the CF should survive
300.000 to 1.000.000 cycles - that should be enough. I'm not going to write often to it, I just want 2 CFs synchonised when I sometimes write to one.<br><br>As I understand there should not be a major increase in writing cycles to the medium by using drbd. Or do you think there is -
e.g. in the area of the metadata?<br></div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">> extremely slow on write (~0,17 MB/s) and slow (~3.4 MB/s) on read - compared
<br>> with harddisks. They are formatted with reiserfs - that could explain the<br>> extremely slow writing speed I got by using cp from ramdisk and calculating<br>> the rate. But anyhow CF is very slow compared with harddisks.
</blockquote><div><br>This might be solved now with my post<br><a href="http://lists.linbit.com/pipermail/drbd-user/2006-April/004858.html">http://lists.linbit.com/pipermail/drbd-user/2006-April/004858.html</a><br></div><br>
Warm regards,<br>Martin<br></div>