Thanks Lars, that really helped! I totally get what you're saying here now.<div><br></div><div>I've sent off a request to the sales linbit folks to ask about a proxy trial. More out of experimental curiosity than anything I'd be interested to give that a shot to see what happens with our SSD/spindle combination... realistically we'll just need to go fork over the money and buy a 2nd SSD for our secondary box and call it a day.</div>
<div><br></div><div><div><div class="gmail_quote">On Fri, Sep 21, 2012 at 9:39 AM, Lars Ellenberg <span dir="ltr"><<a href="mailto:lars.ellenberg@linbit.com" target="_blank">lars.ellenberg@linbit.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
On Fri, Sep 21, 2012 at 07:11:53AM -0300, Andrew Eross wrote:<br>
> Hi guys,<br>
><br>
> I've been doing the pre-requisite Google research and I haven't reached a<br>
> conclusion, so thought I'd ask here. I have an experimental pair of<br>
> identical XenServers setup with DRBD running over a Gigabit cross-over<br>
> cable. The only difference is that the primary has a SSD and the secondary<br>
> is a normal spindle drive.<br>
><br>
> dd tests on the underlying hardware show:<br>
> * the spindle server is capable of writing at ~70MB/s<br>
> * the SSD server at ~250MB/s<br>
><br>
> If I put the primary into drbd standalone mode, I also get about ~250MB/s<br>
> when writing to the DRBD device.<br>
><br>
> When running in primary/secondary mode, however, we only get around the<br>
> ~65MB/s mark, which makes perfect sense with protocol C.<br>
><br>
> I was expecting that if I switched to protocol A, I would be able to let<br>
> the SSD drive write at it's full speed (e.g. 250MB/s) only at the price of<br>
> the secondary potentially falling a little bit behind, however performance<br>
> is almost exactly the same with protocol A, B, or C at around 60-70MB/s.<br>
<br>
Throughput != Latency.<br>
<br>
<br>
<br>
(thanks, <a href="http://ascii-art.de" target="_blank">ascii-art.de</a>)<br>
___________<br>
/=//==//=/ \<br>
|=||==||=| |<br>
|=||==||=|~-, |<br>
|=||==||=|^.`;|<br>
jgs \=\\==\\=\`=.:<br>
`"""""""`^-,`.<br>
`.~,'<br>
',~^:,<br>
`.^;`.<br>
^-.~=;.<br>
`.^.:`.<br>
\ /<br>
\ funnel/<br>
\ /<br>
\ /<br>
\ /<br>
`---- pipe ----<br>
<br>
<br>
<br>
Ok, so if that funnel is big enough for one bucket,<br>
you can pour out one bucket quasi instantaneoulsly.<br>
<br>
During the time it takes you to fetch the next bucket,<br>
the funnel asynchronously drains through the (thin) pipe.<br>
<br>
"Feels" like a "fat pipe", but is not.<br>
<br>
Now, if you fetch the new bucket faster than the funnel can drain,<br>
you reach congestion, and you have to pour more slowly.<br>
<br>
Unless spilling is allowed ;-)<br>
<br>
> I then tried combining that with "on-congestion pull-ahead;" to see if that<br>
> would allow the primary to write at full speed, but still, same result.<br>
><br>
> Is this simply not do-able for some reason to let the primary write at a<br>
> faster speed than the secondary?<br>
<br>
For a short peak period, yes, see above.<br>
To extend that peak period (increase the size of that funnel),<br>
we have the drbd-proxy (contact LINBIT).<br>
<br>
But even with massive buffers (funnels),<br>
the sustained mid term/long term average write rate<br>
obviously cannot exceed the minimum bandwith within the whole system.<br>
<span><font color="#888888"><br>
--<br>
: Lars Ellenberg<br>
: LINBIT | Your Way to High Availability<br>
: DRBD/HA support and consulting <a href="http://www.linbit.com" target="_blank">http://www.linbit.com</a><br>
<br>
DRBD® and LINBIT® are registered trademarks of LINBIT, Austria.<br>
__<br>
please don't Cc me, but send to list -- I'm subscribed<br>
_______________________________________________<br>
drbd-user mailing list<br>
<a href="mailto:drbd-user@lists.linbit.com" target="_blank">drbd-user@lists.linbit.com</a><br>
<a href="http://lists.linbit.com/mailman/listinfo/drbd-user" target="_blank">http://lists.linbit.com/mailman/listinfo/drbd-user</a><br>
</font></span></blockquote></div><br></div>
</div>