<div dir="ltr"><div class="gmail_quote"><div>Hello,</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div><p>- in 1st case (originator is "disk" node) : about 1Gbps on
receiving host<br>
- in 2nd case (originator is "diskless" node) : about 500Mbps on
every of receiving hosts<br>
</p>
<p>From what I see I conclude, that in 1st case (originator is
"disk" node) the single copy of replicated data travels through
network, while in 2nd case (originator is "diskless" node) there
are two copies travel through network.</p></div></blockquote><div><br></div><div>My understanding is the following...<br><br>- If you have 3 nodes, 2 of them Diskful (A,B) and 1 Diskless (C), then if Primary is A or B, then there will be *one* write stream over the the network (either from A -> B or B -> A), reads will be served locally on each Diskful Primary node
(if load-balancing is enabled also over the network). Node C is ignored in this case.<br> If the Primary is the 3rd (Diskless) node (C), then there will be *two* write streams going through the network (C -> A,B) and possibly 2 read streams depending if the load-balancing option is enabled or not. This is probably the scenario you mentioned.</div><div><br>- If you have 3 (A,B,C) nodes and all of them are Diskful, then when any of those nodes is in Primary mode there will be *two* write streams going through the network(A -> B,C or B -> A,C or C -> A,B) reads will be served locally on each Primary node (if load-balancing is enabled also over the network). This is probably the scenario that Robert mentioned.<br><br>I cannot comment on multicast and if it can implemented in DRBD though... :)<br><br>G.<br></div><div> </div></div></div>