Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
We are using the following combination of SSD/HDD for our DRBD cluster and so far it works pretty well. primary: 8x Intel SSD 320, 300GB (in RAID6) secondary: 8x WD 7.2k NL-SAS, 1TB (in RAID6) So all the READ requests willl be handled by the SSDs, while the WRITE has to be done by both of course. This cluster is used solely as a mail-server backend for ~30k mailboxes, thus in general we see much much more READ IOPs than WRITE. (this was also the decider, why we even considered to use such a performance-asynchron disk setup) In our production environment this works out as expected and using the cluster this way around (SSD as primary) delivers around 5-10x the performance as vice versa. We did test this with loading mdir mailboxes with 20k+ messages in a webmail interface that fetches ALL the messages from the server and then sorts in in the frontend afterwards. The loading time of such boxes, while SSD is primary, are by magnitudes faster....i.e. 3-4 seconds vs 20-30 (of course while system was in production and thousands of users online) Still, as long as the expected workload will not be massively dominated by READ IOPs, I also have to aggree that SSD vs 7.2k SATA/SAS may not be the best setup to use overall. regards Christoph Thursday, December 20, 2012, 1:43:39 PM, you wrote: > Thanks, James, you seem to keep forgetting the list CC though ;) > -------- Original Message -------- > Subject: Re: [DRBD-user] Secondary Performance > Date: Thu, 20 Dec 2012 11:40:27 +0000 > From: Prater, James K. <jprater at draper.com> > To: 'ff at mpexnet.de' <ff at mpexnet.de> > What Felix suggested (switching roles) should work. Personally, I > would not use SSDs for that type of deploy. It is best placed to speed > up certain processes (swap, scratch pad database location used for > indexes or anythi > drbd-user at lists.linbit.com > http://lists.linbit.com/mailman/listinfo/drbd-user