Date: Mon, 27 Jun 2005 08:30:59 +0200 From: Oliver Brandmueller <ob@e-Gitt.NET> To: freebsd-stable@freebsd.org Subject: Re: ICP Vortex GDT8546RZ vs 3Ware 9500S-4LP ... ? Message-ID: <20050627063059.GA36150@e-Gitt.NET> In-Reply-To: <20050626232433.O57847@ganymede.hub.org> References: <20050626232433.O57847@ganymede.hub.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi. On Sun, Jun 26, 2005 at 11:26:55PM -0300, Marc G. Fournier wrote: > Just curious as to if ppl consider one better then the other, or are they > about the same? Both are quite fast, both bring software. We replaced our 3ware controllers since we had what seems a compatibility problem between the controller and out harddisks (Western Digital 74 GB 10krpm Raptors): Occasionally on the 3ware a random disk dropped out of our RAID 10. You had to remove power from the disk(!) first, to get it acknowledged again by the controller and then do a rebuild. This seems to occur only on very high I/O loads and we could not explicitly trigger it. Sometimes it happened twice a week in one machine, sometime it did not happen several weeks on all 5 machines with this setup. We have no problem with the 3ware and other disks. We have no problems with these disks on the ICP Vortex. For the software: 3ware gives you a cli, which is quite handy when it comes to writing a nagios plugin or doing the usual maintenance stuff. The commands are not straight forward in the first view, but it's easy to understand the concept behind. The big pro here is, that you can handle everything even in Shell scripts (you can just call the cli with the right commands, the output is somewhat parseable). There's no generic way for SNMP access, though. ICP comes with a management console (it's a curses interface, it's exactly the same as going into the controller BIOS on boot) and a management daemon. The management console (icpcon) is pretty straight forward, you can get a lot of information (like watching io rates on all the subsystems down to disk level, getting S.M.A.R.T. info and so on). This interface is not usable in a batch (like getting a status for a nagios plugin or so), you can write out an info file, which is hardly machine parseable. You have the daemon (srcd) to do the monitoring. The daemon can be connected by management consoles (so you don't need to logon the host to monitor or write the info file from), it can send SNMP traps and it can start commands in case of events. It cannot be queried by SNMP. We're very comfortable with the ICPs now, but exchanging the slightly cheaper 3wares was mainly done because of the drive problems mentioned before. - Oliver -- | Oliver Brandmueller | Offenbacher Str. 1 | Germany D-14197 Berlin | | Fon +49-172-3130856 | Fax +49-172-3145027 | WWW: http://the.addict.de/ | | Ich bin das Internet. Sowahr ich Gott helfe. | | Eine gewerbliche Nutzung aller enthaltenen Adressen ist nicht gestattet! |
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050627063059.GA36150>