Sean Ward <sdward@uswest.com>
Joel Lee <jlee@thomas.com>
Matthew Stier <Matthew.Stier@tddny.fujitsu.com>
Birger Wathne <birger@sdata.no>
Summary of questions and responses :
>Does anybody have any information, experience or advice regarding :
>1. server- v's network-attached RAID in general ...
- Overall, if you can afford the latter, go for it. Tend to be a lot
more expensive than server-attached systems, but very strongly
recommended for large, particularly production-based, environments.
- Cautions: don't get locked into anything proprietary (e.g. backup
schemes), ensure the vendor can provide (and keep you up-to-date with)
multiple protocols.
>2. Symbios Metastor Workgroup systems in particular ...
No takers (should I read something into this :) ?
>3. Network Appliance Corp. products (this name came up frequently in the
>archives, usually with high praise - are they still so highly
>recommended ?)
Yes they are ! All respondents with experience of these systems
commented very favorably about them, with particular reference to :
- stability and reliability
- performance
- ease of maintenance
- native support for Unix / NT
- ease of management
- service - one respondent wrote that his system detected a drive
failure and notified NetApp who over-nighted him replacement parts -
before he was even aware there was a problem (!).
>Lastly, if we do go for a server-attached solution, I would want to use
>multiple Ethernet interfaces (on the same sub-net) ... Is there any good
> documentation on how to set this up with specific reference to load
> balancing ?
No pointers to documentation as such, but the following were suggested :
- manual load balancing via routings tables
- multi-port trunking
- set up routing so that the interfaces on the server are dedicated to
in-and out-going traffic respectively.
My thanks again to all who responded.
Best regards,
-- craig