[Bioclusters] Urgent advice on RAID design requested

[Joe Landman]

Heh ... Chris beat me to it.

Chris Dagdigian wrote:
> 
> Reliable? With a single power supply?  That may be a good home NAS box 

Single points of failure do not reliable units make.  Our JackRabbit box 
has N+1 powersupply configs, multiple redundant gigabit ethernets, and 
lots of other goodies.

Reliable usually means redundant and at least resilient to single 
failures.  RAID is *not* resilient (yes I know what the "R" stands for) 
unless you have multiple paths to the disk such that a single controller 
failure or single disk failure (or single network, or ...) takes away 
your data.

Unfortunately it is very hard to do this "cheaply".  It can be 
in-expensive, but you have to pick what features you want and which ones 
you can live without.

For (smaller) clusters, unless you have other compelling business issues 
on time-constrained access to data, often times a good NAS appliance may 
be sufficient.  Unfortunately most NAS units do not scale in performance 
(JackRabbit does, but I am resisting letting this message become a 
commercial for it).  For better resilience, if your business case 
demands it, replication of data across multiple units is usually a good 
idea, so loss of single boxes (iSCSI bricks or JackRabbit) does not 
impact data accessibility.

You can get cheap or reliable.  You get to pick one (today).  Usually 
you get to pick reliable or fast, but you can have both with the right 
product.

Joe

-- 

Joseph Landman, Ph.D
Founder and CEO
Scalable Informatics LLC,
email: landman at scalableinformatics.com
web  : http://www.scalableinformatics.com
phone: +1 734 786 8423
fax  : +1 734 786 8452 or +1 866 888 3112
cell : +1 734 612 4615