[ale] Giant storage system suggestions

[Alex Carver]

On 7/13/2012 18:55, Ron Frazier (ALE) wrote:
>
>
> Alex Carver <agcarver+ale at acarver.net> wrote:
>
>> On 7/12/2012 04:36, Barlow, Jim D wrote:
>>>
>>>
>>> As Jeff Layton says:
>>>
>>>> If you want to go a bit cheaper, then I recommend going the
>>>> Gluster route. You can get it for free and it only takes a bunch
>>>> of servers. However, if the data is important, then build two
>>>> copies of the hardware and rsync between them - at least you
>>>> have a backup copy at some point.
>>>
>>>> Good luck
>>>> Jeff
>>>
>>> Alex:
>>>
>>> It is a great time to check  out Gluster http://gluster.org as of the
>> 3.3 release.  Red Hat has put a lot into the project lately.   I've
>> been testing it for my own uses and have been delighted with it.   It
>> may complement your ideas.
>>> I've ordered some inexpensive x86_64 hardware  function as Gluster
>> bricks for my own elastically scalable storage.     It mounts as a high
>> performance fuse file system or NFS, with a Samba shim you are good to
>> go with CIFS.
>>
>> I took a look at Gluster but I don't think I want to go that route.  I
>> want to have the fewest number of machines/OSes to maintain (ideally
>> just one but maybe two if the array gets too big for one machine).
>> This
>> would be especially important at work where I can't get more than one
>> or
>> two network IPs.  The maintenance requirements to keep a half a dozen
>> or
>> more machines up to date in software is a lot more work than just one
>> that I can leave under my desk and check on once in a while.
>>
>> So I think I'm still going to be leaning towards one machine and a pile
>>
>> of drives for now.
>> _______________________________________________
>> Ale mailing list
>> Ale at ale.org
>> http://mail.ale.org/mailman/listinfo/ale
>> See JOBS, ANNOUNCE and SCHOOLS lists at
>> http://mail.ale.org/mailman/listinfo
>
> Hi Alex,
>
> Caveat, I'm not a giant storage expert, but ...
>
> * Cost may be obnoxious (don't know) but what about this
>
> http://www.drobo.com/products/business/b1200i/index.php
>
> You could do 24 tb in one chassis with 2 tb drives.
>
> * Re only 2 ip's, could you stick your server behind a nat router at work?  Then you could jave lots of ip's behind the router.  Also, if you could run ipv6, you could probably get 50000 ip's.
>
> Sincerely,
>
> Ron

Well, first, routers are a no-no at work.  Any IP based device must have 
a live IP (for the IT security division to be able to examine it for 
vulnerabilities) or it must run in an isolated network.  No IPv6 yet at 
work (even if there was, 50,000 IP addresses would break the bank 
because each end user leases each network drop/address as the overhead 
to pay for the IT support contract).

I can't discount completely one of these premanufactured systems but 
lock-in concerns me.  If the vendor closes shop I have trouble.  If I 
build a system, then it doesn't matter who makes what part as long as 
there is a device driver available for the OS to read the drives and the 
RAID driver will be able to read the data even if I have new SATA cards 
and a new motherboard.  In this case I still need a separate server to 
do my user interface for data dumping but I'll consider the manufactured 
solution.