[tahoe-dev] https://tahoe-lafs.org/trac/tahoe-lafs/wiki/FAQ Q25

[Jimmy Tang]

2012/2/10 Vikarti Anatra <vikarti at viorsan.com>:

>>
>> * Are you participating in a private or public grid?
> public or VG2?
> * if you are not using public grid - why?(text field)

I purposely left it public or private to reduce the amount of
branching in the survey.

>
>> * Do you run a Storage or Introducer Node or Both?
>> * Is your (Non-client) node a dedicated host with a high uptime (>95%)?
> You mean - only used as Tahoe-LAFS node?what if host has high uptime and node's role is not only it's main role?(may be it's also client, webhost,etc)?

If the node isn't a dedicated host (a server that is meant to be up
all the time) then it isn't a very good server, I'm not sure how to
phrase this.

>
>> * What is the small disk in your cluster
>> * What is the biggest disk in your cluster
> Shouldn't be 'space allocated to node'?for example I have reseverved_space=1400G on one of nodes, this machine is not dedicated in this role
>> * What is the average size of the disks
>> * How many nodes are in your cluster
>> * Your Value of N used (total number of nodes)
>> * Your Value of H used (servers of happiness)
>> * Your Value of k used (minimum number of shares required to recover the data)
>> * If tahoe-lafs didn't exist, what would your alternatives be (e.g.
>> GPG'd files put into LOCKSS or GlusterFS or PLACE DISTRIBUTED SYSTEM
>> HERE...)
> * how you use your grid?tahoe backup/FUSE/pyfilesystem

I added the client access point

> * which host OSes you use on grid, with versions: like CentOS 5.7/WindowsServer2008R/Win7(or may be several checkboxes in style Do you use OS X)

I added this point about OS's

> * approximate balanced storage space:xx Gb

I'm not sure how you can capture that in a survey, I would have
assumed that  (min disk space * number of nodes) would give you that
information. I guess it should be worth asking how many storage nodes
there are in the grid.

> * number of clients(if client is also storage node - count it here too)
>

I added a question about the number of clients in the grid

>>
>> There might be a need to do two slightly different set of questions,
>> one for grid operators and clients. The above is just food for thought
>> for now, I'd be willing to setup a google form for running a survey if
>> there is interest.
>>
> At least I would be interested to read it.
> For clients - just different questioons:grid, N/H/k and so
>
>
>
>
> for example in my case(may be add something from here if you think that's interesting)
>
> Number of storage nodes:11 currently
> Planned future storage nodes:2-4(LAE one when it stabilizes a little more and at my friends locations at other cities)
> Max number of nodes per one location(i.e.  same physical LAN and/or physically near):4
> Number of nodes used as gateway:1(more planned real soon)
> Access method to grid used: tahoe backup, SSHFS+OSXFUSE, Dokan+Pyfilesystem
> Host OSes used: Windows Server 2008R2,Windows 7(does not have it's own node,only uses pyfilesystem for now),Linux(CentOS 5.5-5.7), Mac OS X 10.7.2
> Approximate balanced raw storage space(not accounting for N/k):450 Gb(unbalanced space can be made about 1.5 Tb but only 2 nodes have >1 Tb space so I blocked this)
> Are they public gateways: No, but I'm looking at lafs-rpg project for making R/O access possible so I just could share files with others this way

I'd be interested as well myself to see what sized grids are out there
and the encoding parameters people use.


Jimmy.

-- 
http://www.sgenomics.org/~jtang/