Re: [thredds] TDS performance and NetCDF file

• To: "thredds@xxxxxxxxxxxxxxxx" <thredds@xxxxxxxxxxxxxxxx>
• Subject: Re: [thredds] TDS performance and NetCDF file
• From: Jon Blower <j.d.blower@xxxxxxxxxxxxx>
• Date: Tue, 19 Apr 2011 17:13:58 +0000

Hi Jay,

>From the WMS point of view I don't think file size is an issue in the way you 
>describe.  The main issue is data resolution (i.e. number of grid points in 
>the horizontal), not the number of variables contained in a single file .

I don't know about NetcdfSubset though.

Cheers, Jon

-----Original Message-----
Date: Tue, 19 Apr 2011 09:09:10 -0700
From: Jay Alder <jay.alder@xxxxxxxxxxxxxxxxxxx>
To: THREDDS community <thredds@xxxxxxxxxxxxxxxx>
Subject: [thredds] TDS performance and NetCDF file size
Message-ID: <D361A9DA-DD99-4BD6-82AE-3CAFB761704A@xxxxxxxxxxxxxxxxxxx>
Content-Type: text/plain; charset="us-ascii"

Hello, I'm was wondering if there is a rule of thumb on NetCDF file sizes in 
TDS? I'm building a catalog of regional model data and I'm faced with the 
decision of structuring my files as one variable/file or all variables/file. 
The all in one approach would likely create files between 5-10Gb in the worse 
case scenario. Are there any performance drawbacks with large files, 
specifically WMS and NetcdfSubset? Are there any upper limits on file size 
NetcdfSubset can create? I saw there were some large file size problems last 
May using HTTPServer, but I expect to primarily use NetcdfSubset so users can 
cherry pick the variables they're interested in.

Thanks

-- 
Jay Alder