[RFC 004] Operation Review Comment 8

Here are my responses to the questions. I have not copied the questions into this, however...

1. I set up an OpenDAP HDF and NetCDF server here for the purpose of getting at our direct broadcast MODIS and AIRS data. The installation and setup of these servers is very easy. We have been using OpenDAP for more than 3 years.
2. HYDRA (hyperspectral data viewer and analysis tool) uses OpenDAP to get data from our DB plus the Goddard DAAC. The Integrated Data Viewer (Unidata) is used by our researchers and instructors for general weather analysis and retrieves data from OpenDAP servers at UCAR and NCDC.

2a. Other people use McIDAS, Matlab, and IDL, plus a few other applications. Of these, McIDAS is the only one I know that does not have an OpenDAP connection.

1. We use OpenDAP because it allows applications to be written without programmers and scientists having to learn how to use HDF APIs. Many applications are available that read NetCDF files and OpenDAP allows these to continue to be used, even if the server is actually reading HDF. (All my attempts to get HDF banished have failed, so this is a good alternative...)

3a. We also use ADDE for image data. The advantages to ADDE are that the servers know about their data and can do geographics subsections directly ("give me 1000 x 1000 pixels centered at 43N/-89W"). In addition, the ADDE servers can re-calibrate image data. The disadvantage is that up until this Winter, ADDE was not open (it is now Open Source). There are also some subtle advantages to OpenDAP in that it uses HTTP protocols on Port 80

sometimes in field experiments at military bases, the ADDE port (112) is not open and cannot be opened, but port 80 is available.

1. By "systems" you might mean: a) servers -- very easy to set up, b) applications - depends on the application, of course, but both HYDRA and the IDV hide all the actual URL-generating code from the user.

4a. cannot tell.

1. Let's just say that most of the time it is faster for me to read an HDF file via OpenDAP from the Goddard DAAC than it is to read it off my local hard-drive (granted, it's a laptop, but still...)
2. About the only bump in the workload for our servers is to install and test updates. This usually requires very little time and happens infrequently.
3. Sorry, but we don't keep stats on access....
4. Ditto...