Opensource Tools for Optimizing Server Performance

Question

I use a server with a basic Microsoft OS primarily for data storage and as a central repository for edited vector data. These data are not intended to be used as mapping or GIS products available publicly on the web. Additionally, I have always used a standard file structure for organizing data on the server, where files exist within folders. Given this information, I am interested in optimizing my server performance with opensource database management tools such as PostgreSQL and PostGIS. I do not have a strong IT background and am looking for a starting point. So far, I have downloaded PostgreSQL/PostGIS, although I find the process of loading data cumbersome and find myself asking why go through the hassle?.

Questions:

• Where should I begin in bringing my GIS server and data management practices up to par?
• What opensource/commonly available software can be utilized to optimize a server designed for editing vector data and storing GIS data?

Answer 1

One of the first things I did was replace the built in Disk Defragmenter with the excellent free Defraggler from piriform:

http://www.piriform.com/

for data management and conversion:

http://www.spatialytics.org/projects/geokettle/

Geokettle has many uses from flat file and geo file conversions, to transforms and database import/export tasks. Put simply it can transform anything into anything else.

For organizing your folders and seeing which ones are using what space, JDisk report:

http://www.jgoodies.com/freeware/jdiskreport/

and then on top of that I have a handfull of various command line tools that to be honest I use on all my machines, not just my GIS servers, and that are used for all manner of different data & maintainenence tasks.

Tools such as those from the GNU Win32 project:

http://www.alexa.com/siteinfo/gnuwin32.sourceforge.net

and windows sysinternals:

http://technet.microsoft.com/en-gb/sysinternals

Answer 2

It's worth noting that running Postgresql/PostGIS on a Linux server offers many more optimization options than on a Windows server (kernel tweaks like shared memory, WAL buffers and the like). Also, when migrating your data to PostGIS you can take advantage of the schema concept to duplicate your current file system/directory structure to keep your data organized in way familiar to you. Lastly, on Linux (probably windows also) you can script the upload of data layers to the PostGIS database to make the chore somewhat more hassle-free.