I have been playing with a local installation of PostGIS & GeoServer. I would like to move my work online now to share it with several collaborators.

The proof of the concept map will consist of simple WMS and WFS services mapped through OpenLayers and GeoExt. It will be based on a map of ~3000 administrative units. Initially it needs to serve several people only. From then we will be working on polishing final view and functionality of the application.

There have been several suggestions made under answers to this question. For example, Nicklas is running his site on 512MB Linode (I think...).

I was thinking about using Ubuntu Server on cloud VPS, most likely hosted on rackspace.

My question: what would be the minimal server size I would need to use to start playing with this data online? There are few of them available. Once we decide to open the service to public rackspace promises seamless way of scaling up the service, so I would like to know what bare minimum RAM I need to start with.

Update 1: Similar discussion on GeoServer forum.

Update 2: Useful tutorials on setting up stack on AWS micro instance using Ubuntu and CentOS.

  • 1
    how many users are you expecting to use this at any one point in time? Ubuntu min RAM is 1GB.
    – Mapperz
    Jan 18 '11 at 19:05
  • for testing stage - definitely less than 5 occasional users
    – radek
    Jan 18 '11 at 19:13
  • 1
    128 mb, Ubuntu: help.ubuntu.com/community/Installation/… Jan 19 '11 at 9:10
  • 1
    But I guess Geoserver might need some RAM. I use Mapserver in my application. Jan 19 '11 at 9:20
  • 2
    Is there some type of standard speed test we can perform to illustrate the performance of different setups? I have postgis/mapserver running on shared server at webfaction using CentOS, and also on an ubuntu desktop. I'm not quite sure how best to characterize their performance though.....
    – djq
    Nov 14 '11 at 16:20

It really depends on the size of the datasets you are talking about and the complexity of your queries. I for example run pretty happily on a GoGrid server /PostGIS / Windows 2008 (32-bit) running IIS, a mix of PHP/.NET homegrown webservices with 2GIG ram/ dual core. The main spatial table I query has about 6 million records I think of mostly California data. I do fairly complex spatial queries with that on the fly intersections/transformations, distance checks with geometries some of which are fairly big. They go from single points to large polygons with 8000-10000 someodd points.

Given that Windows tends to be a hog as far as RAM with all that extra junk running. I imagine you can do pretty well on 1G ubuntu (and possibly decently on 512 MB). I'm not sure how much of a hog GeoServer is though but I imagine it can't be much worse than what I'm running.

Forgot to say. You should run with PostGIS 1.5/Geos 3.2. The performance is much better than older versions particular for neighbor queries involving ST_DWithin and so forth (thanks to Nicklas :))

  • Thanks :-) What a cryptic user name. Under cover? :-) Jan 20 '11 at 9:49

Thinkgeo seems to have some info.


There also seems to be an AMI ready built.


Standard Instances
Small Instance (Default) 1.7 GB of memory, 1 EC2 Compute Unit (1 virtual core with 1 EC2 Compute Unit), 160 GB of local instance storage, 32-bit platform

Large Instance 7.5 GB of memory, 4 EC2 Compute Units (2 virtual cores with 2 EC2 Compute Units each), 850 GB of local instance storage, 64-bit platform

Extra Large Instance 15 GB of memory, 8 EC2 Compute Units (4 virtual cores with 2 EC2 Compute Units each), 1690 GB of local instance storage, 64-bit platform

  • Great info there, even had a white-paper about a project for the second link. Thanks for that Brad.
    – D.E.Wright
    Nov 14 '11 at 19:40
  • Wow, @D.E.Wright - that's been a few days ago since I posted it. I am now using the large instance on amazon east. Just noticed a new region started. I may build an instance on it soon.
    – Brad Nesom
    Nov 14 '11 at 19:59
  • Just saw it updated; too bad the costs are so high; would love to do some R&D in the cloud.
    – D.E.Wright
    Nov 14 '11 at 20:29
  • I spent about $150 us in a week worth of spin-up, spin-down, data loading. I think it is well worth some R&d cost like that. If I get to a point that I can't spend time on it you just stop the instance and all costs stop.
    – Brad Nesom
    Nov 14 '11 at 20:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.