5
votes

Background: Our group has full ESRI licenses and applications built with ArcGIS Desktop.

We are invested in the arc geodatabase itself and will not change that tier.

There is a proposal to switch from ARC applications to QGIS and QT with the use of the ESRI geodatabase.

There is a push for:

Being platform agnostic: Issues when windows (client side) moves OS.

True separation between software tiers - This argument is not strong for the case, as this can be accomplished in many ways. Better programming practices, policies etc.

Move away from a dependency on Microsoft: - currently code is written mostly in VB and C++. There is some python, but not much.

The background languages here is python and C++.

Question: Is there a benefit to making this switch? I do not believe that the licensing will change, but it can give us one source code for all platforms.


The push for the change is not due to the backend, but for the client interface and the license costs.

3
  • 4
    Is there a benefit in keeping ESRI geodatabase? Just get rid of all of ESRI stuff. Also we are not aware of your workflow so it is somewhat impossible to advise something here. Maybe in its current state your question could be flagged as too broad... Nov 3 '14 at 18:32
  • 3
    What exactly does "have a more restrictive development framework" mean?
    – KHibma
    Nov 3 '14 at 20:05
  • 1
    As posed, this question can't be answered well in our framework. It will collect opinions, which can be good and useful, but won't get "answered", as we use the term here. A better approach would be along lines of: What needs to be studied, what specific questions need to be asked, what workflows analysed, in order to evaluate whether it's worthwhile to change our GIS stack from ArcGIS to Open Source? Nov 22 '14 at 1:29
11
votes

It's not easy just to switch to QGIS if you've decided to stay invested in the Arc-Geodatabase. I would say don't switch. You'll have more trouble mixing the technologies than it would be worth.

The ESRI Geodatabase is a database meant to work with the ESRI platform. Although there is a 'plugin' for QGIS to use a file geodatabase, there is (as suggested above) no editing capabilities within QGIS, and I see many bugs reported on this forum relating to QGIS and the File Geodatabase. (Additionally, see the note at the bottom of this reply on simply moving workflows to open source...)

For us, we're switching to a spatial database workflow, rather than a geodatabase workflow. This means moving data into Microsoft SQL Spatial so we can use the SQL spatial capabilities of a spatial database. This workflow is based on PostGIS and an enterprise system where spatial data is treated like any other kind of data (versus a model where spatial data is the main data type, and anything else is just an 'attribute').

However, although we're still able to utilize the power of the spatial database, we can still bring the results of queries, custom spatial tables built on SQL views containing spatial data, etc., into ArcMap for visualization and other geoprocessing analysis, as well as publishing these tables to ArcGIS Server.

Why can't we move to open source? Our student information system is based on Microsoft SQL Server. Switching to a full open source stack would cut the ties to this system (for example: PostgreSQL doesn't have a MSSQL wrapper, and I don't want to pull millions of records into Postgresql on a regular basis when the proprietary stack we have works just fine for what we have). If I had my way, the entire system would reside in open source, from the database to the desktop 'GIS' client, to the map server, to the front end. I don't even want to use the Leaflet API for ArcGIS server because it doesn't have query tasks!

There's lots of talk of blending the open source and proprietary models, but I'm not for it. As I've said on this forum before, simply switching the workflows of ArcGIS to open source software doesn't give the full power of the open source geospatial workflow a system like PostGIS offers...

6
votes

Benefits for switching:

  1. No license cost per user!
  2. May integrate with other open source apps. (e.g. GeoServer, MapServer, PostGIS..etc)
  3. Plug-in language built on C++/python (pyQGIS)
  4. Not as buggy

Disadvantage for switching:

  1. Learning curve for users and developers
  2. Cannot edit File GDB data, will have to switch to PostgresSQL/PostGIS or SQLite DB
  3. Migrating scripts from arcpy to pyQGIS
5
  • Great points here. We would still have licence cost for the arc runtime, per user. So I believe that it will not change that cost as much. Unless someone can say correcet me.
    – Omnia9
    Nov 3 '14 at 18:59
  • 1
    I am sorry but this is the general anti-ESRI nonsense that dominates this forum. (4) Not as buggy. QGIS is far more "buggy" than Arc. (3) What about ArcPy. (2) ArcGIS interfaces with many Open Source Apps. Look I am as big a fan of QGIS as anyone but the anti-ESRI nonsense on this board is just tiresome. Nov 4 '14 at 3:19
  • 1
    @user two seven two three nine, I'm not anit-ESRI, but just relaying my experience with the two software's.
    – artwork21
    Nov 4 '14 at 12:48
  • 2
    yep apologies, I am not anti-FOSS. Indeed GDAL QGIS SAGA and WHitebox are stunning but I just get a little annoyed when people state ArcGIS is more "buggy" than QGIS, more bloated no-doubt, but I just do not see how it is more buggy. I think it actually does a reasonable job of using Python and it interfaces pretty well with some open standards and created the singular biggest open gis standard (shapefile). Nov 4 '14 at 14:50
  • 2
    -1 for FLOSS not as buggy. They both have significant roots to trip over. The frequency of encountering them depends on the paths one wanders. What FLOSS does have in spades, & definitely resides in plus column, is bug transparency. It is aggravating that many or most issues I report to Esri are already logged (have a NIM#), but that doesn't come to light until after going through the rigmarole of building a reproducible test case and submitting it. IMO careful analysis of the ArcGIS and Open Source "surface areas" (points of user & data contact) would find "arcgis buggier" perception false. Nov 22 '14 at 1:20
3
votes

The two major reasons (of many) to do this are:

  1. Cost

Software license cost. QGIS will provide far lower cost software purchase price but you will have to weight this alongside staff training costs (will you need to retrain everybody?), support costs (currently ESRI provides software support as part of their license), programming costs (do you have hundreda of scripts that will need to be rewritten?) and other more difficult to measure FOSS costs. Just because licenses are free does not mean costs are always lower.

  1. Customization and Ownership

FOSS software generals allows more modification than commercial OTS GIS products such as ArcGIS. If ownership of code and modification rights are important (and sometimes code transparency although in truth many FOSS are worse than ESRI on this) then switching may become more attractive.

If you are a large enterprise I would recommend a full requirements and cost analysis.

As the Rebel states I would wonder why you would then remain with the ESRI geodatabase. Surely switching to an FOSS spatial DB would be the way to proceed if you transition.

5
  • A number of our scripts are alredy in python and some C++, so script runner could handle that. But im sure there will be cases where it has to be redone. There would be a 'cost' for the team learning to develop in QT/QGIS for sure. Your comments are great.
    – Omnia9
    Nov 3 '14 at 18:57
  • ArcPy or Python? Nov 3 '14 at 19:13
  • python, and I'm reading that it is different.
    – Omnia9
    Nov 3 '14 at 19:14
  • +1 for your comment on code transparency. You may want to read a blog that I wrote on this very topic: whiteboxgeospatial.wordpress.com/2014/05/04/… Nov 3 '14 at 20:51
  • 3
    I almost wrote that WhiteBox is an exception to this but that is why Whitebox exists. In my humble opinion one of the incorrect critiques of ArcGIS is that the processes are "less open" or "less documented" than most FOSS. Indeed I often find ArcGIS gives excellent citations and transparency on geoprocessing algorithms and FOSS does not and other times FOSS provides more information and ArcGIS does not. Both "could do better" and as an industry we should do better. Nov 3 '14 at 21:00

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