Hello, Peter.
My own personal experience ...
We started off connecting Geoserver to MS SQL Server tables to deliver KML
to Google Earth. This was because SQL Server was a corporate direction,
and consequently our data warehouse sits in SQL Server. I was aware that
performance was worse than that of PostGIS but persisted for while, but
then we had an upgrade to 64-bit SQL Server and performance halved. I then
got approval to install PostGIS on the same servers that were running
Geoserver. We use FME to replicate data from SQL Server to PostGIS on a
nightly basis, but we are copying only the data that is required (eg.
attributes required for styling and filtering) into PostGIS - so we are
really just using PostGIS to hold simple 'layers' (in GIS-speak) rather
than for a relational data structure. In some ways conceptually we are
using PostGIS as an alternative to holding layers in shapefiles. When I
explained it that way, the database people were happy.
The advantage of replicating the data into the Geoserver/PostGIS system, is
that you can trim it right down - it is lean and fit for purpose - so you
should get good performance.
Our systems have been running for about 5 years, and we never have to do
anything much with PostGIS.
Regards,
David Collins
Senior Geoscientist
On Wed, Jun 24, 2015 at 9:23 PM, Peter Kovac <peter.ko...@microstep-mis.sk>
wrote:
> Dear GeoServer users,
>
> my company develops systems in meteorology, climatology, radiation
> monitoring etc. It's mostly data collection, storing and later retrieval.
> Our customers started to demand maps, recently.
>
> What we need is a map client serving customers' data, both rasters (e.g.
> forecasts from GRIB files) and vector (e.g. various values measured by
> stations from a national climatology network). We looked out for some
> options and settled on a combination of a GeoServer & OpenLayers
> integrated into our existing web interface.
>
> After several weeks of playing with GeoServer I got an impression that
> GeoServer works great in combination with a local PostGIS instance.
> PostGIS fits well into our scenario, too: it would contain metadata about
> our rasters with time dimensions, and whole vector layers.
>
> The problem is we already have our own infrastructure which does all the
> bookkeeping of all our datasets. E.g. we already store our raster data
> somewhere in the filesystem and there is a table in our database which
> tracks metadata (origin, time, etc.) about these rasters. For climatology
> data, we use Oracle database and have our own schema which stores
> information about weather stations (e.g. name, location) and current and
> historical measured values.
>
> Because of this, my manager does not want to allow me to create a PostGIS
> instance along with a GeoServer instance. His arguments is it's
> unnecessary duplicity, it won't be consistent with our database unless we
> invest heavily in syncing the two and it needs additional maintenance. His
> original idea was that our system will provide data to a GeoServer
> instance via some Web Service / WFS / whatever and GeoServer will just
> "render" it. The problem is our system does not have a WFS interface
> (yet?) and our existing interfaces are not OGC standardized. Of course, if
> we implement WFS we won't need GeoServer (mostly).
>
> I tried to explain to him that we should treat the PostGIS store as some
> kind of internal GeoServer store (just like various index files on disk -
> he is a database guy and does not care about all those files created by
> GeoServer - he cares about databases) and not try to mess with it in order
> to speed up development (deadlines are tight, as usual).
>
> I need either better arguments in this debate or an alternative solution.
>
> I stumbled upon DataStore Development tutorial on GeoTools (
> http://docs.geotools.org/latest/tutorials/advanced/datastore.html ) which
> looks promising, but I'm not quite sure if it's the right thing. I'll try
> to implement a proof-of-concept to see if I'm able to convert our existing
> database schema into a new Data Store type recognized by GeoSever.
>
> In the meantime I'd love to hear your opinion about the case. Is it common
> to implement custom DataStores from existing databases or mirroring it in
> PostGIS is the preferred way? Are there any other options?
>
> --
> Peter Kovac
> IMS Programmer
> MicroStep-MIS
> peter.ko...@microstep-mis.sk
>
>
>
> ------------------------------------------------------------------------------
> Monitor 25 network devices or servers for free with OpManager!
> OpManager is web-based network management software that monitors
> network devices and physical & virtual servers, alerts via email & sms
> for fault. Monitor 25 devices for free with no restriction. Download now
> http://ad.doubleclick.net/ddm/clk/292181274;119417398;o
> _______________________________________________
> Geoserver-users mailing list
> Geoserver-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/geoserver-users
>
------------------------------------------------------------------------------
Monitor 25 network devices or servers for free with OpManager!
OpManager is web-based network management software that monitors
network devices and physical & virtual servers, alerts via email & sms
for fault. Monitor 25 devices for free with no restriction. Download now
http://ad.doubleclick.net/ddm/clk/292181274;119417398;o
_______________________________________________
Geoserver-users mailing list
Geoserver-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-users
