Latest Contributions by VinceAngelo

You should set the CONNECTIONS parameter to the number you need (plus a few more). The N-I desktop heap issue is documented in the Knowledge Base (found by search on "arcsde desktop heap"). It only applies to application server connections, so you'd need 110+ 3-tier connects before you'd need to take more exotic action. - V ... View more

No, each instance is independent with respect to connection count. However, the Windows heap issue is cumulative. - V BTW: The 'sdeping' utility of se_toolkit can be used to diagnose connection count issues -- It has a "-c" option to make multiple connections and a "+DELAY" option to hold onto them (if it succeeds): C:\>set SDEUSER=XXXXX C:\>set SDEPASSWORD=***** C:\>set SDEINSTANCE=5153 C:\>sdeping -c 80 -v +DELAY 17:10 ArcSDE 9.3 Connection Test Utility       Thu Jun 10 17:05:23 2010 ------------------------------------------------------------------------          Server: YYYYY        Instance: 5153        Database: eval93        Username: XXXXX           RDBMS: PostgreSQL         Version: 9.3.1     Description: PostgreSQL Build 2784(PostgreSQL)      Build Date: Tue Oct 27 10:51:14  2009         Release: 93005     Connections: 1   2   3   4   5   6   7   8   9   10  11  12                  13  14  15  16  17  18  19  20  21  22  23  24   Connect Delay: 333.33 ms (avg)   Connect Stats: 312.00 ms (min) / 656.00 ms (max) / 69.32 ms (stddev)      Connection attempt 25 failed --           Error: Operation Failed (-1)   Disconnecting: 24  23  22  21  20  19  18  17  16  15  14  13                  12  11  10  9   8   7   6   5   4   3   2   1 ... View more

The SDE.PROCESS_INFORMATION table tracks connections.  You can use a number of ways to process this information (including 'sdemon', SQL, and a trigger on the table). - V ... View more

Yes, absolutely, the CONNECTIONS parameter controls the number of total connections (sum of app server and Direct Connect). You can update the value with the 'sdeconfig' utility or by correctly updating the row in SDE.SERVER_CONFIG. Keep in mind that Windows hosts have their own upper limit of application server processes (the "non-interactive desktop heap" issue), which is independent of both Oracle and ArcSDE (generating a strange error message somewhere after 66 connections on boxes without the registry hack, and ~110 connections with it). - V ... View more

That product is documented as being limited to 3 seats (see the table two pages down under "How is ArcSDE technology included in ArcGIS?"). - V ... View more

SQL-Server is certainly more forgiving than Oracle, but it also has a completely different storage architecture.  It still takes a little work to obtain optimal performance; I'd like to say I'm amazed that more folks don't make the effort, but when ease of administration is emphasized, tuning is always going to take a back seat. - V ... View more

Which version of ArcSDE did you license?  "Personal ArcSDE" and "Workgroup ArcSDE" have concurrent user limits, while Enterprise ArcSDE does not (well, it does, but it's database resource configuration which is the determining factor -- once you're into hundreds of users, it doesn't really seem like a limit). - V ... View more

You need to upgrade your client to at least 10.2.0.3 (the minimum), or all to the same release as the server 10.2.0.4.  ArcSDE will not function correctly with Oracle 10.2.0.1, which is known to give spurious permission errors and to fail frequently. - V ... View more

You cannot control the allocation of SDE-set registered rowid column values in geodatabases and "in-memory workspaces" aren't, as James pointed out, full geodatabase implementations.  The task of keeping track of pieces of a database, editing them, and merging the result back into the original table is quite challenging, which is why the Disconnected Editing solution was developed.  But you can't pick and choose pieces of that solution (the restrictions it imposes were necessary to make the solution work).   If Disconnected Editing doesn't meet your requirements, you'll have to decide if the days, weeks, or even months of effort to make your own in-memory solution are worth your time, but your design will have to actively track previous objectid values in an additional column, and be prepared to update, delete, and insert appropriately when the transition back to persistent geodatabase occurs. Good luck. - V ... View more

I'd have to see (and reproduce) research that showed that using multiple filegroups (I leave the primary to the database, place sde in one, raster in two, and vector in at least two) hurt performance before I stopped using them.  Using one big file puts you at the mercy of fragmentation, and doesn't allow you to split data across multiple independent disks.  You can always defragment the files on disk if you've got them. Just my $0.0204 (with inflation). - V ... View more

GRANT OPTION is a SQL standard thing, required by Oracle. It is necessary when a non-owner attempts to grant other users access to a table with a view. See the Oracle docs for more info. - V ... View more

You can't control the objectids in either location, and can only keep track of old IDs and edits through a great deal of effort.  It's generally easier to use ArcGIS as it has been designed, so I think you should look into the Disconnected Editing model. A search on "Disconnected Editing" should pull up a wealth of information. - V ... View more

It looks like the search engine issues are closer to being resolved, since a search from http://support.esri.com on "st_geometry trigger" turned up this in the first page.  You'll also need to maintain the objectid appropriately (I recommend a USER-set registered column, but then it can't participate in advanced geodatabase functionality). - V ... View more

The only way the objectids could be consistent is if you a) selected all features, b) inserted them in objectid order, and c) the source dataset started at 1 and did not have any gaps.  If you want to preserve the old ids, you'll need to insert them as an additional column (e.g., "prev_oid integer not null"), since there are no guarantees about SDE-set objectid allocation (except that they will be unique non-zero positive values). - V ... View more

It would be quite wasteful to update the layer extent every time a geometry is inserted.  The layer load utilities generally maintain an envelope of the  newly added features and update once at the end of loading (and only if it would change the extent). I usually set my envelope to the full extent at layer creation, so it doesn't get updated (which also has the side effect of always having the same display extent in ArcCatalog for all layers).  The 'sdelayer -o alter' command allows you to set an extent ('-E minx,miny,maxx,maxy') or calculate one ('-E CALC').  I recommend the former. - V ... View more