Latest Contributions by VinceAngelo

Bind variables just make a certain class of very inefficient queries more efficient.  ArcSDE was altered to support bind variables when needed, but I don't think ArcObjects has a generic interface to use them.  Most of the situations I've seen where bind variables would help would have been helped more by different use of SQL -- if you're generating hundreds or thousands of new queries, there's probably a more efficent way to drive the query so that you're only doing multiple fetches. - V ... View more

Not directly. - V ... View more

You have a lot fewer options than if the column was ST_GEOMETRY or even SDO_GEOMETRY (which doesn't require Spatial -- just Intermedia).  You'd get a lot better performance if you added sort_key or lat/lon columns to the business table.   What application will be making the query?  And what version?  (Accessing shape properties is easy from ArcGIS [and unsupported from other applications], but doing an ORDER BY in ArcGIS is difficult.) - V ... View more

What database are you using?  What is the geometry storage type?  Does it have SQL accessor functions that would allow access to the location (point) or envelope corners (non-point)? - V ... View more

What version of ArcSDE exists in the database?  What version of ArcSDE are you trying to install? Are there tables which have been registered with either ArcSDE or the geodatabase, but then deleted by a means which would not delete orphaned metadata records? - V ... View more

Generally speaking, you *really* want your Direct Connect clients to be the same *exact* release as the database (same service pack *and* patch).  There is some room for backward compatibility, but you need to follow the documentation closely in these cases. Oracle 10.2.0.1 is not supported for use with ArcSDE (or any other Esri product) -- It is known to be unstable and to fail frequently.  The minimim supported Oracle release is 10.2.0.3 (which has been very stable).  The only exception is Oracle Express 10.2.0.1 (which was never upgraded), but support for that wasn't introduced until 9.3.1. Note that Oracle no longer supports Oracle 10g, so support from Esri will be discontinued at a later release of ArcGIS as well. - V ... View more

The Installation Guide provides this information. - V ... View more

I ran your SQL on 32-bit and 64-bit Linux hosts (virtual and physical, respectively) running Oracle 11.1.0.7 with ArcSDE 9.3.1sp2 (32-bit) and ArcSDE 10.0sp0 (64-bit) and saw execution times of 7.5 and 1.6 seconds. Whatever the cause for the difference, it certainly doesn't look like anything that Esri can control. - V ... View more

ArcSDE requires the left and right curly braces ('{6c1d6f56-b4d9-43ba-ac0a-038699229b6b}'). Without knowing what version of ArcGIS you're using, and what kind of data source, it's difficult to give any other guidance. - V ... View more

I don't recommend you follow a ten year old tuning guide. There are plenty of modern tuning steps listed in the documentation. - V ... View more

It's usually best to only have one independent variable when doing statistical evaluations. You've got closer to five (host architecture, operating system, wordsize, Oracle release, and ArcSDE release).  Evaluation of 1500 *different* shapes would be less of a caching exercise -- do the two machines have the same amount of RAM and identical Oracle memory tuning?  Are the character sets identical? If one was using 16-bit characters and the other 8-bit, that could easily account for the 2x performance difference. Did you run each query locally, or from a 3rd host to control for remote client use? How are you capturing the benchmark duration?  Unix has the 'time' utility, but then you're also measuring the time to start the sqlplus utility.  The 'sdesql' utitlity of se_toolkit has a subsecond resolution timer (millisecond resolution on Windows, microseconds on Unix), but it doesn't support PL/SQL directly. It would probably be more interesting to see 150,000 inserts, indexing, and 50 random spatial queries on 64-bit/RISC/Unix, 32-bit/RISC/Unix, 64-bit/Intel/Windows, and 32-bit/ Intel/Windows, all with the same RAM and system load and using Oracle 10.2.0.3 (same CPU) and ArcSDE 9.2 (same service pack & patch).  If you go one step further, and run Linux on the Windows host, you'll come the closest to an apple-to-apples comparison. - V ... View more

Without the LOCAL variable, service start should fail. If your service started correctly, there shouldn't be difficulty connecting through it.  You really need to take this to Tech Support for resolution. - V ... View more

I'm pretty sure the CAD data isn't carried in a SQL-accesible form (I don't see any accessor functions in the SQL standard). Do your tables have a parallel SE_ANNO_CAD_DATA BLOB column? If so, are any of those rows NULL? - V ... View more

You've got a bunch of issues here: ArcGIS 10 requires HIGH precision on SDO_GEOMETRY and ST_GEOMETRY layers -- No workaround. Versioning requires an SDE-set rowid column -- No workaround. "Move edits to base" versioning requires ArcGIS (both to establish it, and to edit it) -- No workaround. I'm afraid you're going to have to alter your plan.  You might be able to use Oracle STREAMS to propagate changes made in a parallel ArcGIS-registered table, but you'd be hard-pressed to get propagation to be two-way. 😞 - V ... View more

I'm confused. You write that that your dbinit.sde has ORACLE_SID and that the service starts and then stops (which is the expected result -- Use LOCAL as the variable), and that sdemon and sdeservice report the service up and running -- which is it? Application server connects are made to the ArcSDE host, not the Oracle host. Even if you're running locally, an enabled firewall may block connection when you refer by the local IP. You really should use robust passwords for administrative database users (and try to avoid publishing them on the Internet). I try to avoid connecting as the SDE user, and never use SYSTEM to connect via ArcSDE. Best practice is to create at least one tablespace to store data and create one or more database users to own spatial data (with default tablespace to (one of) the new tablespace(s)).  I usually create a "read-mostly" user, which is granted only SELECT access (via a "browse" role) to the spatial tables, and use that to connect for mapping applications. - V ... View more