POST
|
Hi, after some further tests we found out that this behaviour - slow display performance in combination with oversized bounding boxes - is a bug that was introduced after 10.2.1 UTUP3. UTUP3 is fine, UTUP5 and later show this problem. This issue is probably only relevant with a large number of layers (> 100) and at large scales, but in our case, screen refresh times are more than doubled with UTUP7 compared to UTUP3. Tech support has created BUG-000109043 for this issue. Martin
... View more
11-09-2017
01:25 AM
|
3
|
1
|
903
|
POST
|
Have a look at PerfQAnalyzer, this can do a lot more than MxdPerfStat. Details at PerfQAnalyzer: New 10.5/10.5.x Version (Build172) Available for Download | ArcGIS Blog Martin
... View more
09-12-2017
02:40 AM
|
1
|
0
|
2879
|
POST
|
Hi, we have a utility customer who has significantly different display performance in ArcGIS 10.2.1 (the "utility version") when displaying the same MXD on two different client machines for the same ST_GEOMETRY feature classes in the same database (Oracle 12.1). We checked the usual suspects (network latency, CPU speed etc.), but we found from Oracle traces that Oracle does significantly more work when the Mxd is displayed from client 1 than from client 2 (data is in the buffer cache for both cases, so no disk reads in either trace). We then looked at the spatial queries which look basically like this: SELECT ... FROM MyFeatureClass WHERE SDE.ST_EnvIntersects(MyFeatuerClass.SHAPE,:1,:2,:3,:4) = 1 Bind variables :1 through :4 contain the bounding box coordinates in map units, and we found that ArcMap generates significantly different bounding boxes for the same Mxd from the two different clients. The Mxd has group layers, and the effect is different for the different group layers: Layers in one group layer show very little difference, whereas another group layer has bounding boxes that are 8 times as wide (and 8 times as high). I would have thought that ArcMap basically uses the visible map extent as a bouding box, and that is what we see on the fast client, but where does it get the large values for the slow client? I'd open a tech support case, but I have no idea how to reproduce the problem ... I have plenty of details if anyone is interested... Thanks for any ideas, Martin
... View more
08-14-2017
06:20 AM
|
0
|
2
|
1623
|
POST
|
Hi Derek, thanks again - seeing a bit clearer now. I will do some experiments now, and if there are any more questions, I'll let you know. Martin
... View more
03-15-2017
06:16 AM
|
0
|
4
|
5042
|
POST
|
Hi Derek, thanks for your reply - that clears things up a bit, if not completely. I have read the documentation and I appreciate the need for abstraction at this level. However, as someone who spends a lot of time troubleshooting performance issues, tuning databases, networks, disk access etc., I find that this abstraction is not enough to understand what to do if there are problems. So from your answer I understand that the "ArcGIS Data Store" is basically a Postgres relational database that runs "under the hood" of ArcGIS Server and is completely managed by the software, i.e. there is no interface for an administrator of troubleshooter to look into if there are problems - correct? Thanks again, Martin
... View more
03-14-2017
08:08 AM
|
2
|
6
|
5042
|
POST
|
Hi, I'm getting a bit lost in the new ArcGIS Enterprise terminology - can someone point me to some document that explains what the ArcGIS Data Store and specifically the relational data store have to do with the good old Enterprise Geodatabase based on RDBMS products such as Oracle, MS SQL Serverv or Postgres? Is this just a new name for the same thing, or is a relational data store truly a new relational database implementation? From the current documentation I get the rough idea the a "managed database" is the old (ArcSDE) Enterprise Geodatabase, so this would indeed indicate that the relational data store is something new. If so, what? Thanks, Martin
... View more
03-13-2017
05:28 AM
|
4
|
19
|
16709
|
POST
|
Hi Chris, we have a number of customers on ArcGIS 10.2.1 that use a seperate viewing database in addition totheir primary editing database. Some of them use Esri geodatabase replication, others use database mechanisms (mostly Oracle). Both methods are viable, here are some considerations from our experience: Changing the replica definition with GDB replication can be a complex process. We find that the schema change propagation mechanisms tend to be a bit unstable with large schemas, so we generally recreate the replica when there are siginificant schema changes. While the replication process itself is fairly mature, we did run into a couple of bugs involving geometric networks that were very difficult to identify, reproduce and fix (all known problems have been fixed in 10.2.1 UTUP 7). We typically have a mix of versioned and non-versioned data in our databases, so when we us GDB replication, this has to be supported by some way of replicating non-versioned data - we use Oracle Materialized Views and Database Links for this purpose. When using SQL/Database replication technology, I suggest copying the whole database (e.g. using export/import) rather than using differential methods (like Materialized Views) in order to avoid confusion with versioning information. Martin
... View more
02-16-2017
02:28 AM
|
0
|
1
|
902
|
POST
|
Hi, we're using ArcMap 10.2.2 (on various Windows platforms) with filebased GDBs that sit on a Windows fileserver (running WS 2008 R2 or WS 2012 R2, doesn't seem to make a difference). Initially, performance is adequate, but sometimes our workflow requires that we make a copy of the file GDB while ArcMap is running, and after that, access to the database slows down dramatically. It doesn't seem to matter whether that copy is made from ArcMap or from another process, and stopping that second process doesn't improve things, either. We did some research into this phenomenon (with Sysinternals Process Monitor) and it looks to me like a second access to the various files seems to change the caching behaviour of ArcMap: File access times increase significantly, and network traffic increases. I realize this is more of a windows than an ArcGIS question, but maybe someone has come across this before, or can point me to some ressource that can explain and preferrably prevent this behaviour. Restarting ArcMap solves the problem, but given the workflow context, that's a very cumbersome workaround that makes a lot of people very unhappy... Thanks, Martin
... View more
09-23-2016
07:24 AM
|
0
|
0
|
1236
|
POST
|
I'm currently researching a problem (on ArcGIS Desktop 10.0 and Oracle 11.2) where we have afeature class with global IDs that shows the wrong field type (Guid rather than Global ID) in ArcCatalog and that also has lots of identical global IDs that look like this: {00000000-0000-0000-0000-000000000000}. In trying to find out if any other feature classes and tables in my database are affected (it's a big schema), I came up with the following SQL query: SELECT name, type FROM (SELECT name, Extractvalue(VALUE(d), '/GPFieldInfoEx/FieldType') type, Extractvalue(VALUE(d), '/GPFieldInfoEx/Name') colname FROM sde.gdb_items_vw items, table(Xmlsequence(Extract(Xmltype(items.definition), '/DEFeatureClassInfo/GPFieldInfoExs/GPFieldInfoEx'))) d WHERE items.type = '{70737809-852C-4A03-9E22-2CECEA5B9BFA}') WHERE colname = 'GLOBALID' AND type <> 'esriFieldTypeGlobalID'; This returned a lot more matches than I expected, one of them being a feature class called FCL_ACCESS. I then looked up FCL_ACCESS in ArcCatalog to make sure that my SQL was OK, but ArcCatalog shows that attribute GLOBALID is of type Global ID. In order to make sure that I didn't have a mix up somewhere, I looked up the OBJECTID of the feature class description with SQL, extracted the XML definition using sdexml, fired up an XML editor and found the SQL information confirmed (attribute GLOBALID has type esriFieldTypeGuid rather than esriFieldTypeGlobalID). Also, the XML descrption showed the attribute HasGlobalID to be false. Now, I'm fairly sure that I'm not mixing up databases here, so why does ArcCatalog tell a different story from GDB_ITEMS?? Mystified, Martin
... View more
09-21-2016
08:07 AM
|
0
|
0
|
1083
|
POST
|
Hi George, as far as I can see, the Export/Import Geodatabase Configuration Keywords GP tools are a replacement for sdedbtune rather than for sdeconfig, no? So is sdeconfig considered obsolete? Thanks again, Martin
... View more
09-09-2016
06:27 AM
|
0
|
4
|
1203
|
POST
|
Hi George, I've noticed that sdeconfig seems to have been left out of the nw GP commands. Is that because there's no need for this any more? Apart from the obvious connections parameter, we used to increase various buffer sizes from their default values and sometimes increased MAXGRIDSPERFEATURE when we had really crazy data in customer databases. So uis this dropped altogether? Thanks, Martin
... View more
09-08-2016
12:17 AM
|
0
|
6
|
1203
|
POST
|
Hi, we've been using MxdPerfStat for years to analyze slow MXDs for years (mostlym on Oracle databases), and it's a great tool. There's one thing that I don't quite understand, though: Whenever MxdPerfStat finds a layer with a definition query that refers an attribute that doesn't have an index, there's a warning ("check if index exists on attribute..."). The same warning comes up when analyzing an MXD for publishing to ArcGIS Server. Why is that? In my experience, attributes used in definition queries tend to be non-selective, with a handful of different values (status, pressure level, type of building, annotation classid...) shared between many features. The database almost never uses these attributes to access the data for display in ArcMap since the spatial query is invariably more selective (unless the feature class is very small, and then indexes don't matter anyway). I've even seen plenty of cases where the presence of an attribute index makes Oracle use that index, resulting in much slower query execution. Are our databases special in the we don't need these indexes and everybody else is happy with them?? Thanks a lot, Martin
... View more
08-12-2016
06:07 AM
|
0
|
1
|
1067
|
POST
|
Hi, here's an update: Tech support replied on our network problems: That was fairly easy, we had done the simplify on the original shape rather than getting a clone of the shape, simplifying that and then assigning it to the feature and storing it. Switching to that second method seems to fix the network problem. Also, we find that some features are split inti multipart features by the simplify, which the network doesn't like, but we can skip and log those features and handle them by hand. As to the other suggestions: The duplicate features are part of a geometric network in a large production geodatabase, and while extracting a large number of network features and the reloading them might be possible in theory, it's not something I'd really like to try. And no, having this many features with duplicate vertices is not a common occurrence but rather very specific to one customer situation. We've started cheking on other databases and generally find a handful of features with "real" duplicate vertices (rather than vertices that are very close) in many databases, but these tend to be a few and can be fixed by hand, if one so desires. Martin
... View more
07-13-2016
12:52 AM
|
0
|
0
|
2500
|
POST
|
Wrong. It does hurt the geometric network. We've contacted tech support about this... Martin
... View more
07-04-2016
11:47 PM
|
0
|
2
|
2500
|
POST
|
Hi Joshua and Christian, thanks for your replies! It very much looks like the ST_GEOMETRY WKT functionality works the same as the Python WKT property (tech support should know for sure), and I wasn't aware of the JSON property of the python geometry type - this has the real geometry and can be parsed to identifiy duplicate vertices. However, this seems to be new with 10.3 as well (like the Data Reviewer checks we're suing now), so it does look like there is no out-of-the-box functionality for identifying duplicate vertices in 10.2.1. Thanks again, Martin
... View more
06-17-2016
02:02 AM
|
0
|
0
|
1177
|
Title | Kudos | Posted |
---|---|---|
1 | 02-18-2015 12:22 AM | |
1 | 09-12-2017 02:40 AM | |
1 | 03-26-2015 01:27 AM | |
1 | 09-29-2014 07:14 AM | |
1 | 09-17-2014 02:17 AM |
Online Status |
Offline
|
Date Last Visited |
11-11-2020
02:23 AM
|