@ggarzaIt took longer than expected, but in FME 2019.1, SQL Server (JDBC) Spatial reading is 3x faster than before (when reading single points).I'd be interested in hearing your "real world" results if you get the chance to report back.
Thank you for the feedback!
Thanks for the extra details!We will improve for FME 2019.0, and see if a 2018.x backport makes sense as well.
@ravenkopelman,here are the answers to your questions:
Yes, I can elaborate.The workspace reads 4 different feature types from a SQL Server 2016 database, passes the data through an AttributeManager (to set attribute) and TestFilter (to split data stream), and outputs to a FGDB Open API writer with 4 feature types.
1.Besides the geometry spatial type, the other types are varchar, int, float, and datetime2.There are no more than 35 attributes in each table.The varchar fields have lengths less than 15 characters.
2.No, a non-spatial table does not have the same problem.I did some further testing using the same tables with only the geometry column removed, and the same data output in one minute flat.空间部分似乎是问题.
3.They are close.They are virtual machines on the same subnet.
Thank you for the (surprising and disappointing) feedback@ggarza。
Can you elaborate on your situation so that we can do comparable experiments?
If we can reproduce this we will make it faster.
Hi@ravenkopelman。Unfortunately, the JDBC SQL Server读者空间is still slow in FME 2018.1.I am running a simple workbench reading from a SQL Server database into an ESRI File Geodatabase (File GDB Open API Writer).The JDBC reader workbench took 20 min to output 175k records.当我换成普通的SQL Server空间阅读器的读者,花了不到2分钟。我知道这个线程是对作家,但我想确保你都知道对读者也是如此。
Thanks for including my case, it will be great to be updated when you get performance improvments.
Thanks for the feedback, I will await the changes in 2018.
Will look at work arounds for now.Using FME Cloud limits us to JDBC only unfortunately.
we are sorry that for this limitation.As@ravenkopelmanalready mentioned, improvements regarding this are on our roadmap!I created a support case on your behalf to make sure you'll be kept up to up to date regarding this.
Thanks
Hi Oliver, this大多comes down to the use (or lack thereof) of bulk interfaces for writing data to SQL Server.We have tentatively scheduled this for improvement in FME 2018.1*;if you want your vote formally counted please create a support ticket and reference PR#61819.Regardless I will try to remember to post an update here when the work is complete.
Thank you for your feedback.
* Our previous goal of 2018.0 had to be revised while we wait for Microsoft to upgrade their bulk loading driver to support spatial columns.