I am trying to migrate the entire backend of an Access application onto SQL Server. The first part of my project involves moving all of the tables whilst making minimum changes after the migration (no SQL Views, Pass-through queries etc. yet).
I have two queries in particular that I am using here:
ProductionSystemUnAllocatedPurchases - Which executes and returns a resultset successfully.
This is the full formula (sorry its extremely complex) for QtyAvailableOnPurchase:
QtyAvailableOnPurchase: I believe this field could be the problem here?
IIf((IIf([Outstanding Qty]>([P-ORDER-T with Qty Balance]![QTY]-[SumOfQty]),
([P-ORDER-T with Qty Balance]![QTY]-[SumOfQty]),[Outstanding Qty]))>0,
(IIf([Outstanding Qty]>([P-ORDER-T with Qty Balance]![QTY]-[SumOfQty]),([P-
ORDER-T with Qty Balance]![QTY]-[SumOfQty]),[Outstanding Qty])),0)
ProductionSystemUnAllocatedPurchasesTotal - Which gives an 'Invalid Operation' error message
Now the strange thing for me is that the first query works perfectly fine, but the second which uses the first as a source table, gives me this error message when executing. This query works perfectly fine with an access backend, but fails with SQL Server tables.
Any Ideas?
Can QtyAvailableOnPurchase be NULL? That would explain why Sum fails. Use Nz(QtyAvailableOnPurchase,0) instead.
My approach is to decompose queries. Create two queries :
First query selects needed data
Second query applies group operations (e.g. Sum)
You'll get easy way to check every step.
I have managed to find a solution to this error.
It seems that the problem is not so much with the query but rather the data type on SQL Server. SQL Server Migration Assistant (SSMA) automatically maps any Number (Double) fields to float on SQL Server. This mapping needed manually changing to Decimal.
Now according to this SO post, Decimal is the preferred for its precision up to 38 points (which is more than enough for my application), While float allows more than this, the data is stored in approximates.
Source: Difference between numeric, float and decimal in SQL Server
TL;DR
I want to migrate data from a MS SQL Server + ArcSDE to a PostgreSQL + PostGIS, ideally using SQLAlchemy.
I am using SQLAlchemy 1.0.11 to migrate an existing database from MS SQL 2012 to PostgreSQL 9.2 (upgrade to 9.5 planned).
I've been reading about this and found a couple of different sources (Tyler Lesmann, Inada Naoki, Stefan Urbanek, and Mathias Fussenegger) with a similar approach for this task:
Connect to both databases
Reflect the tables of the source database
Iterate over the tables and for each table
Create an equal table in the target database
Fetch rows in the source and insert them in the target database
Code
Here is a short example using the code from the last reference.
from sqlalchemy import create_engine, MetaData
src = create_engine('mssql://user:pass#host/database?driver=ODBC+Driver+13+for+SQL+Server')
dst = create_engine('postgresql://user:pass#host/database')
meta = MetaData()
meta.reflect(bind=src)
tables = meta.tables
for tbl in tables:
data = src.execute(tables[tbl].select()).fetchall()
if data:
dst.execute(tables[tbl].insert(), data)
I am aware that fetching all the rows at the same time is a bad idea, it can be done with an iterator or with fetchmany, but that is not my issue now.
Problem 1
All the four examples fail with my databases. One of the errors I get is related to a column of type NVARCHAR:
sqlalchemy.exc.ProgrammingError: (psycopg2.ProgrammingError) type "nvarchar" does not exist
LINE 5: "desigOperador" NVARCHAR(100) COLLATE "SQL_Latin1_General_C...
^
[SQL: '\nCREATE TABLE "Operators" (\n\t"idOperador" INTEGER NOT NULL, \n\t"idGrupo" INTEGER, \n\t"desigOperador" NVARCHAR(100) COLLATE "SQL_Latin1_General_CP1_CI_AS", \n\t"Rua" NVARCHAR(200) COLLATE "SQL_Latin1_General_CP1_CI_AS", \n\t"Localidade" NVARCHAR(200) COLLATE "SQL_Latin1_General_CP1_CI_AS", \n\t"codPostal" NVARCHAR(10) COLLATE "SQL_Latin1_General_CP1_CI_AS", \n\tdataini DATETIME, \n\tdataact DATETIME, \n\temail NVARCHAR(50) COLLATE "SQL_Latin1_General_CP1_CI_AS", \n\turl NVARCHAR(50) COLLATE "SQL_Latin1_General_CP1_CI_AS", \n\tPRIMARY KEY ("idOperador")\n)\n\n']
My understanding from this error is that PostgreSQL doesn't have NVARCHAR but VARCHAR, which should be equivalent. I thought that SQLAlchemy would automatically map both of them to String in its layer of abstraction, but perhaps it doesn't work that way in this case.
Question: Should I define all the classes/tables beforehand, for instance, in models.py, in order to avoid errors like this? If so, how would that integrate with the given (or other) workflow?
In fact, this error was obtained running the code from Urbanek, where I can specify which tables I want to copy. Running the sample above, leads me to...
Problem 2
The MS SQL installation is a geodatabase that is using ArcSDE (Spatial Database Engine). For that reason, some of the columns are of a non-defaultGeometry type. On the PostgreSQL side, I am using PostGIS 2.
When trying to copy tables with those types, I get warnings like these:
/usr/local/lib/python2.7/dist-packages/sqlalchemy/dialects/mssql/base.py:1791: SAWarning: Did not recognize type 'geometry' of column 'geom'
(type, name))
/usr/local/lib/python2.7/dist-packages/sqlalchemy/dialects/mssql/base.py:1791: SAWarning: Did not recognize type 'geometry' of column 'shape'
Those are later followed by another error (this one was actually thrown when executing the provided code above):
sqlalchemy.exc.ProgrammingError: (psycopg2.ProgrammingError) relation "SDE_spatial_references" does not exist
LINE 1: INSERT INTO "SDE_spatial_references" (srid, description, aut...
^
I think that it failed to create the columns referred in the warnings, but the error was thrown at a later step when those columns were needed.
Question: The question is an extension of the previous one: how to do the migration with custom (or defined somewhere else) types?
I know about GeoAlchemy2 that can be used with PostGIS. GeoAlchemy supports MS SQL Server 2008, but in that case I guess I'm stuck with SQLAlchemy 0.8.4 (perhaps with less nice features). Also, I found here that it is possible to do the reflection using types defined by GeoAlchemy. However, my questions remain.
Possibly related
https://stackoverflow.com/questions/34475241/how-to-migrate-from-mysql-to-postgressql-using-pymysql
SqlAlchemy: export table to new database
https://stackoverflow.com/questions/34956523/sqlalchemy-custom-column-type-use-bindparam-as-multiple-function-parameters
SQLAlchemy Reflection Using Metaclass with Column Override
Edit
When I saw the error referring SDE_spatial_references I thought that it could be something related to ArcSDE, because the same machine also has ArcGIS for Server installed. Then I've learned that MS SQL Server also has some Spatial Data Types, and then I confirmed this is the case. I was wrong with this edit: the database is indeed using ArcSDE.
Edit 2
Here are some more details that I forgot to include.
The migration doesn't have to be done with SQLAlchemy. I'd thought that would be a good idea because:
I prefer to work with Python
The solution has to be with FOSS
Ideally, it would be in a way easily reproducible, and possible to launch and wait
After the migration I'd like to use Alembic to conduct further schema migrations
Other things that I have tried and failed (can't remember now the exact reasons, but I'd go through them again if any answer refers them):
Kettle
Geokettle
ogr2ogr (still trying this approach)
Database details:
Small database, ± 3 GB
± 40 tables
There are tables with both spatial and non-spatial data
Both databases (SQL Server and PostgreSQL) in the same server, which is running Windows Server 2008
No big problem with downtime (up to 8 hours would be fine)
Here is my solution using SQLAlchemy. This is a long-blog-like post, I hope that it is something acceptable here, and useful to someone.
Possibly, this also works with other combinations of source and target databases (besides MS SQL Server and PostgreSQL, respectively), although they were not tested.
Workflow (sort of TL;DR)
Inspect the source automatically and deduce the existing table models (this is called reflection).
Import previously defined table models which will be used to create the new tables in the target.
Iterate over the table models (the ones existing in both source and target).
For each table, fetch chunks of rows from source and insert them into target.
Requirements
SQLAlchemy
GeoAlchemy2
sqlacodegen
Detailed steps
1. Connect to the databases
SQLAlchemy calls engine to the object that handles the connection between the application and the actual database. So, to connect to the databases, an engine must be created with the corresponding connection string. The typical form of a database URL is:
dialect+driver://username:password#host:port/database
You can see some example of connection URL's in the SQLAlchemy documentation.
Once created, the engine will not establish a connection until it is explicitly told to do so, either through the .connect() method or when an operation which is dependent on this method is invoked (e.g., .execute()).
con = ms_sql.connect()
2. Define and create tables
2.1 Source database
Tables in the source side are already defined, so we can use table reflection:
from sqlalchemy import MetaData
metadata = MetaData(source_engine)
metadata.reflect(bind=source_engine)
You may see some warnings if you try this. For example,
SAWarning: Did not recognize type 'geometry' of column 'Shape'
That is because SQLAlchemy does not recognize custom types automatically. In my specific case, this was because of an ArcSDE type. However, this is not problematic when you only need to read data. Just ignore those warnings.
After the table reflection, you can access the existing tables through that metadata object.
# see all the tables names
print list(metadata.tables)
# handle the table named 'Troco'
src_table = metadata.tables['Troco']
# see that table columns
print src_table.c
2.2 Target database
For the target, because we are starting a new database, it is not possible to use tables reflection. However, it is not complicated to create the table models through SQLAlchemy; in fact, it might be even simpler than writing pure SQL.
from sqlalchemy import Column, Integer, String
from sqlalchemy.ext.declarative import declarative_base
Base = declarative_base()
class SomeClass(Base):
__tablename__ = 'some_table'
id = Column(Integer, primary_key=True)
name = Column(String(50))
Shape = Column(Geometry('MULTIPOLYGON', srid=102165))
In this example there is a column with spatial data (defined here thanks to GeoAlchemy2).
Now, if you have tenths of tables, defining so many tables may be baffling, tedious, or error prone. Luckily, there is sqlacodegen, a tool that reads the structure of an existing database and generates the corresponding SQLAlchemy model code. Example:
pip install sqlacodegen
sqlacodegen mssql:///some_local_db --outfile models.py
Because the purpose here is just to migrate the data, and not the schema, you can create the models from the source database, and just adapt/correct the generated code to the target database.
Note: It will generate mixed class models and Table models. Read here about this behavior.
Again, you will see similar warnings about unrecognized custom data types. That is one of the reasons why we now have to edit the models.py file and adjust the models. Here are some hints on things to adjust:
The columns with custom data types are defined with NullType. Replace them with the proper type, for instance, GeoAlchemy2's Geometry.
When defining Geometry's, pass the correct geometry type (linestring, multilinestring, polygon, etc.) and the SRID.
PostgreSQL character types are variable length capable, and SQLAlchemy will map String columns to them by default, so we can replace all Unicode and String(...) by String. Note that it is not required, nor advisable (don't quote me on this), to specify the number of characters in String, just omit them.
You will have to double check, but, probably, all BIT columns are in fact Boolean.
Most numeric types (e.g., Float(...), Numeric(...)), likewise for character types, can be simplified to Numeric. Be careful with exceptions and/or some specific case.
I have noticed some issues with columns defined as indexes (index=True). In my case, because the schema will be migrated, these should not be required now and could be safely removed.
Make sure the table and column names are the same in both databases (reflected tables and defined models), this is a requirement for a later step.
Now we can connect the models and the database together, and create all the tables in the target side.
Base.metadata.bind = postgres
Base.metadata.create_all()
Notice that, by default, .create_all() will not touch existing tables. In case you want to recreate or insert data into an existing table, it is required to DROP it beforehand.
Base.metadata.drop_all()
3. Get data
Now you are ready to copy data from one side and, later, paste it into the other. Basically, you just need to issue a SELECT query for each table. This is something possible and easy to do over the layer of abstraction provided by SQLAlchemy ORM.
data = ms_sql.execute(metadata.tables['TableName'].select()).fetchall()
However, this is not enough, you will need a little bit more of control. The reason for that is related to ArcSDE. Because it uses a proprietary format, you can retrieve the data but you cannot parse it correctly. You would get something like this:
(1, Decimal('0'), u' ', bytearray(b'\x01\x02\x00\x00\x00\x02\x00\x00\x00#\xb1\xbf\xec/\xf8\xf4\xc0\x80\nF%\x99(\xf9\xc0#\xe3\xa5\x9b\x94\xf6\xf4\xc0\x806\xab>\xc5%\xf9\xc0'))
The workaround here was to convert the geometric column to the Well Known Text (WKT) format. This conversion has to take place in the database side. ArcSDE is there, so it knows how to convert it. So, for example, in the TableName there is a column with spatial data called shape. The required SQL statement should look like this:
SELECT [TableName].[shape].STAsText() FROM [TableName]
This uses .STAsText(), a geometry data type method of the SQL Server.
If you are not working with ArcSDE, the following steps are not required:
iterate over the tables (only those that are defined in both the source and in the target),
for each table, look for a geometry column (list them beforehand)
build a SQL statement like the one above
Once a statement is built, SQLAlchemy can execute it.
result = ms_sql.execute(statement)
In fact, this does not actually get the data (compare with the ORM example -- notice the missing .fetchall() call). To explain, here is a quote from the SQLAlchemy docs:
The returned result is an instance of ResultProxy, which references a
DBAPI cursor and provides a largely compatible interface with that of
the DBAPI cursor. The DBAPI cursor will be closed by the ResultProxy
when all of its result rows (if any) are exhausted.
The data will only be retrieved just before it is inserted.
4. Insert data
Connections are established, tables are created, data have been prepared, now lets insert it. Similarly to getting the data, SQLAlchemy also allows to INSERT data into a given table through its ORM:
postgres_engine.execute(Base.metadata.tables['TableName'].insert(), data)
Again, this is easy, but because of non-standard formats and erroneous data, further manipulation will probably be required.
4.1 Matching columns
First, there were some issues with matching the source columns with the target columns (of the same table) -- perhaps this was related to the Geometry column. A possible solution is to create a Python dictionary, which maps the values from the source column to the key (name) of the target column.
This is performed row by row -- although, it is not so slow as one would guess, because the actual insertion will be by several rows at the same time. So, there will be one dictionary per row, and, instead of inserting the data object (which is a list of tuples; one tuple corresponds to one row), you will be inserting a list of dictionaries.
Here is an example for one single row. The fetched data is a list with one tuple, and values is the built dictionary.
# data
[(1, 6, None, None, 204, 1, True, False, 204, 1.0, 1.0, 1.0, False, None]
# values
[{'DateDeleted': None, 'sentidocirculacao': False, 'TempoPercursoMed': 1.0,
'ExtensaoTroco': 204, 'OBJECTID': 229119, 'NumViasSentido': 1,
'Deleted': False, 'TempoPercursoMin': 1.0, 'IdCentroOp': 6,
'IDParagemInicio': None, 'IDParagemFim': None, 'TipoPavimento': True,
'TempoPercursoMax': 1.0, 'IDTroco': 1, 'CorredorBusext': 204}]
Note that Python dictionaries are not ordered, that is why the numbers in both lists are not in the same position. The geometric column was removed from this example for simplification.
4.2 Fixing geometries
Probably, the previous workaround would not be required if this issue had not occurred: sometimes geometries are stored/retrieved with the wrong type.
In MSSQL/ArcSDE, the geometry data type does not specify which type of geometry it is being stored (i.e., line, polygon, etc.). It only cares that it is a geometry. This information is stored in another (system) table, called SDE_geometry_columns (see in the bottom of that page). However, Postgres (PostGIS, actually) requires the geometry type when defining a geometric column.
This leads to spatial data being stored with the wrong geometry type. By wrong I mean that it is different than what it should be. For instance, looking at SDE_geometry_columns table (excerpt):
f_table_name geometry_type
TableName 9
geometry_type = 9 corresponds to ST_MULTILINESTRING. However, there are rows in TableName table which are stored (or received) as ST_LINESTRING. This mismatch raises an error in Postgres side.
As a workaround, you can edit the WKT while creating the aforementioned dictionaries. For example, 'LINESTRING (10 12, 20 22)' is transformed to MULTILINESTRING ((10 12, 20 22))'.
4.3 Missing SRID
Finally, if you are willing to keep the SRID's, you also need to define them when creating geometric columns.
If there is a SRID defined in the table model, it has to be satisfied when inserting data in Postgres. The problem is that when fetching geometry data as WKT with the .STAsText() method, you lose the SRID information.
Luckily, PostGIS supports an Extended-WKT (E-WKT) format that includes the SRID.
The solution here is to include the SRID when fixing the geometries. With the same example, 'LINESTRING (10 12, 20 22)' is transformed to 'SRID=102165;MULTILINESTRING ((10 12, 20 22))'.
4.4 Fetch and insert
Once everything is fixed, you are ready to insert. As referred before, only now the data will be actually retrieved from the source. You can do this in chunks (a user defined amount) of data, for instance, 1000 rows at a time.
while True:
rows = data.fetchmany(1000)
if not rows:
break
values = [{key: (val if key.lower() != "shape" else fix(val, 102165))
for key, val in zip(keys, row)} for row in rows]
postgres_engine.execute(target_table.insert(), values)
Here fix() is the function that will correct the geometries and prepend the given SRID to geometric columns (which are identified, in this example, by the column name of "shape") -- like described above --, and values is the aforementioned list of dictionaries.
Result
The result is a copy of the schema and data, existing on a MS SQL Server + ArcSDE database, into a PostgreSQL + PostGIS database.
Here are some stats, from my use case, for performance analysis. Both databases are in the same machine; the code was executed from a different machine, but in the same local network.
Tables | Geometry Column | Rows | Fixed Geometries | Insert Time
---------------------------------------------------------------------------------
Table 1 MULTILINESTRING 1114797 702 17min12s
Table 2 None 460874 --- 4min55s
Table 3 MULTILINESTRING 389485 389485 4min20s
Table 4 MULTIPOLYGON 4050 3993 34s
Total 3777964 871243 48min27s
I faced the same problems trying to migrate from Oracle 9i to MySQL.
I built etlalchemy to solve this problem, and it has currently been tested migrating to and from MySQL, PostgreSQL, SQL Server, Oracle and SQLite. It leverages SQLAlchemy, and BULK CSV Import features of the aforementioned RDBMS's (and can be quite fast!).
Install (non El-capitan): pip install etlalchemy
Install (El-capitan): pip install --ignore-installed etlalchemy
Run:
from etlalchemy import ETLAlchemySource, ETLAlchemyTarget
# Migrate from SQL Server onto PostgreSQL
src = ETLAlchemySource("mssql+pyodbc://user:passwd#DSN_NAME")
tgt = ETLAlchemyTarget("postgresql://user:passwd#hostname/dbname",
drop_database=True)
tgt.addSource(src)
tgt.migrate()
I'd recommend this flow with two big steps to migrate:
Migrate schema
Dump source DB schema, preferably to some unified format across data tools like UML (this and next steps will need and be easier with toll like Toad Data Modeler or IBM Rational Rose).
Change tables definitions from source types to target types when needed with TDM or RR. E. g. get rid of varchar(n) and stick to text in postgres, unless you specifically need application to crash on data layer with strings longer than n. Omit (for now) complex types like geometry, if there is no way to convert them in data modeling tools.
Generate a DDL-file for target DB (mentioned tools are handy here, again).
Create (and add to tables) complex types as they should be handled by target RDBMS. Try to insert a couple of entries to be sure datatypes are consistent. Add these types to your DDL-file.
You may also want to disable checks like foreign key constraints here.
Migrate data
Dump simple tables (i. e. with scalar fields) to a CSV.
Import simple tables data.
Write a simple piece of code to select complex data from source and to insert this into target (it is easier than it sounds, just select -> map attributes -> insert). Do not write migration for all complex types in one code routine, keep it simple, divide and conquer.
If you have not disabled checks while you were migrating schema it is possible that you need to repeat steps 2 and 3 for different tables (that's why, well, disable checks :)).
Enable checks.
This way you will split your migration process in simple atomic steps, and failure on a step 3 of data migration will not cause you to move back to the schema migration, etc. You can just truncate a couple of tables, and rerun data import if something fail.
A bit of background first. My company is evaluating whether or not we will migrate our Informix database to Oracle 10g. We have several ESQL/C programs. I've run some through the Oracle Migration workbench and have been muddling through some testing. Now I've come to realize a few things.
First, we have dynamic sql statements that are not handling null values at all. From what I've read, I either have to manually modify the queries to utilize the nvl( ) function or implement indicator variables. Can someone confirm if manual modifications are necessary? The least amount of manual changes we have to make to our converted ESQL/C programs, the better.
Second, we have several queries which pull dates from various tables etc., and in Informix dates are treated as type long, the # of days since Dec 31st, 1899.
In Pro*C, what format is a date being selected as? I know it's not numeric because I tried selecting date field into my long variable and get Oracle error stating "expected NUMBER but got a DATE". So I'm assuming we'd have to modify how we are selecting date fields - either select a date field in a converted manner so it becomes a long (ie, # of days since 12/31/1899), or change the host variable to match what Oracle is returning (what is that, string?).
Ya. You will need to modify your queries as you described.
long is tripping you up. long has a different meaning in Oracle. There is a specific DATE type. Generally when selecting one uses the TO_DATE function with a format, to get the result as a VARCHAR2, in exactly the format you want.
Probably it didn't hit you yet but be aware that in Oracle empty VARCHAR2 fields are NULLs. I see no logic behind this (probably because I came from Informix land) - just keep it in mind. I think it is stupid - IMHO empty string is meaningful and different from NULL.
Either modify all your VARCHAR2 fields to be NOT NULL DEFAULT '-' or any other arbitrary value, or use indicatores in ALL your queries that return VARCHAR2 fields, or always use NVL().
In order to convert the oracle dates (which are store in Oracle internal format) into a long integer, you will need to alter your queries. Use the following formula for your dates:
to_number (to_char (date_column, 'J')) - to_number(to_char(to_date('12/31/1899', 'MM/DD/YYYY'), 'J'))
The Oracle system 'J' (for Julian date) format is a count of number of days since December 31, 4712BC. If you want to count from a later date, you'll need to subtract off the Julian day count of that later date.
One suggestion: instead of altering all of your queries in your programs (which may create problems and introduce bugs), create a set of views in a different schema. These views would be named the same as all the tables, with all the same columns, but include the NVL() and date() formulas (like above). Then point your application at the view schema rather than the base table schema. Much less testing and fewer places to missing something.
So for example, put all your tables into a schema called "APPS_BASE" (defined by the user "APPS_BASE". Then create another schema/user called "APPS_VIEWS". In the APPS_VIEWS create a view:
CREATE OR REPLACE VIEW EMP AS
SELECT name, birth_date
FROM APPS_BASE.EMP;