Thread: PostGres ODBC too slow
Hello There,
Recently, we've encountered some performance issues with the ODBC driver provided by Postgres. Upon investigation, we noticed that switching from the UNICODE to ANSI driver resulted in performance improvements for some queries, albeit at the expense of others.
To delve deeper into this matter, I conducted tests using a Python script with the psycopg2 library, and the results were significantly better. However, to address this issue comprehensively, I've explored alternative ODBC drivers available in the market. While some minor improvements were observed in a few queries with a different driver, the overall performance remains a concern.
Given your extensive experience in this area, I would greatly appreciate your insights and recommendations on which ODBC driver would be most suitable for our use case. Any advice or suggestions you could offer would be immensely helpful in resolving this performance issue.
To delve deeper into this matter, I conducted tests using a Python script with the psycopg2 library, and the results were significantly better. However, to address this issue comprehensively, I've explored alternative ODBC drivers available in the market. While some minor improvements were observed in a few queries with a different driver, the overall performance remains a concern.
Given your extensive experience in this area, I would greatly appreciate your insights and recommendations on which ODBC driver would be most suitable for our use case. Any advice or suggestions you could offer would be immensely helpful in resolving this performance issue.
Let me know in case of any questions or concerns.
Thanks,
Can you explain where/why you think it is too slow ?
Dave Cramer
www.postgres.rocks
On Mon, 18 Mar 2024 at 10:56, vedant patel <vedantpatel297@gmail.com> wrote:
Hello There,Recently, we've encountered some performance issues with the ODBC driver provided by Postgres. Upon investigation, we noticed that switching from the UNICODE to ANSI driver resulted in performance improvements for some queries, albeit at the expense of others.
To delve deeper into this matter, I conducted tests using a Python script with the psycopg2 library, and the results were significantly better. However, to address this issue comprehensively, I've explored alternative ODBC drivers available in the market. While some minor improvements were observed in a few queries with a different driver, the overall performance remains a concern.
Given your extensive experience in this area, I would greatly appreciate your insights and recommendations on which ODBC driver would be most suitable for our use case. Any advice or suggestions you could offer would be immensely helpful in resolving this performance issue.Let me know in case of any questions or concerns.Thanks,
Hello Dave,
We have tried the retrieval using python library psycopg2 and found that it is taking less amount of time for retrieval. And when we create odbc and use pyodbc it is taking a large amount of time for that odbc. Along with that our main application is in PowerBuilder and we need to use ODBC for that one. We checked the ODBC driver with PowerBuilder and it is taking too much time.
Thanks,
On Tue, Mar 19, 2024 at 12:19 AM Dave Cramer <davecramer@postgres.rocks> wrote:
Can you explain where/why you think it is too slow ?Dave Cramerwww.postgres.rocksOn Mon, 18 Mar 2024 at 10:56, vedant patel <vedantpatel297@gmail.com> wrote:Hello There,Recently, we've encountered some performance issues with the ODBC driver provided by Postgres. Upon investigation, we noticed that switching from the UNICODE to ANSI driver resulted in performance improvements for some queries, albeit at the expense of others.
To delve deeper into this matter, I conducted tests using a Python script with the psycopg2 library, and the results were significantly better. However, to address this issue comprehensively, I've explored alternative ODBC drivers available in the market. While some minor improvements were observed in a few queries with a different driver, the overall performance remains a concern.
Given your extensive experience in this area, I would greatly appreciate your insights and recommendations on which ODBC driver would be most suitable for our use case. Any advice or suggestions you could offer would be immensely helpful in resolving this performance issue.Let me know in case of any questions or concerns.Thanks,
HI Vedant,
Can you share the SQL that is slow. We can't fix this if we don't have a specific problem to look at.
Dave Cramer
www.postgres.rocks
On Wed, 20 Mar 2024 at 02:08, vedant patel <vedantpatel297@gmail.com> wrote:
Hello Dave,We have tried the retrieval using python library psycopg2 and found that it is taking less amount of time for retrieval. And when we create odbc and use pyodbc it is taking a large amount of time for that odbc. Along with that our main application is in PowerBuilder and we need to use ODBC for that one. We checked the ODBC driver with PowerBuilder and it is taking too much time.Thanks,On Tue, Mar 19, 2024 at 12:19 AM Dave Cramer <davecramer@postgres.rocks> wrote:Can you explain where/why you think it is too slow ?Dave Cramerwww.postgres.rocksOn Mon, 18 Mar 2024 at 10:56, vedant patel <vedantpatel297@gmail.com> wrote:Hello There,Recently, we've encountered some performance issues with the ODBC driver provided by Postgres. Upon investigation, we noticed that switching from the UNICODE to ANSI driver resulted in performance improvements for some queries, albeit at the expense of others.
To delve deeper into this matter, I conducted tests using a Python script with the psycopg2 library, and the results were significantly better. However, to address this issue comprehensively, I've explored alternative ODBC drivers available in the market. While some minor improvements were observed in a few queries with a different driver, the overall performance remains a concern.
Given your extensive experience in this area, I would greatly appreciate your insights and recommendations on which ODBC driver would be most suitable for our use case. Any advice or suggestions you could offer would be immensely helpful in resolving this performance issue.Let me know in case of any questions or concerns.Thanks,
Hi
Environment is also important. Are you using connecting through LAN or WAN?
WAN communications are impacted by the size of the data transmitted (and also for not using the network to receive data while processing previously obtained data). ODBC driver as far as I know does use text data serialization which may be bigger than other pg clients. Not having compression does also affect WAN communications but that is postgres related so other clients would be in the same boat.
Best regards
Jacobo
El 20/03/2024 a las 12:17, Dave Cramer escribió:
HI Vedant,Can you share the SQL that is slow. We can't fix this if we don't have a specific problem to look at.On Wed, 20 Mar 2024 at 02:08, vedant patel <vedantpatel297@gmail.com> wrote:Hello Dave,We have tried the retrieval using python library psycopg2 and found that it is taking less amount of time for retrieval. And when we create odbc and use pyodbc it is taking a large amount of time for that odbc. Along with that our main application is in PowerBuilder and we need to use ODBC for that one. We checked the ODBC driver with PowerBuilder and it is taking too much time.Thanks,On Tue, Mar 19, 2024 at 12:19 AM Dave Cramer <davecramer@postgres.rocks> wrote:Can you explain where/why you think it is too slow ?On Mon, 18 Mar 2024 at 10:56, vedant patel <vedantpatel297@gmail.com> wrote:Hello There,Recently, we've encountered some performance issues with the ODBC driver provided by Postgres. Upon investigation, we noticed that switching from the UNICODE to ANSI driver resulted in performance improvements for some queries, albeit at the expense of others.
To delve deeper into this matter, I conducted tests using a Python script with the psycopg2 library, and the results were significantly better. However, to address this issue comprehensively, I've explored alternative ODBC drivers available in the market. While some minor improvements were observed in a few queries with a different driver, the overall performance remains a concern.
Given your extensive experience in this area, I would greatly appreciate your insights and recommendations on which ODBC driver would be most suitable for our use case. Any advice or suggestions you could offer would be immensely helpful in resolving this performance issue.Let me know in case of any questions or concerns.Thanks,