What's New in PostgreSQL 17? - Mydbops MyWebinar Edition 35

What’s New in
PostgreSQL 17
Presented by
Mukul Mahto
Mydbops
Mydbops MyWebinar - 35
Aug 23rd, 2024

Consulting
Services
Consulting
Services
Managed
Services
❏ Database Management and
consultancy provider
❏ Founded in 2016
❏ Assisted 800+ happy customers
❏ AWS partners
❏ PCI & ISO certified
About Us

❏ Introduction to PostgreSQL 17
❏ Performance Optimization
❏ Partitioning & Distributed Workloads
❏ Developer Workflow Enhancement
❏ Security Features
❏ Backup and Export Management
❏ Monitoring
Agenda

PostgreSQL 17
Key Enhancements
- Performance Boosts
- High Availability
- Developer Features
- Security
Why Upgrade?
- Boosted Efficiency
- High Availability
- Cutting-Edge Features

❏ Improved Vacuum Mechanism
New Data Structure: Efficient radix tree replaces old array.
Dynamic Memory Allocation: No 1GB limit.
Faster TID Lookup: Speedier Tuple ID searches.
20x Memory Reduction with New Data Structure
❏ Reduced WAL Volume for Vacuum
Single record for freezing and pruning.
Faster Sync: Reduced sync and write time.
Mark Items as LP_UNUSED Instead of LP_DEAD: Reduces WAL volume

Comparison Performance of Postgres 16 vs 17
maintenance_work_mem is set to 64 megabytes on both of these.
Postgres 16:-
Postgres 17:-
CREATE TABLE test_16 (id INT);
INSERT INTO test_16 (id) SELECT generate_series(1, 10000000);
CREATE INDEX ON test_16(id);
UPDATE test_16 SET id = id - 1;
CREATE TABLE test_17 (id INT);
INSERT INTO test_17 (id) SELECT generate_series(1, 10000000);
CREATE INDEX ON test_17(id);
UPDATE test_17 SET id = id - 1;

Postgres 16 Postgres 17
2024-08-19 10:49:17.351 UTC [21830] LOG: automatic vacuum of table
"pg16.public.test_16": index scans: 1
pages: 0 removed, 88496 remain, 88496 scanned (100.00% of total)
tuples: 3180676 removed, 10000000 remain, 0 are dead but not
yet removable
removable cutoff: 747, which was 0 XIDs old when operation ended
new relfrozenxid: 745, which is 2 XIDs ahead of previous value
frozen: 0 pages from table (0.00% of total) had 0 tuples frozen
index scan needed: 44248 pages from table (50.00% of total) had
10000000 dead item identifiers removed
index "test_16_id_idx": pages: 54840 in total, 0 newly deleted, 0
currently deleted, 0 reusable
avg read rate: 24.183 MB/s, avg write rate: 22.317 MB/s
buffer usage: 119566 hits, 156585 misses, 144501 dirtied
WAL usage: 201465 records, 60976 full page images, 233670948
bytes
system usage: CPU: user: 9.04 s, system: 3.56 s, elapsed: 50.58 s
2024-08-19 10:51:54.081 UTC [21959] LOG: automatic vacuum of table
"pg17.public.test_17": index scans: 0
pages: 0 removed, 76831 remain, 76831 scanned (100.00% of total)
tuples: 0 removed, 10000000 remain, 0 are dead but not yet
removable
removable cutoff: 827, which was 1 XIDs old when operation ended
new relfrozenxid: 825, which is 1 XIDs ahead of previous value
frozen: 0 pages from table (0.00% of total) had 0 tuples frozen
index scan not needed: 0 pages from table (0.00% of total) had 0
dead item identifiers removed
avg read rate: 34.779 MB/s, avg write rate: 19.580 MB/s
buffer usage: 79090 hits, 74631 misses, 42016 dirtied
WAL usage: 4969 records, 2 full page images, 309554 bytes
system usage: CPU: user: 2.71 s, system: 1.14 s, elapsed: 16.76 s

Streaming I/O in Pg 17
❏ Streaming I/O with I/O combining
- Reads multiple pages in one go
- Increases read size to 128K(default)
- New GUC: io_combine_limit
❏ Streaming I/O Enhancements for Sequential Scans
- Faster sequential scans with 128K reads
- Reduces system calls for improved speed
- Improves SELECT query performance
❏ Streaming I/O Benefits for ANALYZE
- Efficient data reading via prefetching
- Improved performance for table statistics
- Supports faster query planning
1
2
3
1
2
3
8k
8k
8k
24k
Postgres 16
Postgres 17
OS
OS

Postgres 16
pg16=# EXPLAIN (ANALYZE, BUFFERS) SELECT COUNT(*) FROM test_16;
QUERY PLAN
------------------------------------------------------------------------------------------------------------------------------
Aggregate (cost=212496.60..212496.61 rows=1 width=8) (actual time=2869.400..2869.401 rows=1 loops=1)
Buffers: shared hit=416 read=88080
-> Seq Scan on test_16 (cost=0.00..187696.48 rows=9920048 width=0) (actual time=145.205..2082.923 rows=10000000 loops=1)
Planning Time: 0.059 ms
JIT:
Functions: 2
Options: Inlining false, Optimization false, Expressions true, Deforming true
Timing: Generation 0.316 ms, Inlining 0.000 ms, Optimization 0.120 ms, Emission 1.452 ms, Total 1.888 ms
Execution Time: 2869.794 ms
(10 rows)

Postgres 17
pg17=# EXPLAIN (ANALYZE, BUFFERS) SELECT COUNT(*) FROM test_17;
QUERY PLAN
------------------------------------------------------------------------------------------------------------------------------------------------
Finalize Aggregate (cost=142142.30..142142.31 rows=1 width=8) (actual time=1287.047..1287.196 rows=1 loops=1)
-> Gather (cost=142142.09..142142.30 rows=2 width=8) (actual time=1281.208..1287.189 rows=3 loops=1)
Workers Planned: 2
Workers Launched: 2
-> Partial Aggregate (cost=141142.09..141142.10 rows=1 width=8) (actual time=1268.267..1268.268 rows=1 loops=3)
-> Parallel Seq Scan on test_17 (cost=0.00..130612.87 rows=4211687 width=0) (actual time=51.831..836.598 rows=3333333
loops=3)
Planning:
Buffers: shared hit=61
(14 rows)

● Configurable Cache Sizes: Adjustable SLRU cache.
● Banked Cache: Efficient data eviction by bank.
● Separate Locks: Each bank uses its own lock.
Improved SLRU Performance

Query Optimization
❏ Faster B-Tree Index Scan
- Improved IN clause performance.
❏ 3x Performance boost

Postgres 16
Postgres 17
pg16=# EXPLAIN (ANALYZE, BUFFERS) SELECT * FROM test_16 WHERE id IN (1, 2, 3, 4, 45, 6, 10007);
QUERY PLAN
------------------------------------------------------------------------------------------------------------------------------
Index Only Scan using test_16_id_idx on test_16 (cost=0.43..31.17 rows=7 width=4) (actual time=0.052..1.277 rows=7 loops=1)
Index Cond: (id = ANY ('{1,2,3,4,45,6,10007}'::integer[]))
Heap Fetches: 0
Planning:
(8 rows)
pg17=# EXPLAIN (ANALYZE, BUFFERS) SELECT * FROM test_17 WHERE id IN (1, 2, 3, 4, 45, 6, 10007);
QUERY PLAN
------------------------------------------------------------------------------------------------------------------------------
Index Only Scan using test_17_id_idx on test_17 (cost=0.43..31.17 rows=7 width=4) (actual time=0.056..0.542 rows=7 loops=1)
Index Cond: (id = ANY ('{1,2,3,4,45,6,10007}'::integer[]))
Heap Fetches: 0
Planning:
(8 rows)

Performance Boost with Constraints
❏ NULL Constraint Optimization
- Skips unnecessary checks for faster queries.
❏ Impact of NOT NULL on Execution
- Reduces processing for more efficient scans.

Execution Plans without NOT NULL Constraint
pg17=# explain (analyze, buffers, costs off) SELECT count(*) FROM employees WHERE name IS NULL;
QUERY PLAN
-----------------------------------------------------------------------
Aggregate (actual time=0.022..0.023 rows=1 loops=1)
-> Seq Scan on employees (actual time=0.016..0.017 rows=2 loops=1)
Filter: (name IS NULL)
Rows Removed by Filter: 8
pg17=# explain (analyze, buffers, costs off) SELECT count(*) FROM employees WHERE name IS NOT NULL;
QUERY PLAN
-----------------------------------------------------------------------
Filter: (name IS NOT NULL)
Rows Removed by Filter: 2
(8 rows)

Execution Plans with NOT NULL Constraint
pg17=# ALTER TABLE employees ALTER COLUMN name SET NOT NULL;
ALTER TABLE
pg17=#
pg17=# explain (analyze, buffers, costs off) SELECT count(*) FROM employees WHERE name IS NULL;
QUERY PLAN
--------------------------------------------------------
-> Result (actual time=0.002..0.002 rows=0 loops=1)
One-Time Filter: false
Planning:
(7 rows)
pg17=# explain (analyze, buffers, costs off) SELECT count(*) FROM employees WHERE name IS NOT NULL;
QUERY PLAN
-----------------------------------------------------------------------
(6 rows)

Faster BRIN Index Creation
❏ Multi-Worker Support
❏ Parallel Processing
❏ Sorted Summary Streams
❏ Leader Integration
❏ Efficient Resource Use

Partitioned and distributed
workloads enhancements

Partition Management
SPLIT PARTITION
- Break large partitions into smaller ones.
MERGE PARTITIONS
- Merge two or more partitions into one.
Note:- Access Exclusive Lock

Exclusion Constraints on Partitioned Tables
Exclusion constraints prevent conﬂicting data in diﬀerent partitions of a table.
pg17=# CREATE TABLE orders (
order_id serial,
customer_id int,
order_date date,
total_amount numeric,
UNIQUE (customer_id)
) PARTITION BY RANGE (order_date);
ERROR: unique constraint on partitioned table must include all partitioning columns
DETAIL: UNIQUE constraint on table "orders" lacks column "order_date" which is part of the partition key.
pg17=# CREATE TABLE orders (
order_id serial,
customer_id int,
order_date date,
total_amount numeric,
UNIQUE (order_date)
) PARTITION BY RANGE (order_date);
CREATE TABLE
pg17=#

Logical Replication
Upgrade Support: No slot drop needed.
Failover Control: Enhanced high availability.
Failover Slot Synchronization: SlotSync Workers
Synchronization:
❏ Automatic Sync: using ‘sync_replication_slots = on’ to sync automatically
❏ Manual Sync: Use pg_sync_replication_slots() to sync slots manually.
Publisher Subscriber
CREATE SUBSCRIPTION sub CONNECTION '$connstr'
PUBLICATION pub WITH (failover = 'true’)

Upgrade Logical Replication Nodes
❏ Seamless Upgrades
- Migrate replication slots during an upgrade without losing data.
❏ Preserve State
- Keeps the replication state unchanged during an upgrade.
❏ Upgrading Subscribers
- Allow Write operations on the primary continue during subscriber upgrades.
❏ Version Requirement
- All cluster members must run PostgreSQL 17.0+ for migration.

New Tool: pg_createsubscriber command.
❏ Helps create logical replicas from physical standbys.
❏ The pg_createsubscriber command speeds up the setup of logical subscribers.
❏ Simplifies the process of upgrading PostgreSQL instances.
New Columns in ‘pg_replication_slots’
❏ inactive_since :- Tracks when a replication slot became inactive.
❏ Invalidation_reason :- Indicates the reason for invalidation of replication slots.

Developer Workflow Enhancement

MERGE Command
Handling unmatched rows with the WHEN NOT MATCHED BY SOURCE clause.
Returning detailed actions with the RETURNING clause and merge_action().
- merge_action Function: report the DML that generated the row.
Views must be consistent for MERGE
❏ Trigger-updatable views need INSTEAD OF triggers for actions (INSERT, UPDATE, DELETE).
❏ Auto-updatable views should not have any triggers.
MERGE

Table product_catalog Table price_updates
pg17=# CREATE TABLE product_catalog (
pg17(# product_id BIGINT,
pg17(# product_name VARCHAR(100),
pg17(# price DECIMAL(10, 2),
pg17(# last_updated TIMESTAMP,
pg17(# status TEXT
pg17(# );
CREATE TABLE
pg17=#
pg17=# INSERT INTO product_catalog (product_id, product_name,
price, last_updated, status)
pg17-# VALUES
pg17-# (1, 'Laptop', 1000.00, '2024-08-01 12:00:00', 'active'),
pg17-# (2, 'Smartphone', 500.00, '2024-08-01 12:00:00', 'active'),
pg17-# (3, 'Tablet', 300.00, '2024-08-01 12:00:00', 'active'),
pg17-# (4, 'Smartwatch', 200.00, '2024-08-01 12:00:00', 'active');
INSERT 0 4
pg17=# CREATE TABLE price_updates (
pg17(# product_id BIGINT,
pg17(# new_price DECIMAL(10, 2),
pg17(# update_type VARCHAR(10)
pg17(# );
CREATE TABLE
pg17=#
pg17=# INSERT INTO price_updates (product_id, new_price,
update_type)
pg17-# VALUES
pg17-# (1, 950.00, 'decrease'),
pg17-# (2, 550.00, 'increase'),
pg17-# (5, 250.00, 'new'),
pg17-# (3, NULL, 'discontinue');
INSERT 0 4
pg17=#

pg17=# WITH merge AS (
pg17(# MERGE INTO product_catalog trg
pg17(# USING price_updates src
pg17(# ON src.product_id = trg.product_id
pg17(# WHEN MATCHED AND src.update_type = 'increase'
pg17(# THEN UPDATE
pg17(# SET price = src.new_price,
pg17(# last_updated = CURRENT_TIMESTAMP,
pg17(# status = 'price increased'
pg17(# WHEN MATCHED AND src.update_type = 'decrease'
pg17(# THEN UPDATE
pg17(# SET price = src.new_price,
pg17(# last_updated = CURRENT_TIMESTAMP,
pg17(# status = 'price decreased'
pg17(# WHEN MATCHED AND src.update_type = 'discontinue'
pg17(# THEN UPDATE
pg17(# SET status = 'discontinued',
pg17(# last_updated = CURRENT_TIMESTAMP
pg17(# WHEN NOT MATCHED BY TARGET AND src.update_type = 'new'
pg17(# THEN INSERT (product_id, product_name, price, last_updated, status)
pg17(# VALUES (src.product_id, 'Unknown Product', src.new_price, CURRENT_TIMESTAMP, 'new product')
pg17(# RETURNING merge_action(), trg.*
pg17(# )
pg17-# SELECT * FROM merge;
merge_action | product_id | product_name | price | last_updated | status
--------------------+-----------------+--------------------------+----------------------------------------------------+-----------------------
UPDATE | 1 | Laptop | 950.00 | 2024-08-19 21:47:23.507675 | price decreased
UPDATE | 2 | Smartphone | 550.00 | 2024-08-19 21:47:23.507675 | price increased
INSERT | 5 | Unknown Product | 250.00 | 2024-08-19 21:47:23.507675 | new product
UPDATE | 3 | Tablet | 300.00 | 2024-08-19 21:47:23.507675 | discontinued
(4 rows)

COPY Improvements
❏ Managing data type incompatibilities using ON_ERROR options
- ignore(text, csv), stop
❏ Enhanced logging using LOG_VERBOSITY
- Handle mismatched data types.
- New attribute added ‘tuples_skipped’ in pg_stat_progress_copy view

root@ip-172-31-3-170:/var/log/csv_test# cat /var/log/csv_test/data_issues.csv
id,name,age,email,is_active,signup_date,last_login
1,Rahul Sharma,30,rahul.sharma@example.com,true,2024-01-01,2024-01-01 12:00:00
2,Neha Verma,-1,neha.verma@example.com,false,2024-01-02,2024-01-02 12:00:00
3,,abc,amit.patel@example.com,true,2024-01-03,2024-01-03 12:00:00
4,Amit Patel,40,amit.patel@example.com,maybe,2024-01-03,2024-01-03 12:00:00
5,Pooja Singh,25,pooja.singh@example.com,true,2024-01-04,2024-01-04 12:00:00
root@ip-172-31-3-170:/var/log/csv_test#
pg17=# SELECT * FROM example3;
id | name | age | email | is_active | signup_date | last_login
----+------+-----+-------+-----------+-------------+------------
(0 rows)
pg17=#
pg17=# copy example3 FROM '/var/log/csv_test/data_issues.csv' (ON_ERROR ignore, LOG_VERBOSITY verbose, FORMAT
csv, HEADER);
NOTICE: skipping row due to data type incompatibility at line 4 for column age: "abc"
NOTICE: skipping row due to data type incompatibility at line 5 for column is_active: "maybe"
NOTICE: 2 rows were skipped due to data type incompatibility
COPY 3
pg17=# SELECT * FROM example3;
id | name | age | email | is_active | signup_date | last_login
----+----------------------+-----+-----------------------------------------+-----------+---------------------+---------------------
1 | Rahul Sharma | 30 | rahul.sharma@example.com | t | 2024-01-01 | 2024-01-01 12:00:00
2 | Neha Verma | 1 | neha.verma@example.com | f | 2024-01-02 | 2024-01-02 12:00:00
5 | Pooja Singh | 25 | pooja.singh@example.com | t | 2024-01-04 | 2024-01-04 12:00:00
(3 rows)

JSON_TABLE: Convert JSON to table.
Converting JSON data into a table format
pg17=# CREATE TABLE customer_orders (order_data jsonb);
CREATE TABLE
pg17=# INSERT INTO customer_orders VALUES ( '{ "orders": [ { "product": "Laptop", "quantity": 2 }, { "product": "Mouse",
"quantity": 5 } ] }' );
INSERT 0 1
pg17=# SELECT jt.* FROM customer_orders, JSON_TABLE(order_data, '$.orders[*]'
pg17(# COLUMNS ( id FOR ORDINALITY, product text PATH '$.product', quantity int PATH '$.quantity' )) AS jt;
id | product | quantity
----+---------+----------
1 | Laptop | 2
2 | Mouse | 5
(2 rows)
pg17=#
SQL/JSON Enhancements

New Functions: JSON constructor
- JSON(): Converts a string into a JSON value.
- JSON_SERIALIZE(): Converts a JSON value into a binary format.
- JSON_SCALAR(): Converts a simple value into JSON format.
SELECT JSON('{"setting": "high", "volume": 10, "mute": false}');
{"setting": "high", "volume": 10, "mute": false}
SELECT JSON_SCALAR(99.99);
99.99
SELECT JSON_SERIALIZE('{"user": "Alice", "age": 30}' RETURNING bytea);
x7b2275736572223a2022416c696365222c2022616765223a2033307d

JSON Query Functions
- JSON_EXISTS(): Checks if a speciﬁc path exists in JSON
- JSON_QUERY(): Extracts a part of the JSON data.
- JSON_VALUE(): Extracts a single value from JSON.
SELECT JSON_EXISTS(jsonb '{"user": "Alice", "roles": ["admin", "user"]}', '$.roles[?(@ == "admin")]');
t(true)
SELECT JSON_QUERY(jsonb '{"user": "Bob", "roles": ["user", "editor"]}', '$.roles');
["user", "editor"]
SELECT JSON_VALUE(jsonb '{"user": "Charlie", "age": 25}', '$.user');
“Charlie”

Reduced Need for Superuser Privileges
pg_maintain Role (VACUUM, ANALYZE, CLUSTER, REFRESH MATERIALIZED VIEW, REINDEX, and LOCK TABLE on all relations.)
- Allows users to perform maintenance tasks like VACUUM and REINDEX without superuser
access.
Safe Maintenance Operations
- Grants specific users maintenance permissions on individual tables.
GRANT pg_maintain TO specific_user;
GRANT MAINTAIN ON TABLE specific_table TO specific_user;

pg_maintain Role
pg17=# create role mydbops;
CREATE ROLE
pg17=# set role mydbops;
SET
pg17=>
pg17=> dt
List of relations
Schema | Name | Type | Owner
--------+----------------------------+-------+----------
public | customer_orders | table | postgres
public | test_17 | table | postgres
(2 rows)
pg17=> vacuum test_17;
WARNING: permission denied to vacuum "test_17", skipping it
VACUUM
pg17=# GRANT pg_maintain TO mydbops;
GRANT ROLE
SET
pg17=> vacuum test_17;
VACUUM
pg17=>

MAINTAIN Privilege
pg17=# GRANT MAINTAIN ON TABLE role_test TO mydbops;
GRANT
SET
pg17=>
pg17=> dt
List of relations
--------+----------------------------+--------+----------
public | role_test | table | postgres
(3 rows)
pg17=> vacuum customer_orders;
WARNING: permission denied to vacuum "customer_orders", skipping it
VACUUM
pg17=>
pg17=> vacuum role_test;
VACUUM
pg17=>

Improved ALTER SYSTEM Command:
ALTER SYSTEM Enhancements
❏ Control Over ALTER SYSTEM
- allow_alter_system

Faster TLS Connections
● Direct TLS Setup:
- sslnegotiation=direct option,
● ALPN Requirement
● Only works on PostgreSQL 17 and newer versions.

Incremental pg_basebackup
❏ Efficient for Large Databases
❏ Using pg_basebackup --incremental
❏ Tracking Changes with WAL - summarize_wal
❏ Restoration with pg_combinebackup

l Enable summarize_wal
Taken Full backup
Taken --incremental backup using pg_basebackup
Scenario
pg17=# show summarize_wal;
summarize_wal
---------------
on
postgres@ip-172-31-3-170:/var/log$ pg_basebackup -D /var/log/full_bck17 -p 5433 -U postgres
postgres@ip-172-31-3-170:/var/log$ pg_basebackup -D incr_bck17 -p 5433 -U postgres
--incremental=/var/log/full_bck17/backup_manifest -v

Taken full backup including incremental backup using pg_combinebackup
postgres@ip-172-31-3-170:/var/log$ /usr/lib/postgresql/17/bin/pg_combinebackup --output=combine_incr_full_bck
full_bck17 incr_bck17
postgres@ip-172-31-3-170:/var/log/combine_incr_full_bck$ tail -10 backup_manifest
{ "Path": "global/6303", "Size": 16384, "Last-Modified": "2024-08-18 13:26:43 GMT", "Checksum-Algorithm": "CRC32C", "Checksum":
"e8d85722" },
"21597da0" },
"b5a46221" },
"865872e0" },
"00000000" }
],
"WAL-Ranges": [
{ "Timeline": 1, "Start-LSN": "2/B000028", "End-LSN": "2/B000120" }
],
"Manifest-Checksum": "0f16e87ab7bd400434c2725cf468d77bbcb702a49d38d749cd0e5af47f8f5fac"}
postgres@ip-172-31-3-170:/var/log/combine_incr_full_bck$

Now started the PostgreSQL server using the newly restored data directory.
postgres@ip-172-31-3-170:/home/mydbops$ /usr/lib/postgresql/17/bin/pg_ctl -D /var/lib/postgresql/17 -l logfile stop
waiting for server to shut down.... done
server stopped
postgres@ip-172-31-3-170:/var/log/combine_incr_full_bck$ /usr/lib/postgresql/17/bin/pg_ctl -D
/var/log/combine_incr_full_bck -l logfile start
waiting for server to start.... done
server started
postgres@ip-172-31-3-170:/var/log/combine_incr_full_bck$ psql -hlocalhost -p 5433 -d postgres
psql (17beta3)
Type "help" for help.
postgres=# c pg17
You are now connected to database "pg17" as user "postgres".
pg17=# dt
List of relations
-------------+--------------------------+-------+----------
public | incrment_test | table | postgres
public | role_test | table | postgres
(4 rows)

EXPLAIN Command Enhancements
❏ SERIALIZE: Network data conversion time.
❏ MEMORY: Optimizer memory usage.
pg17=# EXPLAIN (ANALYZE, SERIALIZE, MEMORY, BUFFERS)
pg17-# SELECT * FROM test_table_17 WHERE data LIKE '%data%';
QUERY PLAN
-----------------------------------------------------------------------------------------------------------------
Seq Scan on test_table_17 (cost=0.00..189.00 rows=9999 width=20) (actual time=0.031..2.401 rows=10000 loops=1)
Filter: (data ~~ '%data%'::text)
Planning:
Memory: used=9kB allocated=16kB
Serialization: time=2.028 ms output=291kB format=text
(9 rows)

pg_wait_events View
- Shows wait event details and description.
pg17=# SELECT * FROM pg_wait_events;
-[ RECORD 1
]-----------------------------------------------------------------------------------------------------------------------------
type | Activity
name | AutovacuumMain
description | Waiting in main loop of autovacuum launcher process

Pg_stat_checkpointer
❏ Displays statistics related to the checkpointer process.
❏ Replaced some of the old stats found in pg_stat_bgwriter.
pg17=# SELECT * FROM pg_stat_checkpointer;
-[ RECORD 1 ]-------+------------------------------
num_timed | 345
num_requested | 21
restartpoints_timed | 0
restartpoints_req | 0
restartpoints_done | 0
write_time | 1192795
sync_time | 1868
buffers_written | 56909
stats_reset | 2024-08-14 10:29:39.606191+00
pg17=#

Pg_stat_statements
CALL Normalization: Fewer entries for procedures.
pg16=# SELECT userid, dbid, toplevel, queryid, query,
total_exec_time, min_exec_time, max_exec_time,
mean_exec_time, stddev_exec_time FROM
pg_stat_statements WHERE query LIKE 'CALL
update_salary%';
-[ RECORD 1 ]-----+-------------------------------
userid | 10
dbid | 16384
toplevel | t
queryid | -6663051425227154753
query | CALL update_salary('HR', 1200)
total_exec_time | 0.10497
min_exec_time | 0.10497
max_exec_time | 0.10497
mean_exec_time | 0.10497
stddev_exec_time | 0
-[ RECORD 2 ]----+-------------------------------
pg17=# SELECT userid, dbid, toplevel, queryid, query,
total_exec_time, min_exec_time, max_exec_time,
mean_exec_time, stddev_exec_time, minmax_stats_since,
stats_since FROM pg_stat_statements WHERE query LIKE
'CALL update_salary%';
-[ RECORD 1 ]------+------------------------------
userid | 10
dbid | 16389
toplevel | t
queryid | 3335012495428089609
query | CALL update_salary($1, $2)
total_exec_time | 1.957936
min_exec_time | 0.06843
max_exec_time | 1.694025
mean_exec_time | 0.489484
stddev_exec_time | 0.6955936256612045
minmax_stats_since | 2024-08-20 14:19:29.726603+00
stats_since | 2024-08-20 14:19:29.726603+00
pg17=#

Removed old_snapshot_threshold Parameter
● Behavior on Querying Old Data
- "Snapshot too old" error
● Reason for Removal

References:-
https://www.postgresql.org/about/news/postgresql-17-beta-1-released-2865/?ref=dailydev
https://www.percona.com/blog/the-powerful-features-released-in-postgresql-17-beta-2/
https://databaserookies.wordpress.com/2024/07/17/postgres-17-copy/
https://pganalyze.com/blog
https://youtu.be/aWScCI5Ko18?si=KyAknBoxmagR18hh
https://andyatkinson.com/blog/2024/04/16/postgresql-17-merge-split-partitions
https://medium.com/@hnasr/combining-i-os-in-postgresql-17-39fe81d2981a
https://databaserookies.wordpress.com/2024/08/09/enhanced-merge-command/
https://www.mydbops.com/blog/postgresql-17-incremental-backup-pg-basebackup-pg-combinebackup/
https://techcommunity.microsoft.com/t5/azure-database-for-postgresql/what-s-new-with-postgres-at-microsoft-2024-edition/ba-p/4140085
https://pganalyze.com/blog/5mins-postgres-17-faster-vacuum-adaptive-radix-trees#the-building-blocks-adaptive-radix-trees-and-tidstore
https://databaserookies.wordpress.com/2024/06/11/exploring-postgresql-17-a-developers-guide-to-new-features-part-2-nulls-constraint-and-perfor
mance/
https://gist.github.com/benoittgt/ab72dc4cfedea2a0c6a5ee809d16e04d?permalink_comment_id=4972955#file-issue_with_multiple_any-sql-L3

Consulting
Services
Consulting
Services
Connect with us !
Reach us at : info@mydbops.com

What's New in PostgreSQL 17? - Mydbops MyWebinar Edition 35

More Related Content

Similar to What's New in PostgreSQL 17? - Mydbops MyWebinar Edition 35 (20)

More from Mydbops (20)

Recently uploaded (20)

What's New in PostgreSQL 17? - Mydbops MyWebinar Edition 35