jsonb_set performance degradation / multiple jsonb_set on multiple documents

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

jsonb_set performance degradation / multiple jsonb_set on multiple documents

Alexandru Lazarev
Hi PostgreSQL Community.

I tried to rewrite some plv8 stored procedures, which process in bulk JSONB documents, to PL/pgSQL.
A SP usually has to delete/update/add multiple key with the same document and do it for multiple documents (~40K) in loop.

When updating a single key PL/pgSQL wins against plv8, but when I need to update multiple keys with jsonb_set, timing increase linearly with number of jsonb_sets and takes longer than similar SP in PLV8.
Below are test-cases I've used.

QUESTION: Is it expected behavior or I do something wrong or there are some better approaches or we can treat datum as object?

test case:
PG 9.6, CentOS 7

CREATE TABLE public.configurationj2b
(
  id integer NOT NULL PRIMARY KEY,
  config jsonb NOT NULL
);
Each jsonb column has 3 top keys, and one of top-key ('data') has another 700-900 key-value pairs e.g. {"OID1":"Value1"}

PL/pgSQL SP
CREATE OR REPLACE FUNCTION public.process_jsonb()
  RETURNS void AS
$BODY$
DECLARE
    r integer;
    cfg jsonb;
BEGIN
RAISE NOTICE 'start';
    FOR r IN
        SELECT id as device_id FROM devices
    LOOP
        select config into cfg from configurationj2b c where c.id = r; --select jsonb one by one

    -- MULTIPLE KEYs, Conditional Busiines Logic (BL) updates
    cfg := jsonb_set(cfg, '{data,OID1}', '"pl/pgsql1"');
    IF cfg@>'{"data" : { "OID1":"pl/pgsql1"} }' THEN
        cfg := jsonb_set(cfg, '{data,OID2}', '"pl/pgsql2"');
    END IF;

    IF cfg@>'{"data" : { "OID2":"pl/pgsql2"} }' THEN
        cfg := jsonb_set(cfg, '{data,OID3}', '"pl/pgsql3"');
    END IF;

    IF cfg@>'{"data" : { "OID3":"pl/pgsql3"} }' THEN
        cfg := jsonb_set(cfg, '{data,OID4}', '"pl/pgsql4"');
    END IF;

    IF cfg@>'{"data" : { "OID4":"pl/pgsql4"} }' THEN
        cfg := jsonb_set(cfg, '{data,OID5}', '"pl/pgsql5"');
    END IF;

   
    update     configurationj2b c set config = cfg where c.id = r;

    END LOOP;
    RAISE NOTICE 'end';
    RETURN;
END
$BODY$
  LANGUAGE plpgsql VOLATILE
  COST 100;

or in pseudo-code I would have

for-each child_jsonb do
begin
  foreach (key-value in parent_jsonb) do
  begin
    child_jsonb  := jsonb_set(child_jsonb , '{key}', '"value"');
  end
  update child_jsonb in db;
end;

plv8 snippet:
$BODY$var ids = plv8.execute('select id from devices');

var CFG_TABLE_NAME = 'configurationj2b';
var selPlan = plv8.prepare( "select c.config from " + CFG_TABLE_NAME + " c where c.id = $1", ['int'] );
var updPlan = plv8.prepare( 'update ' + CFG_TABLE_NAME + ' set config = $1 where id = $2', ['json','int'] )

try {

    for (var i = 0; i < ids.length; i++) {
        var db_cfg = selPlan.execute([ids[i].id]);
        var cfg = db_cfg[0].config;
        var cfg_data = cfg['data'];
        cfg_data['OID1'] = 'plv8_01';
        if (cfg_data['OID1'] == 'plv8_01') {
            cfg_data['OID2'] = 'plv8_02'
        };
        if (cfg_data['OID2'] == 'plv8_02') {
            cfg_data['OID3'] = 'plv8_03'
        }
        if (cfg_data['OID3'] == 'plv8_03') {
            cfg_data['OID4'] = 'plv8_04'
        }
        if (cfg_data['OID4'] == 'plv8_04') {
            cfg_data['OID5'] = 'plv8_05'
        }


        updPlan.execute([cfg, ids[i].id]);
        plv8.elog(NOTICE, "UPDATED = " + ids[i].id);
    }

} finally {
    selPlan.free();
    updPlan.free();
}

return;$BODY$

but for now plv8 has other issues related to resource consumption.

So could I get similar performance in PL/pgSQL?
Reply | Threaded
Open this post in threaded view
|

Re: jsonb_set performance degradation / multiple jsonb_set on multiple documents

Michel Pelletier
I don't know the details of jsonb_set, Perhaps the '||' operator will perform better for you, it will overwrite existing keys, so you can build your new values in a new object, and then || it to the original.

postgres=# select '{"a": 1, "b": 2, "c": 3}'::jsonb || '{"b": 4, "c": 5}'::jsonb;
         ?column?         
--------------------------
 {"a": 1, "b": 4, "c": 5}
(1 row)

-Michel



On Fri, Mar 15, 2019 at 9:02 AM Alexandru Lazarev <[hidden email]> wrote:
Hi PostgreSQL Community.

I tried to rewrite some plv8 stored procedures, which process in bulk JSONB documents, to PL/pgSQL.
A SP usually has to delete/update/add multiple key with the same document and do it for multiple documents (~40K) in loop.

When updating a single key PL/pgSQL wins against plv8, but when I need to update multiple keys with jsonb_set, timing increase linearly with number of jsonb_sets and takes longer than similar SP in PLV8.
Below are test-cases I've used.

QUESTION: Is it expected behavior or I do something wrong or there are some better approaches or we can treat datum as object?

test case:
PG 9.6, CentOS 7

CREATE TABLE public.configurationj2b
(
  id integer NOT NULL PRIMARY KEY,
  config jsonb NOT NULL
);
Each jsonb column has 3 top keys, and one of top-key ('data') has another 700-900 key-value pairs e.g. {"OID1":"Value1"}

PL/pgSQL SP
CREATE OR REPLACE FUNCTION public.process_jsonb()
  RETURNS void AS
$BODY$
DECLARE
    r integer;
    cfg jsonb;
BEGIN
RAISE NOTICE 'start';
    FOR r IN
        SELECT id as device_id FROM devices
    LOOP
        select config into cfg from configurationj2b c where c.id = r; --select jsonb one by one

    -- MULTIPLE KEYs, Conditional Busiines Logic (BL) updates
    cfg := jsonb_set(cfg, '{data,OID1}', '"pl/pgsql1"');
    IF cfg@>'{"data" : { "OID1":"pl/pgsql1"} }' THEN
        cfg := jsonb_set(cfg, '{data,OID2}', '"pl/pgsql2"');
    END IF;

    IF cfg@>'{"data" : { "OID2":"pl/pgsql2"} }' THEN
        cfg := jsonb_set(cfg, '{data,OID3}', '"pl/pgsql3"');
    END IF;

    IF cfg@>'{"data" : { "OID3":"pl/pgsql3"} }' THEN
        cfg := jsonb_set(cfg, '{data,OID4}', '"pl/pgsql4"');
    END IF;

    IF cfg@>'{"data" : { "OID4":"pl/pgsql4"} }' THEN
        cfg := jsonb_set(cfg, '{data,OID5}', '"pl/pgsql5"');
    END IF;

   
    update     configurationj2b c set config = cfg where c.id = r;

    END LOOP;
    RAISE NOTICE 'end';
    RETURN;
END
$BODY$
  LANGUAGE plpgsql VOLATILE
  COST 100;

or in pseudo-code I would have

for-each child_jsonb do
begin
  foreach (key-value in parent_jsonb) do
  begin
    child_jsonb  := jsonb_set(child_jsonb , '{key}', '"value"');
  end
  update child_jsonb in db;
end;

plv8 snippet:
$BODY$var ids = plv8.execute('select id from devices');

var CFG_TABLE_NAME = 'configurationj2b';
var selPlan = plv8.prepare( "select c.config from " + CFG_TABLE_NAME + " c where c.id = $1", ['int'] );
var updPlan = plv8.prepare( 'update ' + CFG_TABLE_NAME + ' set config = $1 where id = $2', ['json','int'] )

try {

    for (var i = 0; i < ids.length; i++) {
        var db_cfg = selPlan.execute([ids[i].id]);
        var cfg = db_cfg[0].config;
        var cfg_data = cfg['data'];
        cfg_data['OID1'] = 'plv8_01';
        if (cfg_data['OID1'] == 'plv8_01') {
            cfg_data['OID2'] = 'plv8_02'
        };
        if (cfg_data['OID2'] == 'plv8_02') {
            cfg_data['OID3'] = 'plv8_03'
        }
        if (cfg_data['OID3'] == 'plv8_03') {
            cfg_data['OID4'] = 'plv8_04'
        }
        if (cfg_data['OID4'] == 'plv8_04') {
            cfg_data['OID5'] = 'plv8_05'
        }


        updPlan.execute([cfg, ids[i].id]);
        plv8.elog(NOTICE, "UPDATED = " + ids[i].id);
    }

} finally {
    selPlan.free();
    updPlan.free();
}

return;$BODY$

but for now plv8 has other issues related to resource consumption.

So could I get similar performance in PL/pgSQL?