Re: nested query on last n rows of huge table

sorry: Postgres 8.0.2 server. The EXPLAIN ANALYZE for the query looks
like this:

explain analyze select
syslog.logs.eventtime,assets.hosts.name,syslog.processes.name as
process
from
syslog.logs,assets.hosts,assets.ipaddrs,assets.macaddrs,syslog.processes
where msg like '%session opened for user root%'
and syslog.logs.assets_ipaddr_id = assets.ipaddrs.id
and assets.ipaddrs.macaddr_id = assets.macaddrs.id
and assets.macaddrs.host_id = assets.hosts.id
and syslog.processes.id = syslog.logs.process_id and
eventtime > timestamp '2006-02-24 15:05:00'

Nested Loop (cost=0.00..328832.34 rows=2 width=254) (actual
time=49389.924..49494.665 rows=45 loops=1)
-> Nested Loop (cost=0.00..328826.32 rows=1 width=90) (actual
time=49365.709..49434.500 rows=45 loops=1)
-> Nested Loop (cost=0.00..328820.30 rows=1 width=90) (actual
time=49327.211..49360.043 rows=45 loops=1)
-> Nested Loop (cost=0.00..328814.27 rows=1 width=90)
(actual time=49327.183..49344.281 rows=45 loops=1)
-> Seq Scan on logs (cost=0.00..328809.04 rows=1
width=16) (actual time=49314.928..49331.451 rows=45 loops=1)
Filter: (((msg)::text ~~ '%session opened for
user root%'::text) AND (eventtime > '2006-02-24 15:05:00'::timestamp
without time zone))
-> Index Scan using "pk_syslog.processes" on
processes (cost=0.00..5.21 rows=1 width=82) (actual time=0.278..0.280
rows=1 loops=45)
Index Cond: (processes.id =
"outer".process_id)
-> Index Scan using "pk_assets.ipaddrs" on ipaddrs
(cost=0.00..6.01 rows=1 width=8) (actual time=0.344..0.346 rows=1
loops=45)
Index Cond: ("outer".assets_ipaddr_id = ipaddrs.id)
-> Index Scan using "pk_assets.macaddrs" on macaddrs
(cost=0.00..6.01 rows=1 width=8) (actual time=1.648..1.650 rows=1
loops=45)
Index Cond: ("outer".macaddr_id = macaddrs.id)
-> Index Scan using "pk_assets.hosts" on hosts (cost=0.00..6.01
rows=1 width=172) (actual time=1.330..1.331 rows=1 loops=45)
Index Cond: ("outer".host_id = hosts.id)
Total runtime: 49494.830 ms