[Bucardo-general] Latency on delete

[Alex Balashov]

Hi,

How long is a DELETE of ~4.5 million rows (in a relatively narrow table) 
supposed to take in a scenario with a simple pushdelta replication 
setup?  Mine took 1 hour and 45 minutes.  While I realise that Bucardo's 
row level-trigger based system incurs some additional latency, the 
latency is just inserting those deltas, by and large, right?  It 
shouldn't take quite so long as that, I would hypothesise.

Both databases - including the 'bucardo' schema of the source - are 
tidily vacuumed nightly.

The servers are both quad-core 2 GHz Xeons with 32 GB of RAM each with 
decent SATA drives on a Dell PERC.

In addition, the following PostgreSQL tuning measures are in place:

    shared_buffers = 512 MB
    work_mem = 1024MB
    wal_level = minimal
    fsync = off
    synchronous_commit = off

The 'bucardo_ctl status' output does not show any delays, failures or 
other anomalies.

The table on which I ran the DELETE statement had about 38 million rows, 
of which I was attempting to delete 4.5.  The table also has two indexes 
(primary key index and one additional indexed column) and a foreign key 
constraint.

Is it reasonable and expected for this DELETE to have taken close to two 
hours?

Thank much!

-- Alex

-- 
Alex Balashov - Principal
Evariste Systems LLC
235 E Ponce de Leon Ave
Suite 106
Decatur, GA 30030
United States
Tel: +1-678-954-0670
Web: http://www.evaristesys.com/, http://www.alexbalashov.com/