This file is indexed.

/usr/share/doc/sqlgrey/README.PERF is in sqlgrey 1:1.8.0-1.

This file is owned by root:root, with mode 0o644.

The actual contents of the file can be viewed below.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
On large sites, database performance can suffer. Here are some tips.

# One SQLgrey on DB vs one SQLgrey on each MTA

Usually, for robustness reasons you should run one SQLgrey on each
MTA and configure Postfix to ask the local SQLgrey instance.
The problem is that when you use a separate database, SQLgrey queries
have to be done accross the network and there usually are several sequential
queries for only one message.
You can minimise latency issues by having only one SQLgrey instance
sitting on the database host. The huge flaw in this architecture is that
if the database host breaks your whole email infrastructure breaks too
(SQLgrey is designed to work around database unavailabilities).
The fact that queries are serialised at the SQLgrey level might also help
performance by avoiding too much parallel queries.

# Use DBClustering

Allows you to distribute load between several database servers. 
See README.DBCLUSTER

# Use Discrimination

Alternative approach to greylisting where you make selective rules to allow some
hosts to bypass greylisting entirely, based on how they "look" and "talk".
See README.DISCRIMINATION

# Optin/optout related performance

See README.OPTINOUT, ## Performance guidelines

# Put most common hosts in whitelists

If your logs show that much of your traffic is coming from a handfull
of servers, put them in the whitelists: SQLgrey won't have to query the
database.

You can use sqlgrey-logstats.pl to point out the most common sources in
your AWLs.

All whitelists types don't share the same speed. The quickest are the IP
based whitelists and the full fqdn entries in the fqdn whitelists.
regexps and domain (*.domain.name) in fqdn whitelists are noticeably slower.

# Add RAM to the database host

If you want to put money somewhere, unless you have an uncommon setup, RAM
on the DB should be the best place to invest.