[squid-users] Throughput slower, when data is in cache instead geting it from the webserver

Hi,

we use squid as shipped with SuSE Linux Enterprise 9
(squid-2.5.STABLE5-42.41).

The throughput is slower, when I get the data from cache and is faster,
when I use the -r option to get the data from the webserver.

--snip--
squid06:~ # time squidclient -p 8080 -r http://mueller.datevnet.de/Richter.pdf >/dev/null

real 0m0.702s
user 0m0.004s
sys 0m0.015s
squid06:~ # time squidclient -p 8080 http://mueller.datevnet.de/Richter.pdf >/dev/null

real 0m6.085s
user 0m0.001s
sys 0m0.002s
squid06:~ # time squidclient -p 8080 -r http://mueller.datevnet.de/Richter.pdf >/dev/null

real 0m0.881s
user 0m0.003s
sys 0m0.012s
squid06:~ # time squidclient -p 8080 http://mueller.datevnet.de/Richter.pdf >/dev/null

real 0m5.785s
user 0m0.001s
sys 0m0.003s
squid06:~ # time squidclient -p 8080 -r http://mueller.datevnet.de/Richter.pdf >/dev/null

real 0m0.742s
user 0m0.005s
sys 0m0.012s
squid06:~ # time squidclient -p 8080 http://mueller.datevnet.de/Richter.pdf >/dev/null

real 0m5.890s
user 0m0.001s
sys 0m0.002s
squid06:~ #
--snip--

here the logfile entries:

--snip--
1153747089.142 698 127.0.0.1 TCP_CLIENT_REFRESH_MISS/200 2274238 GET http://mueller.datevnet.de/Richter.pdf - FIRST_UP_PARENT/127.0.0.1 application/pdf
1153747098.419 6081 127.0.0.1 TCP_HIT/200 2274245 GET http://mueller.datevnet.de/Richter.pdf - NONE/- application/pdf
1153747102.320 878 127.0.0.1 TCP_CLIENT_REFRESH_MISS/200 2274238 GET http://mueller.datevnet.de/Richter.pdf - FIRST_UP_PARENT/127.0.0.1 application/pdf
1153747109.546 5782 127.0.0.1 TCP_HIT/200 2274245 GET http://mueller.datevnet.de/Richter.pdf - NONE/- application/pdf
1153747111.639 739 127.0.0.1 TCP_CLIENT_REFRESH_MISS/200 2274238 GET http://mueller.datevnet.de/Richter.pdf - FIRST_UP_PARENT/127.0.0.1 application/pdf
1153747118.797 5887 127.0.0.1 TCP_HIT/200 2274245 GET http://mueller.datevnet.de/Richter.pdf - NONE/- application/pdf
--snip--

This is my squid.conf

--snip--
squid06:~ # grep -v "^#" /etc/squid/squid.conf | grep -v "^$"
http_port 10.252.104.20:8080
http_port 10.252.104.80:8080
http_port 127.0.0.1:8080
icp_port 0
cache_peer 127.0.0.1 parent 8280 0 no-query no-digest no-netdb-exchange
hierarchy_stoplist cgi-bin ?
acl QUERY urlpath_regex cgi-bin \?
no_cache deny QUERY
cache_mem 512 MB
maximum_object_size 20480 KB
cache_dir diskd /var/cache/squid 15360 16 256
cache_store_log none
ftp_user datevnet@
ftp_list_width 50
auth_param basic children 15
auth_param basic realm DATEVnet Proxy-Server
auth_param basic credentialsttl 2 hours
auth_param basic program /usr/sbin/squid_ldap_auth -b
ou=Accounts,dc=datevnet,dc=de -R -f (&(uid=%s)(dvPerm=WEB)) -s sub -h
haldap.services.datevnet.de -p 389 -D
cn=admin,ou=proxy,ou=systems,dc=datevnet,dc=de -w proxy
refresh_pattern ^ftp: 1440 20% 10080
refresh_pattern ^gopher: 1440 0% 1440
refresh_pattern . 0 20% 4320
half_closed_clients off
shutdown_lifetime 5 seconds
acl all src 0.0.0.0/0.0.0.0
acl manager proto cache_object
acl localhost src 127.0.0.1/255.255.255.255
acl to_localhost dst 127.0.0.0/8
acl SSL_ports port 443 563
acl Safe_ports port 80 # http
acl Safe_ports port 21 # ftp
acl Safe_ports port 443 563 # https, snews
acl Safe_ports port 70 # gopher
acl Safe_ports port 210 # wais
acl Safe_ports port 1025-65535 # unregistered ports
acl Safe_ports port 280 # http-mgmt
acl Safe_ports port 488 # gss-http
acl Safe_ports port 591 # filemaker
acl Safe_ports port 777 # multiling http
acl CONNECT method CONNECT
acl Admin_ports port 1812 8445 # Viruswall-GUI
acl PURGE method PURGE
acl snmpread snmp_community public
acl mrtg_host src 10.252.0.88/255.255.255.248
acl adminclients src 10.252.16.0/255.255.255.0
acl adminusers proxy_auth A00000330020
acl password proxy_auth REQUIRED
acl transon-server dst 212.114.203.97/255.255.255.255
acl frustcenter dst 193.27.49.0/255.255.255.0
acl www_crl_esecure_de dst 193.27.50.195/255.255.255.255
acl nocacheservers dst 193.27.50.178/32 193.27.50.179/32
193.27.50.135/32 193.27.50.137/32
acl blockhostsip dst "/etc/squid/blockhosts.ip"
acl blockhostsdomain dstdomain "/etc/squid/blockhosts.domain"
acl cdbservers url_regex "/etc/squid/squid.cdbservers"
http_access allow manager localhost
http_access deny manager
http_access deny !Safe_ports
http_access deny CONNECT !SSL_ports
http_access deny Admin_ports
http_access allow PURGE localhost
http_access deny PURGE
http_access allow localhost
http_access deny blockhostsip
http_access deny blockhostsdomain
http_access allow transon-server
http_access allow frustcenter
http_access allow www_crl_esecure_de
http_access allow cdbservers
http_access allow password
http_access allow adminusers
http_access deny all
http_reply_access allow all
icp_access deny all
cache_effective_user squid
visible_hostname squid06.services.datevnet.de
unique_hostname squid06.services.datevnet.de
deny_info ERR_CUSTOM_BLOCKHOSTS blockhostsip
deny_info ERR_CUSTOM_BLOCKHOSTS blockhostsdomain
never_direct allow all !CONNECT
error_directory /usr/share/squid/errors/German_Datev
snmp_access allow snmpread mrtg_host
snmp_access deny all
snmp_incoming_address 10.252.104.80
coredump_dir /var/cache/squid
client_persistent_connections off
--snip--

I'm alone on this server, which has 2G Ram and 2 Xeon 3.4 GHz CPUs.
The cache_dir is a hardware raid1 with 36 GByte Space.

Does anybody have an idea why the throughput is lower, when I get the
data from cache instead from the webserver ?

-- 
Gruß
  Dieter
--
I do not get viruses because I do not use MS software.
If you use Outlook then please do not put my email address in your
address-book so that WHEN you get a virus it won't use my address in the
From field.