Hi all

I set up my linux router as a simple NAT router. I use CentOS 5.4. When I set up ISP proxy IP in the browser of client PC, Internet access is fast. When I remove ISP proxy IP from the browser, Internet access is slow.

ISP use transparent caching but I can use manual caching as well. Clients on my network with transparent caching get slow internet access and using ISP proxy get fast internet access.

What's the problem?

ISP announce that we can use transparent caching or manual caching.

So, I set up my linux router with squid. I set up cache_peer TAG point to ISP's proxy IP. But the problem is the same. Using my squid proxy is slow and using ISP proxy directly is fast. All other network settings are correct.

How can I improve my internet connection using transparent caching. I don't want to set up proxy IP address on all clients.

Do please ping to ISP proxy and to www.google.com, like:
ping ISP
ping www.google.com

Then "traceroute" to www.google.com and to ISP proxy.

Post results here, please.

ping result to ISP proxy

Pinging 203.81.27.200 with 32 bytes of data:
Reply from 203.81.27.200: bytes=32 time=530ms TTL=44
Reply from 203.81.27.200: bytes=32 time=531ms TTL=44
Reply from 203.81.27.200: bytes=32 time=545ms TTL=44
Reply from 203.81.27.200: bytes=32 time=508ms TTL=44

Ping statistics for 203.81.27.200:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 508ms, Maximum = 545ms, Average = 528ms

Ping result to Google.

Pinging www.l.google.com [209.85.231.99] with 32 bytes of data:
Reply from 209.85.231.99: bytes=32 time=705ms TTL=45
Reply from 209.85.231.99: bytes=32 time=716ms TTL=45
Reply from 209.85.231.99: bytes=32 time=732ms TTL=45
Reply from 209.85.231.99: bytes=32 time=794ms TTL=45

Ping statistics for 209.85.231.99:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 705ms, Maximum = 794ms, Average = 736ms

where are traceroute results?
what of caching: clients request the same site through your squid and get slow reply? probably problem is not with network then(from your ping, it seems like network itself is a problem), but in your squid conf.

Here's the tracert result.

Tracing route to www.l.google.com [209.85.231.99]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms horizon-server.horizon.local [192.168.0.1]
2 <1 ms <1 ms <1 ms 192.168.1.254
3 * * * Request timed out.
4 * * * Request timed out.
5 94 ms 89 ms * gateway.mpt.net.mm [203.81.65.13]
6 84 ms 95 ms 87 ms 203.81.81.5
7 * 88 ms 70 ms 218.100.58.33
8 234 ms 219 ms 216 ms 195.22.198.49
9 297 ms * 348 ms 72.14.218.58
10 342 ms 343 ms 302 ms 72.14.218.58
11 275 ms 262 ms 263 ms 216.239.47.128
12 358 ms * 288 ms 209.85.249.234
13 * 279 ms 302 ms 72.14.233.104
14 358 ms 335 ms 364 ms 216.239.43.90
15 507 ms 384 ms 395 ms 216.239.43.192
16 470 ms 418 ms 371 ms 216.239.46.14
17 * 395 ms 563 ms 216.239.43.213
18 590 ms 579 ms 590 ms 209.85.255.36
19 739 ms 730 ms 660 ms 216.239.43.213
20 711 ms 599 ms 731 ms 66.249.94.75
21 773 ms 747 ms 719 ms 72.14.238.74
22 941 ms 957 ms 794 ms maa03s01-in-f99.1e100.net [209.85.231.99]

Trace complete.

It is important to compare two traceroute: to your ISP proxy and to somewhere else.

Thanks.

P.S.
Atctually, your traceroute shows:
1 <1 ms <1 ms <1 ms horizon-server.horizon.local [192.168.0.1]
2 <1 ms <1 ms <1 ms 192.168.1.254
5 94 ms 89 ms * gateway.mpt.net.mm [203.81.65.13]
6 84 ms 95 ms 87 ms 203.81.81.5

But ping: Reply from 203.81.27.200: bytes=32 time=530ms

Looks like the problem somewhere between you and ISP

Because if ping to ISP - 530 ms, but to Google 700 ms, 700-530 = 170 ms.
170 ms from Beijing, China to US, California - that is very good.