Dear SingTel, please fix your routing performance!

Slashdot asked for the latency to their site, so I ran a few tests with pretty consistent results in the 250ms range. Slashdot labels that as "still stuck on dialup or in space". Their expectations for overseas users was 80 to 150ms. Puzzled by the result (local websites typically respond in the sub 2ms range), I used traceroute to get to the bottom of this. Here are the results:

Hop	IP	DNS Name	Packet 1	Packet 2	Packet 3	AVG	Delta
1	192.168.1.1	router	0.3730	0.5390	0.3200	0.4107
2	202.166.123.170	(202.166.123.170)	5.9160	4.8020	4.7630	5.1603
3	202.166.123.169	(202.166.123.169)	2.7000	2.6150	2.8210	2.7120	-47.45%
4	202.166.121.101	(202.166.121.101)	3.7700	2.8390	2.5490	3.0527	12.56%
5	202.166.120.186	(ae6-0.singha.singnet.com.sg)	2.8160	7.1180	3.6210	4.5183	48.01%
6	202.166.126.41	(ae5-0.beck.singnet.com.sg)	29.7140	3.1530	2.9850	11.9507	164.49%
7	203.208.190.130	(203.208.190.130)	6.1800	3.6980	3.0630	4.3137	-63.90%
8	203.208.182.153	(ge-4-0-6-0.sngc3-cr2.ix.singtel.com)	3.0040
	203.208.151.177	(ge-0-1-7-0.sngtp-dr1.ix.singtel.com)		3.9300	2.9730	3.3023	-23.44%
9	203.208.171.186	(203.208.171.186)	180.7350
	203.208.153.162	(203.208.153.162)		193.0680	192.9220	188.9083	5620.45%
10	203.208.171.154	(203.208.171.154)	185.7070
	203.208.171.158	(203.208.171.158)		189.3080	189.1780	188.0643	-0.45%
11	67.17.192.141	(67.17.192.141)	185.5320	181.1660	181.5870	182.7617	-2.82%
12	64.208.27.50	(savvis-1.ar4.SJC2.gblx.net)	192.5370	185.9260	185.3730	187.9453	2.84%
13	206.28.98.117	(cr2-te-0-5-0-1.sfo.savvis.net)	185.3240	201.3570	202.9370	196.5393	4.57%
14	204.70.196.246	(cr2-tengig-0-7-0-0.chicago.savvis.net)	239.1180	234.3400	234.1810	235.8797	20.02%
15	204.70.195.122	(hr2-tengigabitethernet-12-1.elkgrovech3.savvis.net)	228.7670	240.4970	240.9600	236.7413	0.37%
16	64.37.207.158	(das5-v3032.ch3.savvis.net)	240.7050	244.1670	243.6270	242.8330	2.57%
17	64.27.160.194	(64.27.160.194)	240.7910	264.4990	253.7380	253.0093	4.19%
18	216.34.181.45	(slashdot.org)	235.7960	242.7860	242.3070	240.2963	-5.02%

Checking for the hosts that created that 5000% jump in latency it turns out, they are SingTel's:

stw@box:~$ whois 203.208.171.186
% [whois.apnic.net node-3]
% Whois data copyright terms    http://www.apnic.net/db/dbcopyright.html

inetnum: 203.208.171.128 - 203.208.171.191
netname: SINGTEL-IX-AP
descr: Singapore Telecommunications Pte Ltd
descr: 31C Exeter Road, Comcenter III
descr: Unit #06-06
descr: Singapore 239734
country: SG
admin-c: SAK3-AP
tech-c: SAK3-AP
status: ASSIGNED NON-PORTABLE
notify: shanali@singtel.com
mnt-by: MAINT-SINGTEL-IX
changed: shanali@singtel.com 20070615
source: APNIC

stw@box:~$ whois 203.208.153.162
% [whois.apnic.net node-5]
% Whois data copyright terms 

inetnum: 203.208.153.128 - 203.208.153.191
netname: SINGTEL-IX-AP
descr: Singapore Telecommunications Pte Ltd
descr: 31C Exeter Road, Comcenter III
descr: Unit #06-06
descr: Singapore 239734
country: SG
admin-c: SAK3-AP
tech-c: SAK3-AP
status: ASSIGNED NON-PORTABLE
notify: shanali@singtel.com
mnt-by: MAINT-SINGTEL-IX
changed: shanali@singtel.com 20070615
source: APNIC

Looks very much like an underperforming switch/router/filter. SingTel, please fix that!