Hi all,

we have 2 datacenters in different region. They both use the same ISP (connected to 2 differen POPs of course) with connection in one DC 1 GIG in the second 10GIG.

I'm trying to find out why the connection between the systems in both datacenters is so slow (I'm using iperf to show the result, but the real application shows something like that):

first with TCP:

./iperf -c XXX -p 10111

------------------------------------------------------------

Client connecting to XXX, TCP port 10111

TCP window size: 16.0 KByte (default)

------------------------------------------------------------

[ 5] local YYY port 35822 connected with XXX port 10111

[ ID] Interval Transfer Bandwidth

[ 5] 0.0-10.7 sec 728 KBytes 557 Kbits/sec

Second with UDP:

./iperf -u -c XXX -p 10111

------------------------------------------------------------

Client connecting to XXX, UDP port 10111

Sending 1470 byte datagrams

UDP buffer size: 64.0 KByte (default)

------------------------------------------------------------

[ 5] local YYY port 53908 connected with XXX port 10111

[ ID] Interval Transfer Bandwidth

[ 5] 0.0-10.0 sec 1.25 MBytes 1.05 Mbits/sec

[ 5] Server Report:

[ 5] 0.0-10.0 sec 1.23 MBytes 1.03 Mbits/sec 0.343 ms 19/ 893 (2.1%)

[ 5] Sent 893 datagrams

One can start in paralell many connections between the same pair of hosts in different datacenter then you can see each connection get the same speed ~1Mbit/s. if you start 100 paralell sessions (e.g. ftp downloads) then you'll get 100Mbit, but for the single conection only ~1Mbit/s.

Internal communication in both DCs is OK - up to the limit of access port speed 100MBit.