Hello, I'm writing to add more information. I've done a few test to compare ssh -L and stunnel4 with spiped.
In 5 seconds, the server could run 50 queries with spiped. With ssh -L, the server ran 2700 queries. And with stunnel4, it ran 3000 queries. Always testing with the same hardware and same network. Am I missing something? I've used wireshark to capture traffic and noticed big gaps between packets in Spiped (I'm not expert at all regarding TCP). Instead with, ssh -L there are no gaps. Is there any kind of buffering in spiped? I guess the problem is the latency added in every query. This shouldn't be a problem regarding bandwith. Thanks! -- Mauro Ciancio On Wed, Jan 21, 2015 at 9:28 AM, Mauro Ciancio <[email protected]> wrote: > Hi there, I've started using spiped to connect a client with a mysql > server. > I'm using spiped in the server and in the client. These are the commands > I'm running: > > In the client: > */bin/spiped -F -e -s 'localhost:3306' -t 'lan.server:13306' -k key.key* > > In the server: > */bin/spiped -F -d -s 'lan.localhost:13306' -t 'localhost:3306' -k key.key* > > And then the connection between the mysql client and server is transparent. > > The problem I'm having is that spiped is adding latency (I guess so). > Obviously I'm not creating a new connection every time I need one (I have > a connection pool in the application server). > > I've done a quick test and replaced *spiped* with *ssh -L* and it > performed a lot better. > > Any ideas or things I could look to? > > I'm sorry for the vague details, I'm not sure how to provide numbers. I'm > willing to provide all the information needed to track down this problem. > > Ping between servers is as follows (I guess is not bad ~1mseg ping, right?) > $ ping lan.server > --- lan.server ping statistics --- > 14 packets transmitted, 14 received, 0% packet loss, time 13012ms > rtt min/avg/max/mdev = 0.750/0.959/1.190/0.137 ms > > Thanks you! > -- > Mauro Ciancio >
