I really appreciate your advices because I’m wandering at Google for clarifying how to check the effect of the number of servers at backend. Thank you so much.
I summarize your advices.
- Rather than the
number of servers, theconcurrent connections(Memory Usage) and thetraffic rate, thethe number of TLS handshakes per second(CPU Usage) will consume the most of computing resources, CPU and RAM. - Servers do not waste a buffer anymore when the checks are not running.
- And
NICandHAProxyshould run on the same CPU.
Could I ask some questions more?
- As I’m a newbie at sw and network area, I don’t know what to do for estimating the consumed resources (such as 2.7kB memory per server, 17 and 35 kB memory per active client connection). How can I estimate these things by myself? For example, for estimating the effect of
number of servers at backendat HAProxy, we can’t buy larger number number of servers` just for testing. I try to estimate it by reading the HAProxy source code but it seems not to help. - You said
data rateandTLS handshakeburden CPU. Is it from establishing connections towards backend and forwarding data and connection? - To make
NICandHAProxyrun on the same CPU, do I need to set some config athaproxy.cfgor some linux kernel config? (I check your other post, Architectural limitation for nbproc? - #6 by willy. It saidnumactl)
I’m sorry if these questions are too basic and Thank you for your time.