A question about nodebalancer limits

about 10,000 concurrent connections

does the concurrent counting also keep alive connection? so if my nginx has keep alive for 30 seconds every request will be counted for at least 30 seconds?

if I have the math correctly for a 30 seconds keep alive I can get top of 333 consist requests per seconds with 1 nodebalancer?

if it correct so for supporting 4000-3000 rps I will need do dns round robin a 12 nodebalancer?

yes I know that 3000 rps is a lot but I have a specific use case of plain html serving so I can pretty easily support this load

6 Replies

If the same client sends multiple requests through the same connection (which is the whole point of keepalive), it will only count as one connection.

If 3000 clients connect to your server and each client sends 1 request per second, that's still only 3000 connections if they all use keepalive.

@hybinet:

If the same client sends multiple requests through the same connection (which is the whole point of keepalive), it will only count as one connection.

If 3000 clients connect to your server and each client sends 1 request per second, that's still only 3000 connections if they all use keepalive.

thanks but if I every connection with keepalive taking a 1 spot of the 10,000 for 1 minute and I have a burst of 3000 unique users per second for a period of 1 minute after 3 seconds I will not be able to accept new connections

just to note: linode nodebalancer http and https does not support keepalive it is off by default and you cant turn in on

too many limitation with this nodebalancer

@boynet:

just to note: linode nodebalancer http and https does not support keepalive it is off by default and you cant turn in on

too many limitation with this nodebalancer
thanks for that heads up

probably better just to roll out your own haproxy based load balancer setup :)

@centminmod:

@boynet:

just to note: linode nodebalancer http and https does not support keepalive it is off by default and you cant turn in on

too many limitation with this nodebalancer
thanks for that heads up

probably better just to roll out your own haproxy based load balancer setup :)

yap the problem with own haproxy you need to:

create cluster of haproxy for 1 point of failure

and also you network speed will be limited by each haproxy node

indeed though linode network in speed is already quite fast

linode newark

./speedtest-cli --share --server 6030
Retrieving speedtest.net configuration...
Testing from Linode (45.xxx.xxx.xxx)...
Retrieving speedtest.net server list...
Selecting best server based on ping...
Hosted by fdcservers.net (New York City, NY) [14.26 km]: 2.946 ms
Testing download speed................................................................................
Download: 1811.86 Mbit/s
Testing upload speed................................................................................................
Upload: 1429.16 Mbit/s
Share results: http://www.speedtest.net/result/6477884908.png

Reply

Please enter an answer
Tips:

You can mention users to notify them: @username

You can use Markdown to format your question. For more examples see the Markdown Cheatsheet.

> I’m a blockquote.

I’m a blockquote.

[I'm a link] (https://www.google.com)

I'm a link

**I am bold** I am bold

*I am italicized* I am italicized

Community Code of Conduct