[concurrency-interest] Blocking vs. non-blocking
stanimir at riflexo.com
Thu Aug 7 02:08:58 EDT 2014
On Thu, Aug 7, 2014 at 1:25 AM, Zhong Yu <zhong.j.yu at gmail.com> wrote:
> I mean there's a constant ratio between the two numbers, therefore we
> can estimate the magnitude of connections from the number of requests
> per second. We can put a generous upper bound on the ratio at 100
> second, and bet that Google Search maintains less than 10M concurrent
> connections globally
This is true only for http-alike workloads. For instance pushing market
data (stocks/forex) results into hundreds (or thousands) of
packets/messages a second sent per connection and virtually no requests as
the model is largely subscribe and forget. And sometimes it's "forget for
In such cases sending the messages would require context switch on each
message which is non-trivial.
> 4 at 5:15 PM, Zhong Yu <zhong.j.yu at gmail.com> wrote:
> >> On Wed, Aug 6, 2014 at 2:12 PM, Vitaly Davidovich <vitalyd at gmail.com>
> >> wrote:
> >> > what would happen if you have 100k+ connections?
> >> You'd rule the world. To put the number in perspective, google
> >> receives less than 100k searches per second.
> >> Zhong Yu
> >> bayou.io
> Concurrency-interest mailing list
> Concurrency-interest at cs.oswego.edu
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Concurrency-interest