Hi, Luca Morandini wrote:
> On 03/21/2011 09:16 AM, Arne Kepp wrote: > >> > > Not sure about the intermittent slowdowns, > > That could suggest some swapping taking place (hence slowing > it down) until the > memory load relents and no swapping is needed any more... > unless it drops > requests, freeing up resources. Requests were not dropped, service was fine but every now and then there were a few seconds dead period. > > but in general it is not that > > simple. mod_proxy doubles the number of threads*, and the number of > > TCP/IP connections needed. > > Hmm... I supposed the HTTPd ran on a different server than > the one hosting > GWC/GeoServer, hence the additional load should not fall on > the latter. Wrong, Apache was running on the same server. > > Last time I checked GWC throughput peaked > > somewhere between 2000 or 4000 simultaneous threads > (against a realistic > > set of clients, a mixture of slow and fast, not gigabit > ethernet clients). > > Meaning some connections were longer-lived, hence heavier on > resources, both on > HTTPd and GWC/GeoServer. > > > > At that point the additional TCP/IP connections + threads > in Apache are > > not neglible anymore. > > Indeed: I stand corrected. > > By the way, when the load is expected to be heavy, I would > push GWC to the edge of > the network, de-copupling it from GeoServer and putting it on > the front-end > (HTTPd) server, where any static content should be. > > Regards, > > Luca Morandini > http://www.lucamorandini.it -Jukka Rahkonen- ------------------------------------------------------------------------------ Colocation vs. Managed Hosting A question and answer guide to determining the best fit for your organization - today and in the future. http://p.sf.net/sfu/internap-sfd2d _______________________________________________ Geoserver-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/geoserver-users
