I'dd add to look at the metrics. If you have enterprise version installed, setup the jrun metrics. Google on 'jrun metircs' and you will get hits to how. Also, check out blogs from Robi Sen, he's got good info on things like this. Once you have metrics logging on, you can measure changes to simultaneous requests under load. I don't recall how to get this under a standard version install, sorry. Typical starting points are 3-4 simultaneous requests per CPU. After that its all up to your app, users, etc. For example, we have apps that have heavy DB work going on all the time and not much work for CF to do but sit and wait. We have this set to about 50 (25 on each of two clustered JRun instances). We get near 7000 logins, that's logins now not just hits, between 7am to 7pm. Again, things vary wildly depending on your application(s).
DK On 6/25/07, Wil Genovese <[EMAIL PROTECTED]> wrote: > > Joshua, > > We have ours set to 100 simultaneous requests per second. Here is our > situation. We have 1.5 million page view per day with an average of 67 > thousand visitors per day each spending about 20 minutes on our site. > > The only limit is your CPU power, memory your app uses and DB capacity. > It is my opinion that you can increase your simultaneous requests until > your server loads reaches about 50-60% average load during normal usage. > This will give you room for traffic spikes. Setting your # of requests > too high will have no effect on your server if you don't get enough > traffic to max out that setting. However, if you do get enough traffic > to max out that number and your CF and or DB servers can't handle the > load then your going to have problems. If you set your # of requests too > low and your traffic has that maxed out then your also going to have > problems. Keep adjusting until you find the right amount for your systems. > > Also, ignore the threads your seeing. They are listeners. > > > > -- > Wil Genovese > > One man with courage makes a majority. > -Andrew Jackson > > A fine is a tax for doing wrong. A tax is a fine for doing well. > > > > Joshua Woodcook wrote: > > I'm running CFMX7 on a dedicated linux machine w/apache. I ran a > standard install of cf and have not done any tuning/config changes for > accepting incoming connections. > > > > I'd like to determine if there is anything I need to do to ensure the > server can handle ~7000 unique visitors per day (with an avg of 3-4 > pageviews each; 98% of that traffic between 7am - 8pm) > > > > I was thinking I could just bump up the simultaneous requests in the CF > Admin interface . It's currently set for 8 simul. connections. How high > can I set this? I'm thinking it will be directly proportional to my > system's RAM (2GB) > > > > One thing I noticed when viewing running processes on the server (ps > -auxf), is that CF has spawned around 40 - 50 sub threads that look like > this: (/opt/coldfusionmx7/bin/cfmx7 -jar cfusion.jar -start coldfusion) > > > > Is each one of these listening and ready to accept incoming > connections? Why would CF spawn this many threads if the Admin setting is > configured to only accept 8 simultaneous connections? > > > > > > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| CF 8 â Scorpio beta now available, easily build great internet experiences â Try it now on Labs http://www.adobe.com/cfusion/entitlement/index.cfm?e=labs_adobecf8_beta Archive: http://www.houseoffusion.com/groups/CF-Server/message.cfm/messageid:6357 Subscription: http://www.houseoffusion.com/groups/CF-Server/subscribe.cfm Unsubscribe: http://www.houseoffusion.com/cf_lists/unsubscribe.cfm?user=11502.10531.10
