On 8/30/05, Dustin M Snell [Network Automation] <[EMAIL PROTECTED]> wrote: > Adam, > > Hey! Speaking of cheap -- that's kind of a cheap shot! I'm not cheap, but we > are talking about a small business here that still has a requirement for > maximum uptime. > > I wouldn't mind paying twice as much for two servers - even though the 2nd > isn't even doing anything most of the time. I don't think it's much to ask > that it be able to scale to that level without having to increase our > initial investment 8x over. >
While not speaking for Adam (he can certainly do that for himself), I think his point, indirectly, was that you have to weigh the actual importance of constant uptime vs. cost. For my company, for example, it is run entirely through our web site -- there's no shrink-wrapped product, no magazines, etc. Therefore, as a smaller company ourselves, we made the business decision that we would lose money beyond recovery if our site was down for an extended period of time. As such, the thousands of dollars we paid to be in a clustered environment (two physical servers each running full versions of JRun with CF Enterprise deployed in each behind a BigIP load balancer) is a drop in the bucket and a no-brainer. That being said, the situation is might be entirely different for you. In Adam's case, I believe he works for the US State Department. That's not one of my government agencies that I'd want to hear has constant downtime! (And $10K in the context of the State Department's budget is nothing anyway.) As others have noted, there are ways to achieve high availability without buying top shelf hardware and Enterprise versions of ColdFusion. Personally, I'd start by looking at your code, and, in particular, any potentially long running queries to see if that's what's causing your spikes. Make sure you're using the recently released JDBC driver upgrades and that they're properly installed. Turn on the "log long running requests" option in the CF Admin and set the threshold to a value right at what you think the longest page response should be. I think what you'll find is that most issues are with the codebase and not the CF server product itself. I've consulted on numerous engagements where the client was blaming the hardware, but there was zero effort to cache anything and their code was constantly making query calls to pull back 10,000 records just to get a record count! All of the above being said, there is great value in failover availability. Others have already chimed in with relatively cheap solutions you could try out if your code analysis is fully optimized. Regards, Dave. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| Find out how CFTicket can increase your company's customer support efficiency by 100% http://www.houseoffusion.com/banners/view.cfm?bannerid=49 Message: http://www.houseoffusion.com/lists.cfm/link=i:4:216880 Archives: http://www.houseoffusion.com/cf_lists/threads.cfm/4 Subscription: http://www.houseoffusion.com/lists.cfm/link=s:4 Unsubscribe: http://www.houseoffusion.com/cf_lists/unsubscribe.cfm?user=11502.10531.4 Donations & Support: http://www.houseoffusion.com/tiny.cfm/54