Mike,
1) Is this a single machine, or multiple load balanced? Normally in prod environments you'd have multiple to provide some level of redundancy. One Server with Application (Ofbiz/BigFish) One Server with DB One Server with Apache 2) What type of machine is this? Is a VM, Amazon AWS, stand alone, number of cores, RAM, etc. Amazon AWS/EC2 instance with 2 CPUs, 7.5GB RAM 3) What were the java memory settings (you mentioned that you had to tweak them). Xms/Xmx = 5120M MaxPermSize = 1024m The teak was to allocate the same amount of memory to Xms and Xmx allocating JVM on application startup. 4) The cache optimizations you mention. Are these already included in the bigfish download/demo? 5) Are any of these documented? I noticed that bigfish has impressive documentation (well done!). Yes, you can also review the Cache settings in the BigFish Help Documentation. http://bigfish.solveda.com/help/guideCachingAndMemory.htm Regards Len -----Original Message----- From: Mike [mailto:mz4whee...@gmail.com] Sent: Thursday, February 13, 2014 4:29 PM To: user Subject: Re: OFBiz Performance, a good story These are pretty impressive numbers. A couple of questions Nick; 1) Is this a single machine, or multiple load balanced? Normally in prod environments you'd have multiple to provide some level of redundancy. 2) What type of machine is this? Is a VM, Amazon AWS, stand alone, number of cores, RAM, etc. 3) What were the java memory settings (you mentioned that you had to tweak them). 4) The cache optimizations you mention. Are these already included in the bigfish download/demo? 5) Are any of these documented? I noticed that bigfish has impressive documentation (well done!). Thanks! On Thu, Feb 13, 2014 at 4:06 AM, Nick Rosser <nros...@solveda.com> wrote: > All, > > I thought I'd share some encouraging news from a recent implementation we > were involved in: > > * we built a site for a Philly based jeweler who has some unique > products geared for Valentine's Day > * they start their sale on 1-Jan, it typically gains real momentum by > the end of Jan, and peaks around the 11th, 12th Feb > * peak #visitors (day): 20,000 > * peak #pageviews (day): 150,000 > * peak #orders (day): ~4000 > * peak #concurrent users: 275 > > And perhaps the most impressive fact around performance was that during > peak the site was flying, no noticeable difference if there was a single > user or 275 concurrent users. > > My conclusion is that the OFBiz caching model works really well when > configured correctly (together with good memory allocation and settings). > We also spent a great deal of time tuning our BigFish solution, > specifically around the usage of the entity model layer (really just making > sure we always use cache calls). We logged all SQL calls, and then analyzed > the code to remove a second execution of any SQL call (since it should be > cached). The Grinder was used to monitor and gauge performance > improvements. All in all very encouraging and gives great confidence of > marketing this solution into the largest organizations. > > I would personally like to thank the key committers who built the caching > functionality -- it's a little tricky to understand, use and configure but > once that is overcome it works impressively. > > We have been involved in some IBM Websphere Commerce implementations and > I'm really beginning to wonder why the software and implementations are > 10-20 times more expensive. With OFBiz we have an enterprise solution, with > affordability for a small business. > > I'd be interested in any stories in the community around performance and > high levels of order taking --- these case studies can really increase > adoption of OFBiz. Stability and scalability are often at the top of > anyone's list when looking at a new platform. Any stories out there? > > Check out the site: www.ihatestevensinger.com. It's built using our > BigFish solution. More at http://bigfish.solveda.com. > > Nick > >
