Thank you, Rupert. I'll give these guidelines a try as soon as I can.
I can't wait to get started :) -----Original Message----- From: Rupert Westenthaler [mailto:rupert.westentha...@gmail.com] Sent: Thursday, April 03, 2014 3:04 AM To: dev@stanbol.apache.org Subject: Re: Installation Guidelines Hi, Let me try to give you an overview ... On Wed, Apr 2, 2014 at 4:10 PM, Curry, Kalika [IT] <kalika.k.cu...@sprint.com> wrote: > Hi, I'm looking to install and use the Apache Stanbol services on a web > server and am having some trouble identifying the requirements for > installation. I apologize ahead of time if there's another resource I should > be using. Please direct me to it and I'll gladly use it. > > I can see what we need to install the services (Java 6 and maven 3.0.3), but > am having difficulty locating information pertaining to what hardware and > operating system necessary to use them. > Stanbol 0.12.0 [1] runs on Java 6. The 0.12.1-SNAPSHOT [2] and 1.0.0-SNAPSHOT versions [3] do require Java 7. Java 8 is not (yet) supported. For building I suggest to use the most recent maven release (3.2.1). Personally I am currently using maven 3.0.4 on Java 1.7.0_45. > Currently I have a Windows 2008 web server that duals as a database server > our traffic load is fewer than 50 users but we plan for it to increase to 150. > Most Stanbol installations do run under Linux (Ubuntu, Debian) but Windows is also supported. However you might encounter the following issues * There are some reports that unit tests do have problems on Windows to successfully delete files. So depending on your environment you might need to skip the tests when building on windows. * IOExceptions with a message like "Unable to establish loopback connection": [4] has some information about this. Typically this is caused by the (personal) firewall configuration. If you see this exceptions Stanbol may still work fine as this might also be related to the Apache Sling Control Port [10]. * Apache Sling FileInstaller may have an issue on windows (see [5] for more information and a workaround). > What's the recommended RAM requirements and how much hard drive space is > necessary? > > Being a very new developer any other information you can provide with regards > to installation and implementation would be greatly appreciated. We are still > very much in the planning phases of our project. > For production use I would suggest you to use the 0.12.0 release [1] or 0.12.1-SNAPSHOT [2]. Prebuild launchers are available at [6] but I would also strongly suggest you to configure a custom launcher fitting your needs. See [7] for more information on that. RAM and hard drive requirements highly depend on the use case. I will try to give you some figures: * Stanbol requires ~500MByte of RAM. * Holding the english NLP models will require ~160Mbyte. Add NLP models for other languages will add another ~500MByte. * EntityLinking (with the FST linking engine [8]) does use a Entity Cache. If you increase the size of this cache you will also need to provide the required RAM. * If you want to enhance long documents Stanbol will also need some memory to hold enhancement results in memory. So if you plan to enhance 50+ page long PDF files you will need to consider that. * If you have a Server with a lot of CPU cores and you are expecting a lot of concurrent requests you need also to consider that all the enhancement results of concurrent requests need to be kept in memory. Experience shows that for most use cases 2GByte of RAM are sufficient Hard drive space fully depends on the usage scenario. Most of the space is required by the SolrIndexes used for EntityLinking. If you use the EntityhubLinking engine [9] or the FST linking engine [8] with a small Entity Cache linking performance highly depends on the performance of the hard disc. If you use FST linking with a big Entity Cache hard disc performance is not an issue. best Rupert [1] http://stanbol.apache.org/downloads/releases.html [2] http://svn.apache.org/repos/asf/stanbol/branches/release-0.12/ [3] http://svn.apache.org/repos/asf/stanbol/trunk/ [4] https://weblogs.java.net/blog/binod/archive/2006/12/glassfish_and_w.html [5] http://markmail.org/message/3tuv36oyge64ib63 [6] http://dev.iks-project.eu/downloads/stanbol-launchers/ [7] http://stanbol.apache.org/docs/trunk/production-mode/ [8] http://stanbol.staging.apache.org/docs/trunk/components/enhancer/engines/lucenefstlinking [9] http://stanbol.staging.apache.org/docs/trunk/components/enhancer/engines/entityhublinking [10] http://sling.apache.org/documentation/the-sling-engine/the-sling-launchpad.html#control-port -- | Rupert Westenthaler rupert.westentha...@gmail.com | Bodenlehenstraße 11 ++43-699-11108907 | A-5500 Bischofshofen | REDLINK.CO .......................................................................... | http://redlink.co/ ________________________________ This e-mail may contain Sprint proprietary information intended for the sole use of the recipient(s). Any use by others is prohibited. If you are not the intended recipient, please contact the sender and delete all copies of the message.
