Jun 29 2009

A Really Great Conference - O'Reilly Velocity - Want Performance?

Posted by Mike Brunt at 12:42 PM
6 comments
- Categories: Web Servers | .NET | DataBase | CloudComputing | ColdFusion | JRun-J2EE

 

I won't go on again waxing too lyrically about the O'Reilly Velocity conference but I fully recommend all who read this post to go and attend the conference, if it occurs again next year.  What I thought might be of use is to go through some of the products and services we were exposed to over the three days of the conference.  What it seems like to me is that web technologies are evolving and morphing so fast that those of us interested in maintaining and furthering our careers need to keep tabs on what Web 2.0 is pushing us into.  

I will add a pre-note before going into more detail, Adobe were not evident at all at this conference, of course my focus is ColdFusion-JRun-J2EE but obviously there are many other Adobe technologies aimed squarley at the web and I really think Abobe could have made an impact in an environment which featured the current movers and shakers in all things web.

 

  • CouchDB - This is a verbatim quote from the Apache CouchDB documentation..."Unlike SQL databases where data must be carefully decomposed into tables, data in CouchDB is stored in semi-structured documents. CouchDB documents are flexible and each has its own implicit structure, which alleviates the most difficult problems and pitfalls of bi-directionally replicating table schemas and their contained data."  So it seems to me that CouchDB is a de-normalized, distributed database paradigm.  Data is structured for presentation using "views" and the whole construct, including the views and data-structure etc is easily replicable across multiple physical servers-locations.  The views are built using Javascript.  Replication is peer-to-peer and one thing I found interesting is the fact that CouchDB is built using ERLang which is of interest to me as processors move to a more multi-core model.  There is much more detail here.
  • Drizzle - This is another initiative at the database tier and it has to be said that a lot of the Velocity conference addressed databases and how data can be moved closer to the user and that involves safely storing and manipulating data in memory.  I will touch on a caching mechanism later on in this blog piece but this section is focussed on a re-look at MySQL in the light of Cloud Computing and the emerging impact of multi-core processes, here is a verbatim quote from the Drizzle project web site. "Drizzle is building a database optimized for cloud and [network] applications. [Drizzle] is being designed for massive concurrency on modern multi-cpu/core architectures. The code is originally derived from MySQL."  This pretty much explains what Drizzle is and note that this is charactarized as a lightweight adapatation based on MySQL.   One item that did strike me during the Velocity conference is how widely used MySQL is in this high-end web world, high-end in terms of user traffic volume that is.  I did note a comment by John Adams from Twitter that they view the datatabase itself, the infrastructure, hard-drives etc as as nothing more than a storage mechanism and that all transactions basically take place in a memory; I hope I explained that clearly. 
  • Hadoop - This is another Apache based project and applies to clustering data and in that sense delivering redundanacy and scalability to immense applications.  Data is playing an ever larger part in our daily lives, this might sound like an obvious statement as data has always been around, however Web 2.0 interfaces are almost exclusively dynamic and that means database-driven.  In addition, as Ad revenues decline both in traditional and digital media, the mining of data to extract an edge of some kind in targeted marketing becomes more of an imperative.  This is the Hadoop web site with more information for those interested.
  • Memcached - Once again aimed squarely at the data tier, I will use the verbatim from the memcached web site again http://www.danga.com/memcached/ as I could not explain it better; memcached "is a high-performance, distributed memory object caching system, generic in nature, but intended for use in speeding up dynamic web applications by alleviating database load."  Once again we see the trend to put data closer to the user, in this case by using a distributued caching mechanism.  There are what are called "offspring" to extend memcached with replication etc and they can be found here.
  • Puppet - Puppet provides a language for automating system administration tasks across many multiple systems.  In this day of growing Cloud Computing usage it could be easy to forget that there are still mundane servers at the core of things, well of course servers are not mundane.  Puppet is aimed at Unix-like operating systems and once again the puppet blog has more detail than I can possibly cram into this short blog piece, you can find that here...
  • Guerrilla Capacity Planning, Dr Neil J. Gunther - One on the evening sessions at Velocity was called "Ignite" and featured a series of 5 minutes vignettes from various luminaries.  As many of you know it is a constant mantra of mine to pay attaention to all involved in an overall infrastructure and the resulting required performance.  Dr Gunther is an expert on quantified capacity planning as it pertains to all systems whether they be Cloud, Web, Client-Server etc.  He has published several books on the subject and I recommed taking the time to check out his blog.

 

There is so much more to point out and this is already a long blog post.  I will try to return to this and point out other items that we encountered at the conference. 

 

 

 

 

 

 

Comments

Mike Brunt

Mike Brunt wrote on 07/04/09 7:16 AM

Adding my own comment here in response to a Twitter comment from Sean Corfield about how CFML could work with the constructs I mentioned about and the answer is very easily. I think the main area of question is in the area of caching as the CFML engines already have their own caching schemes.
bcp software

bcp software wrote on 02/09/11 1:19 AM

I had the opportunity of attending the conference but fall sick. So that was something that I wasn’t expecting. Anyway I have heard that it was something wonderful and some of my colleges were thrilled out of the experience they got. So I am probably looking to attend the next one and I hope that it goes well the next time around, and I have read every single product and it all sounds wonderful.
solo hd

solo hd wrote on 05/03/11 12:19 AM

This <a href="http://www.solohdheadphone.com">solo hd</a> of headset timbre and its appearance style are very alike -- again, very thin. <a href="http://www.solohdheadphone.com">beats solo hd</a> voice very open seem a bit too thin, <a href="http://www.solohdheadphone.com/fabulous-monster-limited-edition-gold-c-81.html">Fabulous Monster Limited Edition GOLD</a> low frequency partial hard, descend not beautiful but speed feeling good. <a href="http://www.solohdheadphone.com ">discount solo hd</a> hf performance is good, accurate and not mellow, intermediate frequency performance is regular. Overall <a href="http://www.solohdheadphone.com/classic-monster-powered-isolatio-black-s-c-102.html">Classic Monster Powered Isolatio black</a> voice more features, more suitable for listening to electronic music or part of the pop.
formation seo

formation seo wrote on 10/30/12 4:23 PM

Merci pour le partage, c'est très bon à savoir. Cela me paraît fort intéressant. Bonne journée

prix référencement google http://googlepandacestquoi.wordpress.com référencement web
http://www.lancelpascherfr.fr

http://www.lancelpascherfr.fr wrote on 11/05/14 7:46 PM

There are 10 major phases to this when applied to existing applications and this post is the first of 10 successive ones which I genuinely hope will be of help to the community that has been good enough to provide with a good living.

Write your comment



(it will not be displayed)



Leave this field empty: