- Posts tagged Performance
- Explore Performance on posterous
Google Speed initiative
Let's make the web faster
Great initiative with already a good set of tutorials and compilation of tools, and potentially an active community.
Liked "Java run-time performance and availability monitoring"
Run-time performance monitoring and availability monitoring is critical to achieving and maintaining a well-performing system.
An overview of modern SQL-free databases
SQL-free databases are quickly moving forward those days. They aren't "embedded databases" nor "key-value databases" anymore. Modern SQL-free databases are now featureful, while remaining fast, light and reliable. They really can replace SQL databases (even column databases) for a lot of tasks.
Via http://00f.net/2009/an-overview-of-modern-sql-free-databases
Update 1: check-out also the post from Richard Jones on a list of distributed key-value stores with a quick decent analysis on most of them.
Update 2: Leonard Lin over at random($foo) posted yesterday a great post on his own experience with key-value stores. Worth it the read as he ended up writing his own thin consistent hashing layer in front of Tokyo Tyrant, i.e. pretty equivalent to what LightCloud is.
Update 3: Even if done in Python, here i an interesting performance comparison of key/value stores for language model count.
Update 4: And what about another key-value database: Redis, written in ANSI-C for Posix systems?
Update 5: Yahoo is talking about their own key-value store Sherpa.
Update 6: Liked this post on the rational why one will want to go beyond a relational database.
Update 7: But SQL-free DBs do have some serious drawback regarding mostly ad hoc reporting and ad hoc data fixing as alighted in this post on the dark side of NoSQL.
A Comparison of Approaches to Large-Scale Data Analysis - MapReduce vs. DBMS Benchmarks
There is currently considerable enthusiasm around the MapReduce (MR) paradigm for large-scale data analysis [17]. Although the basic control flow of this framework has existed in parallel SQL database management systems (DBMS) for over 20 years, some have called MR a dramatically new computing model [8, 17]. In this paper, we describe and compare both paradigms. Furthermore, we evaluate both kinds of systems in terms of performance and development complexity.
Here is the link of the article (PDF).
SSD Flash drive performances: finding good resources (first pass)
I really started (maybe a little bit late) to look more closely at SSD Flash drive with the recent news around Fusion-io ioDRIVE; see a review at DVNation. Then, I looked at Sun with the new SSD Flash array storage offering; see also the Sun white paper on SSD-based server performances (PDF). An external look at this offering from Spinn3r's Kevin Burton.
My main concern is mostly around Oracle I/O performances, as generally this is one of the main bottleneck of any Java (or not) server based on Oracle (RAC or not in case of sharding for example) once the code has been tuned, proper usage of the DB being implemented , proper caching, etc. On the Oracle 11g performance page you can find this presentation from Oracle Development offering some insights (PDF) on running an Oracle database on Flash drive. I also found this good presentation from Texas Memory Systems on faster Oracle performance with SSD drives (PDF).
Even if I am focusing now on Oracle, the pros & cons are the same for other databases. Sun has some interesting blog posts on SSD performances and database (even if using DB2) here and here. Big DBA Head is mostly focusing on MySQL for most of their SSD drive performances work that you can find here and here. Kevin Burton has also a good write-up on using SSD drives for Spinn3r MySQL databases; and here is his list of posts on SSD drives with mostly a MySQL angle.
Finally there is a good aggregation of SSD related news and articles at Storage Search, including this Flash Performance (PDF) article from EasyCo CTO.
Art of scalability - Scalability principles
Good introduction or refresher for people having to deal with scalability issues or architecting some services that will need to be scalable. See the series at hfadeel.com. Couple this series with his Performance Anti-Patterns post and you'll get a good overview.
Comment: those posts are just scratching the surface of scalability, and is missing for people starting to deal with scalability a list of good resources providing more specific details and information. I guess I could do something here...
