Please read Part 1 for my explanation of what I am trying to do, and why I am testing these out. It is good to note that libredis performed almost identical to pecl-memcache when running 10 concurrent connections (each request puts 5 keys and gets those 5 keys back)

wget http://127.0.0.1/path/to/benchmark

1 connection pecl-memcache 10000 sets/gets
set: 0.6813(s)
get: 0.6937(s)

1 connection php-memcached 10000 sets/gets
set: 0.6988(s)
get: 0.7722(s)

1 connection libredis 10000 sets/gets
set: 0.8227(s)
get: 0.7697(s)

1 connection predis 10000 sets/gets
set: 2.2110(s)
get: 1.0792(s)

1 connection rediska 10000 sets/gets
set: 5.0501(s)
get: 3.3987(s)

—————————————
ab -c 10 -t 10 http://127.0.0.1/path/to/benchmark

10 concurrent connections pecl-memcache – 5 sets/gets per request
Complete requests: 4402
Requests per second: 440.14 [#/sec] (mean)

10 concurrent connections libredis – 5 sets/gets per request
Complete requests: 4754
Requests per second: 475.35 [#/sec] (mean)

10 concurrent connections predis – 5 sets/gets per request
Complete requests: 2179
Requests per second: 216.43 [#/sec] (mean)

10 concurrent connections php-memcached – 5 sets/gets per request
Complete requests: 2200
Requests per second: 204.78 [#/sec] (mean)

10 concurrent connections rediska – 5 sets/gets per request
Complete requests: 1678
Requests per second: 167.76 [#/sec] (mean)
—————————————
nginx/0.8.36 PHP print “hello” (php-fpm latest SVN, PHP 5.3.2)

ab -kc 100 -n 10000 http://127.0.0.1/dummy.php 100 concurrent keepalive connections
Requests per second: 1781.36 [#/sec] (mean)

ab -n 100000 -kc 10000 http://127.0.0.1/dummy.php 10,000 concurrent keepalive connections
Requests per second: 1559.65 [#/sec] (mean)

b -n 100000 -kc 10000 http://127.0.0.1/dummy.php 10,000 concurrent NON-keepalive connections
Requests per second: 1416.88 [#/sec] (mean)

—————————————
ab -kc 100 -n 30000 http://127.0.0.1/404.html

nginx 404.html – 100 concurrent keepalive connections
Requests per second: 15954.15 [#/sec] (mean)

Update: 04/27/2010 – Scroll to bottom for benchmark which includes the almost unknown Libredis library for PHP.

I am new to using key-value memory caches, so today I decided to run some tests. I’ve searched the web quite a bit today (ok.. a LOT) and found some old benchmarks, but I wanted to run some modern benchmarks on the latest versions and see how it worked for my needs.

$expire = 120;

$key = md5($query);

$begin = microtime(true);

for ($i=0;$i<10000;$i++)
{
    $this->_server->set($key.$i,$value,$expire); // different for different libraries, but you get the idea
}

$taken = microtime(true) - $begin;
$taken = number_format($taken,4);
print "set: $taken(s)<br />";

$begin = microtime(true);
for ($i=0;$i<10000;$i++)
{
    $this->_server->get($key.$i); // different for different libraries, but you get the idea
}

$taken = microtime(true) - $begin;
$taken = number_format($taken,4);
print "get: $taken(s)<br />";

Bing Twitter Search Beta

Bing has just announced on their community search blog here that they now have real-time access to the entire public Twitter feed.

Here is an excerpt from their announcement:

We’re glad you asked that. Because today at Web 2.0 we announced that working with those clever birds over at Twitter, we now have access to the entire public Twitter feed and have a beta of Bing Twitter search for you to play with (in the US, for now). Try it out. The Bing and Twitter teams want to know what you think.

Now.. this all sounds great, except for the fact is at the time of writing this blog post, the link to their Twitter search (bing.com/twitter) is 404.

Update – Bing Twitter Search went live (from 404) about 15 minutes after this post was published!

I don’t see anything odd about this day being April Fool’s or anything, so I am assuming they are working on that page as I am writing this article (or that post got published sooner than it should have on Bing’s Search Blog).

Let me know what you think

I thought my first informative post should be a good one, so here goes.

It may not be the biggest list, but they are verified by me as having dofollow comments, a good amount of content, and a decent mozRank, which usually means a decent PageRank.

I decided not to list PageRank, since I much prefer the mozRank floating point numbers, but if someone wants to go through and provide the PageRank info, I will update it.

I’ll also throw in some other sites that provide more lists, no guarantees, but I believe in sharing, so here they are:

http://buckdat.blogspot.com/2009/10/effective-link-building-galore-unique.html

http://niceblogger.com/2009/08/21/dofollow-list/

http://www.seotrafficgenerator.com/website-seo/high-pr-do-follow-blogs-list/2009/09/07/

http://www.clickonf5.org/google/dofollow-blogs-google-reader/4570

If anyone has any more to add to the list, just let me know via comment!

A little background on the testing done so far. We began using and testing the system in May 2009. Originally the service was conceived for Go Web Solutions, a web design and marketing firm. The goal was to provide staff with monitoring tool to deliver website management and marketing services, and to provide clients with reporting that shows how their site is performing at all times. As soon as the tool was implemented, it was a huge hit with clients.

Since May, we have had several dozen client domains in the system and being monitored. In that time we have identified and fixed several bugs, implemented numerous enhancements, and we have been creating a list of future enhancements that our users will find useful. Now that we are in Beta, it’s time to start putting a larger load on the system to test its stability and scalability.

Aaron, the amazing developer and co-creator of EZWM, has put some excellent tools in place that allow us to monitor the monitoring queues in real-time. This allow us to visually watch all types of monitors, across all servers, and see the real-time load on each server. This level of detailed monitoring of the queues will let us easily identify when upgrades are needed to the network and deploy additional resources.

We are looking for the following types of beta testers:

1. Website Owners. The average website owners doesn’t know how their site is doing, they simple watch the traffic (if that). Our goal is to give website owners and simple way to actually monitor uptime, quality, keyword positions and popularity in one place. We need website owners that are looking for better tools to monitor their website and want to take the next step to controlling their success.

2. Power Users and Website Managers. EZ Website Monitoring is a great tool for tracking multiple websites, making it easy for power users to track more than one website and always know how it’s performing. Typically this type of user has several sites they are responsible for, they have one or more hosting providers, and are responsible for getting traffic to those sites and converting that traffic into sales. This group will help us test the multi-domain capability of EZWM and the ease-of-use of monitoring more than one domain.

3. Agencies. Internet Marketing, SEO, Advertising agencies, New Media companies…whatever the name, we need testers like you to really push the system from the multi-domain/multi-client stand point. The idea is to provide you with the ability to manage multiple customer web sites from one interface and easily switch between website monitoring reports. We also want to test the notification and weekly reports, eventually giving you the ability to white-label the reports and notifications as a value-added service to your client.

If you know anyone that would be interested in testing our service, please have them sign up for the Beta and for 6 months of free website monitoring. The Beta test is scheduled to end October 18th, 2009.