Ich habe vor ein paar Wochen meinen Rechner ein wenig aufgerüstet.

Folgende Komponenten sind neu dazu gekommen – quasi ein Aufrüstkit:

• CPU: AMD Phenom II 905e mit 4 x 2,5 GHz

• Mainboard: Gigabyte MA785GT-UD3H, mit 785G Chipsatz, AM3-Sockel und 6 Sata2-Anschlüssen

• Ram: 2x 2GB GEIL PC-10667 CL7 DDR3

Die alten Komponenten werkeln jetzt im Rechner eines Freundes.

Seit ein paar Tagen nun habe ich weiter aufgerüstet:

• OS: Windows 7 Home Premium 64bit

• SSD: Super Talent Ultradrive GX 64Gb

Interessanterweise habe ich alle Komponenten günstiger bekommen, als sie zur Zeit angeboten werden.

Vor allem Das Betriebssystem habe ich – dank der von Micrsosoft veranstalteten Promotion-Aktion – deutlich billiger, nämlich für 39,95€ bekommen.

War nur mit dem Aufrüstkit der Geschwindigkeitsvorteil eher “dezent”, so kann ich zu gefühlten Geschwindigkeit jetzt nur sagen:

WOW!

Ich kann nicht genau festmachen, welche Komponente welchen Stellenwert beim Zuwachs hat. Sei es die SSD, sei es Windows 7, sei es das Aufrüstkit, sei es alles zusammen.

So lässt sich auf jeden Fall arbeiten

Gefühlte Geschwindigkeit:

Programme, die vorher bis hin zu einer Minute gebraucht hatten, um vollständig geladen zu sein, laden nun in 2-3 Sekunden.

Firefox anklicken – nach 2 Sekunden ist er komplett geladen. Vorher brauchte er gerne mal 30 Sekunden dafür.

Man kann nun auch ohne Bedenken 30 Programme gleichzeitig laufen lassen. Und die CPU langweilt sich trotzdem.

Unter XP ging die Prozessorlast gerne mal bei 2 oder mehr Kernen in den Bereich Volllast. Unter Windows 7 dümpelt die CPU meistens auf Minimum herum.

Ein Bootvorgang (reines Windows) dauert bei mir etwa 10-15 Sekunden, vorher waren es etwa 60 Sekunden. Das BIOS benötigt tatsächlich mehr Zeit zum Booten, als das OS mit all seinen Autostart-Programmen.

Auch das Kopieren von Daten zwischen zwei Festplatten ist deutlich schneller geworden, ebenso das packen und entpacken von Daten.

Konnte ich vorher mit maximal 35 MB/s entpacken, geht dies nun mit bis zu 80 MB/s. Auch das verschieben/kopieren von Festplatte x zu Festplatte y wird jetzt nur noch durch die maximale Transferrate der langsameren Festplatte begrenzt.

Windows 7 :

Das neue Betriebsystem gefällt mir recht gut.

Der Umstieg von Xp zu Windows 7 verlief relativ reibungslos.

Die Installation auf der SSD ging flott von Statten. Insgesamt dauerte es etwa 20 Minuten.

Lediglich einige lieb gewonnene Einstellungen gibt es nicht mehr oder nicht mehr in der Form.

Ich habe bspw. noch nicht herausgefunden, wie man den Explorer dazu bekommt, beim einfachen anklicken eines Ordners seine Subfolders zu kaskadieren/auszuklappen. Das ging unter XP immer.

Ein wenig Sorgen bereitet mir, dass ich nicht mehr via ATI Catalyst Suite die Helligkeit regeln kann. Da kann ich munter Werte editieren, es zeigt sich keine Änderung.

Ich bin nun den Weg über die Farbkalibrierung in Windows selbst gegangen – jedoch schaltet  jedes Mal die Helligkeit hoch, wenn ich ein Spiel beendet habe, welches eigene Gamma-Werte nutzt.

Und ein bisschen nervig ist, dass der Windows Explorer sich partout nicht merken will, wie ich meine Ordner-Ansicht wünsche – nämlich überall als Detailansicht. Er will es einfach nicht übernehmen und zeigt mir bei jedem Ordner wieder eine andere Ansicht.

Ansonsten bin ich vor allem mit dem Design zufrieden.

Probleme hatte ich – mal wieder – mit dem Creative Treiber. Zur Abwechslung ließ er sich problemlos installieren. Unter XP hatte das gerne mal zwei Dutzend Anläufe gebraucht. Jedoch musste ich feststellen, dass 90% der Programme und Konfigurationsmöglichkeiten fehlten. Ich konnte nicht einmal irgendwo eine Bassumleitung oder Bassverstärkung eingeben. Es existierte lediglich die Audio-Konsole. Dementsprechend war auch die Tonqualität suboptimal. Der Bass ließ sich gar nicht zur Mitarbeit überreden.

Mit dem X-Fi Support Pack 2.0, einem alternativen Treiber, war dann der gewohnte Funktionsumfang vorhanden.

Jetzt klingt auch wieder alles, wie es soll.

Was ich nicht wusste: XP-Antispy läuft auch unter Windows 7. Damit wurde ich auch einige nervige Komponenten los.

Auch die Benutzerkontensteuerung habe ich direkt deaktiviert. So sinnvoll sie auch sein mag – wenn ich schon beim einfachen entpacken einer Datei Adminsistratorrechte brauche und ständig Nachrichten aufpoppen, dann muss das weg.

Den Bug, den Avira Antivir verursacht, indem es bei jedem Neustart einmal chkdsk durchlaufen lassen will, habe ich glücklicher Weise nicht.

CPU:

Ich nutze hierbei K10Stat, um die CPU zu regeln. Dabei müssen im BIOS und im System Cool & Quiet aktiviert sein, K10Stat baut darauf auf.

Der minimale Wert liegt bei mir bei 0,85 Volt. Wobei das Gigabyte Board leider von sich aus etwas übertaktet. Weiter runter habe ich mich noch nicht getraut, zumal dies auch einigen Aufwand zum Testen benötigt.

Der Maximale Wert liegt bei 1,075 Volt. Jeweils mit 0,075 Volt Abstufungen zu den P-States.

Interessanterweise habe ich keine Aufgabe definiert, wie es unter Vista/Win7 wohl üblich ist.

Ich habe lediglich eine Verknüpfung mit den nötigen Parametern (-ClkCtrl:1 -NoWindow -StayOnTray -LoadPstate) in den Autostart-Ordner gepackt – und es läuft.

SSD:

Als ich das Paket mit der SSD bekam, war ich ein wenig erstaunt. Die SSD ist deutlich kleiner, als ich dachte. Sie ist nicht einmal halb so groß, wie eine 3,5 Zoll Festplatte und wiegt nur wenige Gramm.

Glücklicherweise blieb mir das herum flashen mit verschiedenen Firmware-Versionen erspart. Mein Modell entstammt einer neueren Baureihe und weist bereits die neueste Firmware-Version 1819 auf.

Sonst hätte ich zunächst von der 17er auf die 15er und dann auf die 18er flashen müssen. Das hat bei einigen Leuten nicht so gut geklappt – und führt in jedem Fall zu einem kompletten Datenverlust.

Ich habe die SSD einfach unten auf den Boden gelegt – sie Bedarf ja keiner exakten Ausrichtung. Das sieht zwar ziemlich doof aus, aber aufgrund es unglaublichen Kabel-Wirrwarrs im Gehäuse ist das jetzt ohnehin wurscht.

Hauptsache es funktioniert!

Windows 7 hat automatisch die richtige Blockgröße eingerichtet, die Defragmentierung aus- und TRIM eingeschaltet.

Ob Trim eingeschaltet ist, kann mit einfach mit folgender Zeile in der Eingabeaufforderung herausfinden:       fsutil behavior query DisableDeleteNotify . Wenn eine Null ausgegeben wird, ist Trim seitens des OS aktiviert. Bei einer 1 nicht.

Für diese SSD, wie auch für einige andere, wird gerne der Rat gegeben, mindestens 10% der Speicherkapazität ungenutzt zu lassen. Man kann den Platz also komplett partitionieren und formatieren, sollte aber dann 5-6 GB frei lassen, damit es nicht zu Geschwindigkeitseinbrüchen kommt.

Mit Windows 7, all meinen Programmen und sogar einem Spiel (Battlefield 1942+Desert Combat) sind nicht einmal 20Gb belegt.

Sicherlich wäre eine Intel Postville noch einen Tick schneller gewesen, aber die Lieferzeiten liegen jenseits von Gut und Böse. Frühestens Ende Dezember hätte bspw. Mindfactory eine liefern können. Intel gibt die SSDs offenbar nur Tröpfchenweise ab.

Benchmarks:

Die SSD läuft bei mir auf dem Niveau 217 MB lesend und 160MB/s schreibend. Wie ich mir versichern ließ, ein durchaus guter Wert für diese SSD. Selbst einige 128er Versionen des Ultradrive – welche eigentlich schneller agieren sollen – sind bei einigen langsamer, als meine 64er.

Wie zu erwarten war, brechen die Werte bei kleinen Dateien im 0,5 bis 16Kb-Bereich ein, liegen jedoch auf üblichem Niveau.

Ohnehin – wie einem jeder SSD-Kenner genervt entgegnet – kommt es nicht auf die maximalen Transferraten an, sondern auf die Operationen pro Sekunde an. In Zusammenspiel mit einer Access-Time von etwa 0,2 ms schlägt diese SSD jede andere Festplatte um Längen. Selbst eine Kombination aus zwei kleinen, schnellen Raptor-Festplatten von Western Digital im Raid 0, ist langsamer.

Somit ist eine SSD für den Einsatz als Systemplatte geradezu prädestiniert. Als Datengrab ist sie aber ohnehin zu klein

Andere Benchmarks, wie bspw. 3d Mark 06 oder Super Pi haben von dem neuen System wenig profitiert.

Lediglich der CPU Score ist hierbei deutlich gegenüber meinem alten Dualcore (Athlon BE-2400)  angestiegen.

Dies war aber auch nicht zu erwarten. Vor allem Grafikbenchmarks  und Spiele profitieren wenig von Windows 7 und einer SSD. Hier sind höchstens die Ladezeiten geringer.

Interessant ist jedoch der mehr als doppelt so hohe Durchsatz beim kopieren/verschieben/entpacken – vor allem beim Arbeiten auf eine anderen Festplatte.

Konnte ich vorher mit maximal 35mb/s auf eine andere Festplatte entpacken/kopieren, werde ich jetzt lediglich durch die maximale Performance der langsameren Festplatte ausgebremst – das sind meistens um die 80 MB/s.

Fazit:

Als Fazit bleibt mir nur zu sagen: ich bin mit der gebotenen Leistung sehr zufrieden!

Ich kann nicht genau ausmachen, welche Komponente jetzt welchen genauen Anteil hat, aber das ist mir eigentlich auch egal.

Es arbeitet sich mit Windows 7, aktiviertem AHCI, einer guten SSD und einem nicht allzu schlechten PC-Grundgerüst deutlich angenehmer

Wer also wie ich viel Zeit vor dem PC verbringt, sollte sich durchaus überlegen, ob er nicht doch einmal eine andere Komponente weniger aufrüstet und statt dessen eine SSD einbaut.

Bei Preisen ab 150€ für 64Gb sicherlich kein günstiges Vergnügen.

Fest steht: ich gebe meine SSD nicht mehr her

Und aufrüsten muss ich in absehbarer Zeit nichts mehr

Bisiklet kullanmayı alışkanlık haline getirenler için, bisiklet sadece bir ulaşım aracı değildir şüphesiz. Bisikletler romantiktir…

Özgür olduğumuzu hatırlatırlar. Şöyle güneşli bir günde, kulağımızda yankılanan sevdiğimiz bir müzik ile birlikte, tüm kaygılardan kolaylıkla uzaklaştırabilirler bizi.

Bunun yanında günümüzde kullanılan diğer ulaşım araçları ile karşılaştırdığımızda, bisiklet en temiz ve sürdürülebilir olanıdır. Kendi enerjisini kendi yaratacak kadar mütevazidir kendisi.

Yaptığım girişten anladığınız kadarı ile 2. benchmarkımız bisikletler ile ilgili, Barselona ve Kopenhag’dan geliyor bu sefer.

Bicing Barselona’da kullanılan, belediyenin sağladığı toplu taşıma araçlarından biri. Alternatif bir toplu taşıma aracı diyebiliriz. Amaç bir bisiklet istasyonundan aldığınız bisikleti, gideceğiniz yerdeki, şehrin başka bir noktasında bulunan diğer bir bisiklet istasyonunda bırakmak.

Tüm şehire yayılmış durumda bulunan 400 tane bisiklet istasyonu bulunmakta, toplamda 3000 adet bisiklet kullanılmakta sistemde.

Sistemde bulunan bisikletleri kullanabilmek için, yıllık kayıt yaptırmış olmak gerekiyor. Kayıt yaptırdıktan sonra, evinize gelen RFID kartınız ile sistemi kullanmaya başlıyorsunuz. Her bisiklet istasyonunda, kartı okutabileceğiniz ekranlar bulunmakta. Böylelikle sistem kim olduğunuzu tanımakta, siz de ödünç aldığınız bisikletin sorumluluğunu üstlenmektesiniz.

Sistemin esas amacı bisikletleri kısa ve orta mesafeler için kullanmak. İlk 30 dakika için kullandığınız bisikletler ücretsiz. Diğer her 30 dakika için 50 cent ödemeniz gerekmekte. Bir seferde en fazla kullanabileceğiniz süre ise 2 saat.

Bicing tarzı sistemlerin benzeri Toulose, Lyon ve Paris’de de bulunmakta.

Bunun dışında geçenlerde bir sergide karşılaştığım Kopenhag’da kullanılmakta olan diğer bir bisiklet sistemi de ilgi çekici. Bisikletin içine entegre edilmiş bozuk para sistemi ile, bisikletleri kart gerekmeden kullanabiliyorsunuz. Yine bisiklette bulunan şehir haritası da turistler için çok kullanışlı olsa gerek.

7 tepeli İstanbul’da bisiklet kullanmanın ne denli zor olduğunu biliyorum, ancak kolaylaştırmak için hiç bir şeyin yapılmadığı da unutulmamalı. Bu alanda neler yapılabileceği konusunda hali hazırda bir çok yaratıcı örnek mevcut.

Bu alanda bir şeyler yapılana kadar, bisikleti olan şanslı azınlık için: Bisikletli, huzurlu, özgür ve mutlu bir gün geçirmeniz dileği ile!

The benchmarks were done while Bifferboard was running Linux kernel 2.6.30.5 and Debian.

CPU speed
Calculated BogoMips: 56.32
According to a quite complete BogoMips list table, this is an equivalent of Pentium@133MHz or 486DX4@100MHz.

According to another CPU benchmarks comparison table for Linux, Bifferboard falls into the category of Pentium@100Mhz.

Memory speed
A “dd” write to “/dev/shm” performs with a speed of 6.3 MB/s.
The MBW memory bandwidth benchmark shows the following results:

bifferboard:/tmp# wget
http://de.archive.ubuntu.com/ubuntu/pool/universe/m/mbw/mbw_1.1.1-1_i386.deb
bifferboard:/tmp# dpkg -i mbw_1.1.1-1_i386.deb
bifferboard:/tmp# mbw 4 -n 20|egrep ^AVG
AVG Method: MEMCPY Elapsed: 0.15602 MiB: 4.00000 Copy:
25.637 MiB/s
AVG Method: DUMB Elapsed: 0.06611 MiB: 4.00000 Copy:
60.502 MiB/s
AVG Method: MCBLOCK Elapsed: 0.06619 MiB: 4.00000 Copy:
60.431 MiB/s

For comparison, my Pentium Dual-Core @ 2.50GHz with DDR2 @ 800 MHz (1.2 ns) shows a “dd” copy speed to “/dev/shm” of 271 MB/s, while the MBW test shows a maximum average speed of 7670.954 MiB/s. Bifferboard is an embedded device after all…

Available memory for applications
My base Debian installation with “udevd”, “dhclient3″, “rsyslogd”, “sshd”, “getty” and one tty session running shows 24908 kbytes free memory. You surely cannot put CNN.com on this little machine, but compared to the PIC16F877A which has 368 bytes (yes, bytes) total RAM memory, Bifferboard is a monster.

Disk system
All tests are done on an Ext3 file-system and a very fast USB Flash 8GB A-Data Xupreme 200X.
A “dd” copy to a file completes with a write speed of 6.1 MB/s.
The Bonnie++ benchmark test shows the following results:

Version 1.03d       ------Sequential Output------ --Sequential Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
bifferboard.lo 300M   822  96  5524  61  4305  42   855  99 16576  67 143.2  12
                    ------Sequential Create------ --------Random Create--------
                    -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP /sec %CP
                 16  1274  94 14830 100  1965  85  1235  90 22519 100 2015  87

bifferboard.localdomain,300M,822,96,5524,61,4305,42,855,99,16576,67,143.2,12,16,1274,94,14830,100,1965,85,1235,90,22519,100,2015,87

Therefore, the sequential write speed is about 5.5 MB/s, while the sequential read speed is about 16.5 MB/s.

It’s worth mentioning that while the write tests were running, there was a very high CPU System load (not CPU I/O waiting) which indicates that the write throughput of Bifferboard may be a bit better if the file-system is not a journaling one. However, the tests for “Memory speed” above show that writing to “/dev/shm” (a memory-based file-system) completes with a rate of 6.3 MB/s. Therefore, this is probably the limit with this configuration.

Network
Both Netperf and Wget show a throughput of 6.5 MB/s.
The packets-per-second tests complete at a rate of 8000 packets/second.
Modern systems can handle several hundred thousand packets-per-second without an issue. However, the measured network performance of Bifferboard is more than enough for trivial network communication with the device. During the network benchmark tests, there was very high CPU System usage, but that was expected.

Conclusion
Bifferboard performs pretty well for its price. It’s my personal choice over the 8-bit 16F877A and the other 16-bit Microchip / ARM microcontrollers, when a project does not require very fast I/O.

This article is about measuring network throughput and performance using a Linux box, in a very approximate way.

There are many network performance benchmarks and stress-tests which measure the maximum bandwidth transfer rate of a device – that is how many bytes-per-second or bits-per-second a device can handle. Good examples for such benchmark tests are NetPerf, Iperf, or even multiple Wget downloads which run simultaneously and save the downloaded files to /dev/null, so that we eliminate the hard disk throughput of the local machine.

There is however another very important metric for the throughput of a network device – its packets-per-second (pps) maximum rate.

Today I struggled a lot to find out a tool which measures the packets-per-second throughput of a network device and couldn’t find a suitable one. Therefore, I tried two different methods both using the ICMP echo protocol, in order to approximately measure the packets-per-second metric of a remote network device which is directly attached to my network.

Method #1: The old-school “ping“. Execute the following as root:

ping -q -s 1 -f 192.168.100.101

This floods the destination “192.168.100.101″ with the smallest possible PING packets. Here is a sample output:

--- 192.168.100.101 ping statistics ---
20551 packets transmitted, 20551 received, 0% packet loss, time 6732ms

You can calculate the packets-per-second rate by dividing the count of the packets to the elapsed time:

20551 packets / 6.732 secs ~= 3052 packets-per-second

Note that this is the count of the replies at 0% packet loss. Which means that the same count requests were sent too. So the final result is:

3052 x 2 = 6104 packets-per-second

Note that if you run multiple “ping” commands simultaneously, you will get more accurate results. You need to sum the average values from every “ping” command, in order to calculate the overall packets-per-second rate. For example, when I performed the measurement against the very same host but using two or three simultaneous “ping” commands, the average packets-per-second value was settled to 8000.

Method #2: The much faster and extended “hping3“. Execute the following:

taurus:~# time hping3 192.168.100.101 -q -i u20 --icmp|tail -n10

This will bombard the host “192.168.100.101″ with ICMP echo requests. After a few seconds, you can interrupt the ping by pressing CTRL+C. Here is the output:

--- 192.168.100.101 hping statistic ---
67932 packets transmitted, 10818 packets received, 85% packet loss
round-trip min/avg/max = 0.5/7.8/15.6 ms

real 0m2.635s
user 0m0.368s
sys 0m1.160s

If the packet loss is 0%, decrease “-i u20″ to something smaller like “-i u15″ or less. Make sure to not decrease it too much, or you may temporarily lose connectivity to the device (because of switch problems?).

The calculations for the packets-per-second value are the same as for “ping” above. You need to divide the received packets to the time:

10818 packets / 2.635 secs ~= 4105 packets-per-second

Because the calculated packets-per-second represent the rate of successful replies, we need to multiply it by two, in order to get the actual packets-per-second, as the count of the replies is the same as the count of the successfully received requests. The final result is:

4105 x 2 = 8210 packets-per-second

The calculated value by “hping3” is the same as the one by “ping”. We are either correct, or both methods failed…

Some notes which apply for both methods:

• You should run multiple instances of the ping commands, either from the same or from different machines, in order to be able to properly saturate the connection of the machine which is being tested.
• The machines from which you test have a hardware rate limit too. If the host being tested has a similar rate limit, you surely need to run the ping commands from more than a single machine.
• You should run more and more simultaneous ping commands until you start to receive similar results for the calculated packets-per-second rate. This indicates a saturation of the network bandwidth, usually at the remote host which is being tested, but may also indicate a saturation somewhere in the route between your tester machines and the tested host…
• The ethernet switches also have a hardware limit of the bandwidth and the packets-per-second. This may influence the overall results.
• If the tested device has any firewall rules, you should temporarily remove them, because they may slow down the packets processing.
• If the tested device is a Linux box, you should temporarily remove the ICMP rate limits by executing “sysctl net.ipv4.icmp_ratelimit=0″ and “sysctl net.ipv4.icmp_ratemask=0″.
• If the tested device is a router, a better way to test its packets-per-second maximum rate is to flood one of its interfaces and count the output rate at the other one, assuming that you are actually flooding a host which is behind the router according to its routing tables.

Thanks to zImage for his usual readiness to help people out by giving good tips and ideas.

On a recent trip through DFW Airport an advertisement caught my eye.  Not for a product, but for DFW’s customer satisfaction metrics.  One way to improve service is to survey customers.  DFW took it a step further by making those results public!

Their metrics were things like: terminal cleanliness, ease of finding your way, courtesy and helpfulness of security and check-in staff, courtesy of airline check-in staff, and restaurant/eating facilities.  And their scores (on a 5-point Likert scale) ranged from 4.0 for washroom cleanliness to 4.3 for ease of finding your way.

DFW further segments results by terminals.  I like this idea because it fosters healthy competition among terminal staffs.  There’s little harm in sharing the details because, as consumers of the service, we already know which terminals are ‘better’.  As an experiment during a long layover, check out the difference between terminals in an airport.  Atlanta’s Hartsfield airport is a good test case.  The international terminal is spacious and clean while some of the others are narrow, dark, and dirty.

DFW misses two opportunities.

1. DFW should post their results relative to an industry benchmark.  Management at DFW should determine which airports (globally and locally) set the standards and match their results to those of DFW.  For good results, that would help DFW showcase their performance.
2. DFW should use the advertisements to highlight planned improvements.

What DFW has done, and what so many companies are afraid to do, is share customer satisfaction survey information.  This level of transparency creates a bond between customer and provider and enables a dialogue.

The pivot point is that service transparency forces service improvements.  Without sharing results, management can hide in blissful ignorance.  By sharing results, by opening a forum so that customers can comment, critique, and yes, complain about service, companies take a visible step in acknowledging what customers want and hold themselves accountable to improvements.  Which kind of company would you rather do business with, one that purports to deliver high quality service, or one which  shares its results transparently?

WOD 9226- Tuesday

Safe Travels to Arjun. See you in few weeks.

“Grace”
30 Clean and jerks for time 135 lbs/95lbs
Post time and load to comments

Grace (wmv/mov) CrossFit
Finding Balance…or Not? CrossFit NorCal

Che cos’è?

Uno strumento utile per valutare il rischio tipico del mercato. In finanza il termine inglese “Benchmark” è utilizzato per definire un indice rappresentativo di un mercato o di un suo comparto. Un Benchmark quindi, non è un’attività finanziaria o un contratto. A cosa serve il Benchmark? La sua funzione originaria è quella di informare gli investitori, gli intermediari e le autorità di controllo sull’andamento delle quotazioni di un insieme di titoli, che riveste interesse per la totalità dei partecipanti al mercato di riferimento. Il gestore è obbligato per legge a comunicare ogni sei mesi ai propri sottoscrittori qual’è stato il comportamento del fondo rispetto a quello di un altro investimento che rifletta l’andamento generale del mercato.

Il benchmark, in pratica, è l’avversario contro cui il fondo misura la sua forza.

Sulla base della tipologia dei titoli inclusi, i Benchmark possono essere

classificati in obbligazionari, azionari e misti.

Inoltre, i Benchmark possono essere: nazionali o internazionali.

Ogni benchmark dovrebbe essere caratterizzato da quattro elementi fondamentali:

• Trasparenza: gli indici devono essere calcolati con regole chiare e replicabili dall’investitore
• Rappresentatività: gli indici devono essere rappresentativi delle politiche di gestione del portafoglio;
• Replicabilità: gli indici dovrebbero essere completamente replicabili con attività acquistabili direttamente sul mercato;
• Hedge ability: per una copertura tempestiva dei portafogli e l’abbassamento dei costi di transazione

Ti è piaciuto questo articolo? Iscriviti gratis alla nostra newsletter!

Kali ini saya membandingkan enam PHP Framework berbasis MVC (Model View Controller) yang populer. Keenam framework itu adalah CakePHP, CodeIgniter, Kohana v2 dan v3, Yii, dan “new kid on the block” DooPHP. Semua menggunakan versi terbaru saat ini (2009-11-22) kecuali Yii. Berikut informasi detail mengenai versi yang di-test:
- Cake PHP 1.2.5
- Code Igniter 1.7.2
- Kohana 2.3.4
- Kohana 3.0.1.2
- Yii 1.0.10
- Doo 1.2

Mesin yang digunakan
Acer Aspire 4520
AMD Turion 64 X2 TL-58 1.9 GHz
RAM 3 GiB

Software yang digunakan
- Windows 7 Ultimate
- Apache 2.2.12
- PHP 5.3

Hampir semua framework konfigurasi-nya tidak diubah sama sekali (kondisi default). Kecuali Cake PHP yang sudah dimatikan konfigurasi session, cache, database, log, dan debug.

Metode test yang digunakan mencetak tulisan “Hello World!” lalu langsung exit;.

Hasil yang didapat adalah nilai rata-rata dalam satuan rps (request per second). Berikut dari yang terlambat hingga yang tercepat:

- CakePHP ==> 19.72 rps
- CodeIgniter ==> 71.23 rps
- Yii ==> 73.68 rps
- Kohana v2 ==> 76.31 rps
- Kohana v3 ==> 103.39 rps
- Doo ==> 200.32 rps

Terlihat Doo buatan negri tetangga adalah yang tercepat dalam test “Hello World!” ini.

Tapi dalam memilih suatu framework, bukan mana yang tercepat. Tapi yang paling nyaman anda gunakan sehingga mempersingkat waktu developing dan meningkatkan produktivitas anda.

Berikutnya akan ada tambahan dari PHP Framework berbasis MVC lainnya seperti Symfony, Zend Framework, dan APF, Solar Framework. Dan agar lebih fair, semua session, cache, database, dll akan di-non aktifkan. “Hello World!” juga akan dicetak pada view, bukan pada controller lalu exit.

Nb: Konfigurasi module dan extension yang sedang berjalan pada Apache dan PHP tidak diikutsertakan dalam tulisan ini.

Eigentlich gab es ja jedes Jahr mindestens einen Gig in der Kantine in Augsburg. Mir ist gerade aufgefallen, dass ich aber in der Neuen Kantine noch nicht gespielt habe. Das ändert sich am Freitag, wenn Kollege Krimeforce UK (aka Asam the teacher) einläd oben im Schwimmbad seinen Geburtstag zu zelebrieren. Neben dem Geburtstagskind gibt es Unterstützung von Benchmark und meiner Wenigkeit. Musikalisch wirds wohl irgendwo zwischen WE LOVE BASS, Triple Trouble, WTF!?? und FORZAelectroniCA liegen. Ich bin gespannt…
Werde mich also am Freitag zum zweiten Mal diesen Monat in den Zug setzen und nach Augsburg fahren und wohl im Gegensatz zur letzten WE LOVE BASS auch mehr/länger und zu besseren Zeiten auflegen.

Wer mal sehen will was ich momentan noch laufen habe, einfach http://myspace.com/aektschndisco checken. Dicke Bilder auf jeden Fall.

Most of the small businesses I talk to or have worked with, have a common concern for lack of available industry trends. We depend on either inside contacts, published competitor collateral or just plain old snooping around, in order to find out the current industry trends and offerings. As a result , small business leadership is plagued by lack of actionable industry benchmarks . ..how to understand where exactly their performance stand in terms of either the brand position or established benchmarks.

Google Analytics does a pretty good job with it’s benchmarking tool. It lets the businesses assess how their website compares against some of the industry verticals. I find this tool to be one of GA’s biggest strength.More so, since, there is so little actionable benchmark data available for small accounts( read business).

There’s one catch though. You have to have your data sharing with Google & other benchmarking tools enabled. Sometimes that can be the deal breaker. However, I’ve always been in favor of data sharing when it comes to Google ,since they do this anonymously . All identifiable data from your website is removed before it’s merged with other sites in comparable industry. There is no way you identify any of yours or your competitor’s data from the aggregated result . There is a bonus factor too. With your data sharing enabled, you also gain access to Google’s Conversion Optimizer that lets you adjust your bids to maximize conversions at the minimum price in Adwords .

The initial data presented takes into account all businesses of similar size. But you might want to drill down to your specific trade. GA actually lets you do that. This way you can ensure that the data is more pertinent to your particular business. In this example, I wanted to look at not only travel but even further ; Adventure travel players.

6 top level metrics data over the given timeline is displayed against the benchmark. It’s important to note that all figures are time dependent. Daily variations are possible due to parallel marketing campaigns and other promotional efforts. But comparing similar industry rules out any skew from the seasonality of the business since the seasonal trends of these companies are mostly alike.

Visits is the most basic top level measurement for any industry follower. The fact that in this example webpage visits are over 25% down from the industry benchmark, right away tells them to get their acts together and start working on improving the web site, spending more on paid search ,improve the quality of the links , landing pages and keywords , get more affiliate partners and network better on social media. Unless, this is a new brand, in which case, a benchmark has been established for them to work on. More on that in a min.

New visits= prospects= incremental sales. This is therefore such a key metric ,especially for new products or brands that just launched. The fact that in this example the new visits are about 5% over benchmark, almost tells me that it’s a new brand working to attract prospects and is generating a lot of interest. Over time, however, they can expect this figure to travel south and as long as it stays around 72% (which is the established benchmark)there should be no concern. Expectations set..Bingo!

Bounce rate – Anyone who has worked with me is aware of my obsession with this metric. I don’t know if I am obsessed or despise it, but for sure I can’t live without it. The fact that I’ve worked so hard to get this visitor to come witness my great offering and that he/she is leaving my site right away , without even giving me a second chance, kills me. If it was possible I would recommend the whole world wide web community to solely work towards bringing all bounce rate down to zero, and ban the word bounce rate. But hey.. that’s the extremist me. Sobering down..the bounce rate should never climb above 30%.. end of discussion. Even the benchmark bounce rate( 60%) in this example is super high. But it is what it is. Let’s accept it & move on.

Average time on site – I would definitely like my visitor to stay on. But there are times when we have them land on a page where all information is provided and they choose to just not look around. I am not sure I like that but that definitely brings down the average time spent on my site. That’s why we need benchmarking. If the visitor is spending around 5 mins on competitor’s site , am I not providing the quality experience for him/her to spend that much time on mine? What else can I do in terms of value proposition or in terms of uniqueness to improve that? On the other hand, if I am above the Industry benchmark.. oh well.. what can I say.. I ‘ve stopped bragging these days.

Pageviews & Pages /Visit – Benchmarking this can be counterproductive. That’s just me. Maybe it’s due to the nature of the business I handle. I am currently establishing a brand whose web site have a lot of information on single pages. Also, the page views depends on your product mix .Similar information can be served in different combination and layouts. But you might have a different perspective on this. Why don’t you share?

There is something else about benchmarking in GA that stands out for me- Able to visualize my brand positioning. Let’s look at these two examples.

The first one exceeds the benchmarking for every possible metric.From a top level, this is a leader to me. The alfha male!

Compared to that.. here’s another brand in the same category tracked over the same period of time. Does it tell you anything?. Look carefully. Compare it’s results to the industry benchmark. To me, it’s either a weak and a struggling brand or a new brand trying to make it’s mark.More likely, the later.. looking at the new visits.

What did I miss? How are you guys benchmarking your data? What are your sources? How do you validate? Please share with us.
Till then,
Happy Benchmarking

Bibi

En la entrada anterior he descrito más o menos las diferencias iniciales entre VMware y Parallels en sus últimas versiones. He estado usando VMware 3.0 con Windows Vista 32bits (2 Cores asignados, 3 GB de RAM). Esa VM ha fue “actualizada” desde VMware 2.06. Comencé a oír “maravillas” del nuevo Parallels 5.0, que salió pocos días después que VMware, de modo que decidí “mirarlo”. El resultado fue bastante satisfactorio, digamos que me sorprendieron varias cosas, ciertas tareas son muy rápidas, la interfaz está bastante más cuidada que en VMware, y en general me pareció un producto muchísimo más completo que VMware. Pero yo ya he comprado el VMware 3.0 y mi licencia de Parallels es 4.0 y el trial de la versión 5 acaba en dos días.

Cuando probé el nuevo Parallels (el cual estuve usando estas últimas dos semanas), lo hice migrando mi VM de VMware a Parallels (el Windows Vista 32 bits), porque allí tenía todo mi entorno Visual Studio 2008, herramientas de SQL Server, etcétera. Esta última semana me llegó el Windows 7 Home Premium (OEM), así que decidí instalarlo de “cero” en Parallels. Lo hice, instalé mi Visual Studio, Service Pack, configuré todo y dejé la máquina limpia y lista para trabajar. Probé de todo, Crystal, Coherence, Full Screen, Screen, Mac Look, etcétera. Todo lo que se os pueda ocurrir que ofrece Parallels (que es mucho más que lo que ofrece VMware). El nivel de conformidad y satisfacción fue bajando conforme pasaron los días. ¿Por qué? Porque si bien pienso que Parallels ofrece más opciones y cosas para tocar, no todas funcionan bien o acaban siendo 100% funcionales. Intentaré explicar brevemente a qué me refiero a lo largo de la entrada, cuando vaya recordando las cosas.

Dada esta pequeña suma de micro-insatisfacciones, decidí que mi VMware no estaba muerto y me dije, ¿por qué no instalar también el Windows 7 en VMware y repetir toda la configuración…?

Y eso hice.

El motivo es que ahora tengo dos VMs “iguales”, instaladas del mismo modo, con todo prácticamente igual y puedo comprar el rendimiento y sobre todo las prestaciones en el día a día con un poco más de precisión.

Hardware y Software
El ordenador en el cual estoy haciendo todo es mi ordenador personal y de trabajo. Tiene OS X Snow Leopard 1.6.2.

Model Name: Mac Pro
Model Identifier: MacPro3,1
Processor Name: Quad-Core Intel Xeon
Processor Speed: 2.8 GHz
Number Of Processors: 2
Total Number Of Cores: 8
L2 Cache (per processor): 12 MB
Memory: 10 GB
Bus Speed: 1.6 GHz

Dos discos duros internos (el de 1TB es Time Machine solamente)

Hitachi HDS721075KLA360:
Capacity: 750.16 GB (750,156,374,016 bytes)
Model: Hitachi HDS721075KLA360

ST31000528AS:
Capacity: 1 TB (1,000,204,886,016 bytes)
Model: ST31000528AS

Ambos son de 7200RPM.

La tarjeta de Vídeo es: NVIDIA GeForce 8800 GT 512 MB.

Y todo está conectado a estas dos pantallas:

BenQ FP92Wa:
Resolution: 1440 x 900 @ 75 Hz
Pixel Depth: 32-Bit Color (ARGB8888)

HP L1740:
Resolution: 1280 x 1024 @ 75 Hz
Pixel Depth: 32-Bit Color (ARGB8888)

Instalación y Preparativo
El proceso que utilicé es completamente extraoficial y no está basado en ningún estricto control de igualdad. El “host” (OS X) tenía varias aplicaciones abiertas, entre ellas: Adium, Mailplane, iTunes, 1Password, varios Menú Items, el Activity Monitor, etcétera. Sin embargo, no quise cerrar nada porque justamente a mi me interesan los benchmarks “reales” y no los que son pruebas de laboratorio. Yo quiero saber como anda mi VM en el día a día.

La instalación fue la misma para ambos. Inserté el DVD de Windows 7 Home Premium. File -> New VM. En ningún caso usé la instalación “Easy”, sino que fui en el modo manual e intenté poner los mismos parámetros para ambas VMs. 3GB RAM, 2 Processors, 60GB Disco, autoexpandible.
Las aplicaciones no están compartidas, ni las carpetas ni nada, salvo la carpeta de “Downloads” de OS X, que está compartida automáticamente y se puede acceder desde las VMs.

Parallels tiene Adaptive Hypervisor, AutoCompress Disks, Tune Windows for Speed, Optimize Performance for Virtual Machine y Power Consumption Better Performance.
VMware no tiene nada de esas opciones, pero tiene Preferred virtualization Engine Automatic y Hard disk buffering automatic. El disco de VMware está dividido en ficheros de 2GB (Split into 2 GB Files), opción que vino por defecto.

La red es Bridged en ambos. Y ambos tienen la misma configuración de red, ip fija, misma puerta de enlace, todo igual.

Cuando acabó el Windows 7 (mas o menos tardaron lo mismo, la verdad es que no me importa el tiempo que tardan en instalar porque no es algo que haga todos los días) ejecuté Windows Update y vinieron unos 40 MB de parches. Tras el reinicio y la configuración de red (Ip fija, cambio de Workgroup, etc), monté la imagen del Visual Studio 2008, lo instalé como viene (sin cambiar opciones). A continuación, el Service Pack de Visual Studio (SP1).
Luego instalé Total Commander (no puedo vivir sin él) y SQL Express 2008 w/advanced services. Luego Service Pack 1 de SQL 2008. Todo instalado, restauré dos bases de datos diferentes (que uso para desarrollo). Configuré el mismo wallpaper y los mismos colores, Aero naturalmente está activo y listo.

Luego me puse a hacer unas pruebas “informales” y a continuación expongo los resultados. En morado los “mejores tiempos” (siempre menos es mejor). Luego hice una segunda pasada y supongo que por motivos de caché y actividad del ordenador Host, algunos resultados variaron. Pero salvo algún caso aislado, los números fueron más o menos similares:

Los valores que no están en el segundo run han sido literalmente iguales a la primera pasada.

Conclusión
A simple vista todo parece indicar que Parallels es, en definitiva, más rápido en la “mayoría” de las cosas que VMware, excepto en abrir la aplicación; no sé por qué Parallels tarda tanto comparado con VMware.

Personalmente lo siento un poco más ágil (lo sentí desde el minuto uno y los números me están mostrando que no estaba tan equivocado). Parallels tiene algunas cosas que no están 100% pulidas. He tenido algún problema con el portapapeles que no funcionaba bien (nada que cerrando y abriendo todo de nuevo no se solucione) pero detalles de ese tipo.

Parallels parece usar un poco más el disco duro, sobre todo en operaciones como suspender una VM. En el gráfico no se ve, pero el Parallels, luego de suspender una VM y cerrar la aplicación, se quedó más de dos minutos escribiendo en el disco duro (según el Activity Monitor, unos 20MB/Seg). Este tiempo uno “no lo ve”, pero en un ordenador como un portátil o con un disco duro más lento esto puede ser realmente muy molesto.

VMware por el contrario sólo escribió a unos 5MB/seg durante unos 20 segundos luego de la misma operación.

¿CON CUAL ME QUEDO?
La primera impresión es saltar e ir a por Parallels; pero no todo es oro tampoco aquí. Hay que tener en cuenta que ni el modo Coherence ó Crystal en Parallels, ni el modo Unity en VMware son cómodos y ágiles para trabajar con Visual Studio. Al hacer Debug, y movernos con paletas, se pierde mucho la gracia de todo el trabajo, se hace más difícil encontrar bien las ventanas y bugs (en ambos) impiden que a veces se pueda encontrar la ventana que se busca, dado que un click en el Dock no siempre restaura como es debido. A eso hay que sumarle mucho flickering.

La solución es usar modo “Pantalla” o “Pantalla Completa”, que por suerte ahora, funciona con dos pantallas, pero el VMware tiene un Bug espléndido que hace que no se pueda trabajar con dos pantallas en modo Pantalla Completa por más de un rato porque se vuelve loco. No le ocurre a todo el mundo, pero ocurre y está en los Foros.

No me pasó lo mismo con Parallels (aunque yo lo uso con una sola pantalla, he probado usando todo) de momento.

Seguiré usando ambos por unos tres días más (que es el tiempo que tiene el Windows 7 de mi VMware para activarse, dado que estoy usando la misma licencia en ambos). Si me quedo en VMware, reactivaré mi Windows allí y no podré trabajar con Parallels (es un coñazo porque hay que llamar por teléfono a Microsoft).

Si el Parallels es tan rápido como parece, entonces quizás me cambie una temporada; lo que es seguro es que seguiré contándoles como me va con ambos. Seguro que en unos meses veremos actualizaciones y allí será donde podremos ver en qué han estado pensando ambos equipos. A ver si oyen a los usuarios (Algo que Parallels tiene fama de no haber hecho en sus comienzos) o si hacen lo que les da la gana.

Para trabajar con una aplicación más simple o quizás algún juego (no he probado nada de eso), habrá que ver otras pruebas y esperar un poco a ver que dice “la gente”. Yo sólo uso el Windows para Visual Studio y si no tuviese que hacerlo, es porque posiblemente me quedé sin trabajo

Se echan en falta algunas opciones en VMware… quizás peca de ser demasiado simple.

En todo caso, tengo tres días para decidirme, seguiré leyendo en Twitter y por ahí, a ver si aparece algún factor decisivo ¡hasta la próxima!

21-15-9 Reps of:
95# Barbell Thrusters
Pull ups

My Time = 9:00
Last Time = 22:50

Need to get into the 3’s!

xceeding is launching a Benchmarking programme for SME’s in January 2010. See if your industry is covered [www.xceeding.com/benchmarking] and what you need to do to get a free (time limited) study, based on your organsiation!

We have seen another flurry of research releases, to help guide the digital community on what online success looks like.

 The latest was Nielsen’s 3 year study, which thudded on our desks late last week.

Highlights included:

“One third of consumers (33%) exposed to an online ad are able to recall it (when prompted), intention to purchase increased by 4.9% following exposure to an online advertising campaign, with brand sentiment increasing by 5.3%. Display advertising also correlated with a rise in awareness, with top-of-mind awareness jumping 3.1%. The likelihood of a consumer recommending a brand following exposure to an online advertising campaign also increased by 4.4%”

On first impression, this paper only seemed to build on already substantial evidence highlighting online display is a powerful brand ‘lever’ rather than simply a response channel.

However, it was surprising to see mainly traditional brand measures were being applied and reported in the research paper, the further I read. No sign of dwell ratio/time (sorry eyeblaster – it only tells half the story!) and thank goodness, no stray CTR graphs!

This then got me thinking…

Many recent papers and campaign evaluations measuring brand get themselves in a knot when using analytics and online ‘buzz’ data. Does this still mean many feel compelled to use it all, just because they have it (research publishers and campaign managers included)?

The trick, is both clearly articulating what your campaign objectives are and focusing on the right methods to test/measure success will be, up front.

Sounds simple, but when you have many tools to measure, selecting the right one isn’t easy. particularly, if you have to feed the results into a predefined global online performance spreadsheet, which nine times out of ten only include response fields and certainly not a focus group sentiment section!

However, like the Nielsen research, the best methods tend to be a careful blend of traditional brand (pre and post sentiment/NPS online polls/focus groups) and the right online analytics (engagement).

That way you deliver a clearer and more robust campaign evaluation with less tools. Far better than wasting time with more tools that simply present a cloudy, half story.

I suppose it’s about using less, but seeing more?

Posted via email from digispeak’s posterous

Cambridgen yliopiston tutkijat ovat kehittäneet video-, suunta-, kiihtyvyys-, magneettikenttä- ja GPS-dataan perustuvan paikannusjärjestelmän. Viimekädessä tarkka paikannus järjestelmässä tehdään sovittamalla videokuva 3D-malliin. Menetlmän virhe vaikutti raportoidun datan perusteella olevan useimmiten alle metrin ja bonuksena järjestelmä pystyi automaattisesti palautumaan virheellisestä paikasta uuden datan perusteella, joten virheen kumuloituminen ei vaikuta suurelta riskiltä. Tutkijat myös ajoivat järjestelmää reaaliaikaisena VAIO U-71:lla (1,1 GhZ Pentium III) noin 17 Hz paikkatiedon näytteenottotaajuudella.

http://mi.eng.cam.ac.uk/~gr281/outdoortracking.html

If you don’t already have Ubuntu Tweak I recommend it. From there you can enable the development repos for Firefox development versions. Under Applications>Third-Party Sources. Otherwise you can do it strait from launchpad.

If you run the update manager Firefox 3.5.6pre will be installed, and Under Synaptic Firefox 3.6b3 and 3.7a1 will be available to install. I only tried installing 3.6. It wont overwrite your current install of FF, and I noticed I could even run two versions at the same time. In the applications menu “Firefox” will be replaced with Shiretoko (the code name for the 3.5 release) and Namoroka (the code name for 3.6) will now show up.

There are other more complete browser benchmarks out there, but I just wanted a quick rundown of how much of a performance improvement is coming with 3.6.

Browser Benchmarks

Ubuntu 9.10 – Nov 12th 2009 – Sunspider:

Epiphany 2.28 [w/WebKit] (64bit) – 580.2ms
Chromium 4.0.245.0 (64bit) – 583.2ms
Chromium 4.0.245.0 (on slower 32 bit system) – 616.2ms
Epiphany 2.28 [w/WebKit] (on slower 32 bit system) -  954.8ms
Firefox 3.6b3pre (on slower 32 bit system) – 1385.4ms
FF 3.5.5 (on slower 32 bit system) – 1642.8ms
Firefox 3.5.6pre (on slower 32 bit system) – 1677.0ms
Firefox 3.6b3pre (64bit) – 2084.2ms
Firefox 3.5.6pre (64bit) – 2755.8ms
Opera 10.01 (64bit) – 3701.4ms
Opera 10.01 (on slower 32 bit system) – 6089.0ms (yikes!)

For whatever reason FF does poorly on the sunspider test on my 64 bit machine. I switched this machine to 64 bit when I did a fresh install of Karmic. I have run previous tests on in it as 32bit and FF did better then. see old post. Surprisingly Epiphany won the race overall.

That inspired me to play with Epiphany just a bit and found out it can be customized to be a pretty slick browser! I found it has an inspector for web developers, automatically opens source (ctrl+U) in gedit, and in many way behaves very much like chrome. It opens fast, plays flash well, and the chrome can be stripped down so it wastes less screen space. For example if you turn off the status bar link URL’s show up in the bottom left corner just like in Chrome, a nice little touch. The unified address/search bar also serves as a progress bar like Safari used to/does(?). here’s a screenshot:

Epiphany and web inspector looking good

Une excellente façon de progresser, c’est bien sur l’échange. Echanges entre acheteurs, entre “professionnels”;  Le temps de se poser un peu, de se regarder fonctionner, de constater aussi comment raisonnent ses “pairs”.

Et cette opportunité peut avoir lieu de bien des manières: Par exemple, en participant avec plusieurs autres responsable achat à un “focus group”, c ‘est à dire une réunion informelle, bien souvent organisé sous l’égide d’un fournisseur;

Un cas très concret à venir: François Odolant gérante de AFMBooster, recherche des acheteurs dans le secteur voyages et services généraux, pour un focus group qui aura lieu en fin d’après midi à Paris, le 10 Décembre prochain.

Elle vous invite à participer à une réunion pour évaluer un nouveau concept de véhicule VIP

(qui aura lieu en présence du commercialisateur représenté par sa Direction Générale et sa Direction Marketing Communication).
Cette réunion aura lieu le jeudi 10 décembre 2009 de 17h à 19h30.

Elle se tiendra dans Paris intra muros (lieu à confirmer).

Si vous êtes intéressé, veuillez me faire parvenir contacter par mail : francoise.odolant@neuf.fr

Faban 1.0 has just been released. This is a major milestone for this open-source workload creation and test framework. Faban is widely used by many performance geeks to performance test various types of server applications. Amongst open source tools, Faban is unique in that it not only provides a framework to create and run performance/load tests, but also has robust functionality to run monitoring tools. It collects all sorts of configuration information as well to truly help performance engineers keep track of configuration and tuning settings.

Here are some major new features in 1.0 which I think will make Faban a very attractive proposition (when compared to the likes of some very expensive proprietary tools).

Automated Form Generation

If you have used Faban before to create workloads, this is the feature you probably have been craving. But let me explain for those who have never used Faban.

Unlike most other tools, Faban does not provide a standard GUI to drive workloads. Instead, a workload developer needs to create a custom form for each workload using XForms. This provides great flexibility and power in designing workloads as you can tailor the information you collect to the workload. On the other hand, editing XForms is a nightmare and error-prone and most benchmark developers spend the most amount of time in this one task. No longer. Faban 1.0 auto-generates the XForms for you from your run configuration file (run.xml) at deploy time. How cool is that !  And the generated forms are hard to tell apart from good hand-crafted ones – they are that good !

Dynamic Services and Tools

Faban always had the concept of Services (meaning objects that were responsible for managing a particular server application like Oracle, Apache etc.) A Service is used to start/stop/restart a server, grab it’s configuration and error logs etc. But to add a new service, one had to add a class to the Faban code. This meant a user had to not only understand Faban code but also had to get the new service code committed. The same was true for monitoring tools (For example, I added ‘mcstat’ – a tool for monitoring memcached but this had to go in as part of the Faban harness code).

With 1.0, both services and tools can be dynamically deployed. This means you can write your own services and tools and deploy them in your own environment. What’s more – Faban comes with several sample services to get you started. These include services for Apache, Glassfish, Memcached, MySQL, Oracle etc. I have added services for Thin and Nginx as part of Olio using this framework. See the documentation for more details.

Tags and Feeds

As the number of tests and results grow, we have all faced the situation where we can’t find a particular run result. Tags can greatly help here. Faban 1.0 has  web2.0 features including tags and RSS feeds. You can add tags when defining a new test (actually it’s easy enough to add them later as well, so you can go back and tag all your old runs if you want to). To find a particular test (or series of tests), you can do a Tag Search from the main Faban page.

You may wonder what in the world is the use of a Feed in a performance tool ? Well, if you kick off a series of runs before you leave work, you can sign up for the RSS Feed (located at the top right corner of the screen). This will send you an email whenever a run completes. So even if you don’t have access to your test systems, you can view the results anywhere – perhaps on your iPhone while you are at dinner ?

This post is already too long. I’ll cover the other new features in a future post. But meanwhile, do check out Faban 1.0.

Translate to:English

Benchmark GoLang vs Python

Manos a la obra. Una de las principales motivaciones de GoLang (el lenguaje de programación Google) es obtener en un sólo lenguaje la flexibilidad y rápido desarrollo que proporciona Python pero con la velocidad que un código nativo como C proporciona.

Quiero comprobar esto con un ejercicio inmediato y práctico: hacer un benchmark comparando un servidor web en python contra otro en GoLang.

Las herramientas que he usado son:

• Servidor web simple en Python (Python 2.5.2) [Fuente]
• Servidor web simple en GoLang (GoLang hg 10/11/2009 snapshot) [Fuente]
• Apache HTTP server benchmarking tool (ab v2.3)

La máquina donde he ejecutado los tests es:

• Intel Core 2 Extreme X9770 3.2 Ghz
• Ubuntu 8.10 Intrepid amd64
• Kernel 2.6.27-15-generic

Vamos allá.

1. Copy/paste del código de ejemplo del servidor web para GoLang (con modificación de la plantilla)
2. Copy/paste del código de ejemplo del servidor web para Python (con modificación de la plantilla)
3. Compilamos el código GoLang
   % 6g webserver.go
   % 6l -o webserver.out webserver.6
4. Lanzamos el servidor web en GoLang
   % ./webserver.out
5. Ahora lanzamos el test para 40.000 peticiones con nivel de concurrencia 8 (ese nivel de concurrencia da un mínimo local en tiempo total de respuesta)
   % ab  -n 40000 -c 8 http://localhost:1718/
6. Matamos el servidor y lanzamos ahora el de Python
   % python webserver.py &> /dev/null
7. Aplicamos el test anterior
   % ab -n 40000 -c 8 http://localhost:1719/
8. Por último comparamos los resultados

GoLang:

Python:

Concurrency Level:      8
Time taken for tests:   10.644 seconds
Complete requests:      40000
Failed requests:        0
Write errors:           0
Total transferred:      8000000 bytes
HTML transferred:       3360000 bytes
Requests per second:    3758.13 [#/sec] (mean)
Time per request:       2.129 [ms] (mean)
Time per request:       0.266 [ms] (mean, across all concurrent requests)
Transfer rate:          734.01 [Kbytes/sec] received
Connection Times (ms)
min  mean[+/-sd] median   max
Connect:        0    0   0.0      0       0
Processing:     0    2  33.8      1    6448
Waiting:        0    2  33.8      1    6448
Total:          0    2  33.8      1    6448
Percentage of the requests served within a certain time (ms)
50%      1
66%      2
75%      2
80%      2
90%      2
95%      2
98%      2
99%      2
100%   6448 (longest request)

Conclusión:

En este sencillo test (poco riguroso, la vedad), el servidor de Python tarda en servir 40.000 solicitudes 10,644 segundos, mientras que el servidor web GoLang tarda 4,849 segundos.

El servidor web GoLang es 119% más rápido.

Requests/s GoLang vs Python

Entradas relacionadas:

GoLang, nuevo modelo de programación
GoLang no es para Windows

Translate to:English
Menéame Twitter

Postal is used to benchmark SMTP servers.
I am using it so that I can send mails continously to my test server and hence check if it is working properly. Hence this saves me from the pain of composing mails everytime I do some changes in the server and check if the changes I made is what I intended.

All I have to do is execute one command given below.

postal -m 100 -f test -t 5 192.168.31.100 users_mail

users_mail has the list of all the users in my test server.