Matt Callanan's Blog

Tuesday, July 31, 2012

Lead Time vs Cycle Time

http://www.isixsigma.com/methodology/lead-time-vs-cycle-time/

( http://hbswk.hbs.edu/archive/1460.html )

1. Lead Time and Cycle Time don't have the same unit although their names are both "Time." Lead Time is measured by elapsed time (minutes, hours, etc.), whereas Cycle Time is measured by the amount of time per unit (minutes/customer, hours/part, etc.). It does not make any sense to add one to, or subtract one from, another.

2. Cycle Time is actually a measure of Throughput (units per period of time), which is the reciprocal of Cycle Time. This relationship is analogous to Takt Time (amount of time per unit), which is the reciprocal of customer demand rate (units per period of time). Note that by definition, Cycle Time (or Takt Time) is an average value.

3. Lead Time and Cycle Time are related by Work-in-progress (WIP) in the entire process, in a relationship described by the Little's Law:

Lead Time = Cycle Time * WIP
Or,
Lead Time = WIP/Throughput

4. The Cycle Time above must be the process cycle time, which is determined by the bottleneck. Cycle Times of individual steps cannot be used alone to calculate the process Lead Time without knowing the WIP.

Wednesday, July 25, 2012

Ansible

Ansible: http://ansible.github.com/

Simple config management
No server
No daemons
Uses ssh and simple scripting language for:
- (1) Configure
- (2) Deploy
- (3) Ad-Hoc Tasks

Tuesday, July 24, 2012

Kanban from the horse's mouth

DevOps Cafe Episode 30

http://devopscafe.org/show/2012/6/13/devops-cafe-episode-30.html

A great podcast interview with David Anderson and Dominica DeGrandis about the roots of Kanban and how to use it to introduce process improvement in an organisation.

Notes:

Kanban = culture of evolution
Original Kanban experiement at Microsoft was digital (ticketing system with database triggers)
Toyota let people see their processes because they know by the time they've copied it Toyota will have evolved further and they know that culture is much harder to copy especially for western cultures at a neuropshycological level
Guiding principles

Start where you are now
Agree to pursue change
Respect the current situation

Drum-buffer-rope analogy from "The Goal"
Blog article: Understanding process of knowledge discovery
Value Stream Mapping?

Metaphor breaks down
Stream = water flow
Water can't flow uphill - IT releases involve rework that's hard to map
Better: "Workflow mapping"
Go around objections

Create environment where people can experiment and come up with own unique processes

Don't create new process for people - create a culture where they can evolve their own process & understand individual changes.
Make many small adjustments over the long haul - evolving with each change
Snowball effect
Evolution gets you there faster
Like switching to Mac - can't explain it until you experience it

Pull

Toyota don't do a pure pull system
Users don't always know what they want

Eg jaguar asked jaguar owners to describe what they want in the new jaguar design - they designed the same car - can't imagine

"But we're a scrum/agile shop"

Tribal reaction - emotional response
Resistance to change
Can't have logical discussion about devops
Suggest a "super tribe"

Lessons agile mgmt

Introduction to Chef

http://learnchef.getharvest.com/introduction.html

Sunday, July 22, 2012

Cost of software / Capers Jones links

Tech debt article: http://swreflections.blogspot.com.au/2012/02/technical-debt-how-much-is-it-really.html

Capers Jones PDF: http://sqgne.org/presentations/2011-12/Jones-Sep-2011.pdf

Capers Jones facts and figures: http://allankelly.blogspot.com.au/2011/02/more-facts-and-figures-from-capers.html

Monday, July 16, 2012

Installing RVM on Ubuntu

Install RVM instead of Installing Ruby.

http://rubysource.com/installing-ruby-with-rvm-on-ubuntu/

bash < <(curl -sk https://raw.github.com/wayneeseguin/rvm/master/binscripts/rvm-installer)
echo '[[ -s "$HOME/.rvm/scripts/rvm" ]] && . "$HOME/.rvm/scripts/rvm"' >> ~/.bashrc
source ~/.bashrc

Sunday, July 15, 2012

BDD and REST

Testing REST with BDD
Brian Sletten 

Presentation: http://skillsmatter.com/podcast/ajax-ria/sletten-rest-bdd?goback=%2Egde_4079964_member_112404082

The move to a good REST API yields an explosive combination of options due to content-negotiation and arbitrary workflows. At the same time, the uniform interface you project (representations, how you respond to verbs, response codes, etc.) becomes your contract with your clients. The API itself becomes a completely transportable, reusable bit of organizational knowledge when done well. Testing it sufficiently will give you a safety net but may drive you crazy unless you do it right. We will walk through the use of Behavior-Driven Development (BDD) to test these kinds of APIs.

We will highlight the   problems of sufficiently testing REST APIs while providing a rich and relatively straight-forward solution to the problem. We will use a Behavior-Driven Development (BDD) tool like Cucumber to establish reusable steps and comprehensive, but lightweight testing strategies for testing REST APIs.

How to sell BDD to the business
Dan North

Presentation: http://skillsmatter.com/podcast/soa-rest/how-to-sell-bdd-to-the-business
Slides: http://skillsmatter.com/custom/presentations/sellingbddtothebusiness_bdd.pdf

I don't care how you do it - I just want it better, faster and cheaper and I don't want you to keep bugging me

How do you present the benefits case of BDD to a sceptical, time-starved, budget-focused business sponsor? When we talk about Writing software that matters it is from the perspective of the business as much as anybody else.

Software delivery shouldn't be a zero sum game where you have to choose between writing quality software, showing bottom line results and having fun. In this talk, Dan shares his experiences of introducing BDD and Agile principles to senior stakeholders and business executives.   Download the slides here: http://skillsmatter.com/custom/presentations/sellingbddtothebusiness_bdd.pdf

Blog: BDD for secure REST APIs using cucumber, pickle, and rack-test
http://blog.tddium.com/2011/04/22/bdd-for-secure-rest-apis-using-cucumber-pickle-and-rack-test/

Wednesday, July 11, 2012

How to show Maven classpaths

From: http://maven.apache.org/plugins/maven-antrun-plugin/examples/classpaths.html

Add this under <build><plugins> in your pom, then run it with mvn compile :

<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-antrun-plugin</artifactId>
<version>1.7</version>
<executions>
<execution>
<id>compile</id>
<phase>compile</phase>
<configuration>
<target>
<property name="compile_classpath" refid="maven.compile.classpath"/>
<property name="runtime_classpath" refid="maven.runtime.classpath"/>
<property name="test_classpath" refid="maven.test.classpath"/>
<property name="plugin_classpath" refid="maven.plugin.classpath"/>

<echo message="compile classpath: ${compile_classpath}"/>
<echo message="runtime classpath: ${runtime_classpath}"/>
<echo message="test classpath: ${test_classpath}"/>
<echo message="plugin classpath: ${plugin_classpath}"/>
</target>
</configuration>
<goals>
<goal>run</goal>
</goals>
</execution>
</executions>
</plugin>

Tuesday, June 26, 2012

UberConf 2012 notes

Complexity of Complexityby Ken Sipe - Wednesday 3:15 PM
 

Essential complexity = plane 
Accidental complexity = Indian power lines 
V1 user does not like 
V2 overly complex 

Everything is diffing mental models 
Size of team matters adding ppl is counterproductive 

Application of knowledge, level of experience 

Dreyfus model 

Experts create recipes for novices and advanced beginners 

Brian Goetz, extending Java with backward comp 

1952 
Make coding  comprehensible 

Test reviews more important than code reviews 

2.0 - 0.9 = 0.89999999 
Have to have humans working on code that understand this 

Possible to create obj hierarchy where legal equals is not possible 

Must use most powerful language at disposal 

Increased flexibility = increased complexity 

ESB - complexity does not go away, need to be smart 
Internals are not standardized 

CAP Availability = caching 
"if u can't split it u can't scale it" shoelz eBay 

Level 3 = competent
Super observer sees all consistency - not necessary? 

What is core   business?  Can solve with fault tolerant hardware instead of sware complexity? 

Context... Hard problem 
Google talk hardest problem is social graph (friend status) 
IO throughput, CPU throughput, two solutions 

Request, asynchronous mq, response 
Don't cache preemptively
MongoDB: Scaling Web Applicationsby Ken Sipe - Wednesday 5:00 PM
 

Sharing 3x3 is magic 
3 shards, replicate sets
Mobile Performance Tips n' Tricksby Pratik Patel - Wednesday 5:00 PM
 

Blaze.io/testing
Visualizing Data on the Webby Brian Sletten - Wednesday 8:30 PM
 

Preattentive 200ms 

Stevens power law ppl find length easier to perceive than area - pie charts are harder to understand 

Show the data - don't accentuate, "lie factor" 

Max ideas presented min time, min ink, min   space 
"Chart junk" 

Spark lines, word-size graphics, eg glucose level inline with text 

Jquery sparkling 
Rgraph 
Highcharts commercial eg Gatling 
Raphael 
D3 data drives with little code, based on protovis 
-  multivariate relationships 
- parallel coordinates 100g 
Tangle worrydream, Bret Victor 
- interactive document 
exhibit, simile-widgets 

Flowingdata.com "visualize this" book 

Many eyes IBM 
Selection.datavisualization.ch 
inmaps.linkedinlabs.com
Clojure Workshopby Neal Ford - Thursday 9:00 AM

Watch Rich Hickey videos 

Dynamic 
Strongly  typed 
Functional 
High-performance 
Lisp 

Homoiconic 
Only references are mutable 

Regex #"" 
Symbols 
Keywords special symbols :foo ::foo, constant intern strings 
Data literals 
-List. Singly linked insert at front () 
-Vector. Indexed, insert at rear [] 
-Map key/value {:} 
-Set #{} 

Few data structures many functions 

Seq interface

Convert Groovy NodeChild xml element to XML string

new groovy.xml.StreamingMarkupBuilder().bindNode(nodeChild) as String

Monday, June 25, 2012

ElasticSearch terminology

Relational DB ⇒ elasticsearch

database ⇒ index
table ⇒ type
row ⇒ document
column ⇒ field
schema ⇒ mapping
index ⇒ everything is indexed
SQL ⇒ query DSL

Sunday, June 24, 2012

Kibana: A brower based log analysis tool for Logstash and ElasticSearch

http://rashidkpc.github.com/Kibana/

You have logs. Billions of lines of data. You shipped, dated it, parsed it and stored it. Now what do you do with it? Now you make sense of it. Kibana helps you do that. Kibana is an alternative browser based interface for Logstash and ElasticSearch that allows you to efficiently search, graph, analyze and otherwise make sense of a mountain of logs.