I've been quiet lately for a couple of reasons.
First, I'm thrilled to say that I'm joining the No Fluff, Just Stuff stable of speakers. It's an honor and a pleasure to be invited to keep such company. The flip side is, I'm spending a lot of my free time polishing up my inventory of presentations. More frankly, I'm rebuilding them all with Keynote. (Brief aside, I'm coming to love Keynote. It has some flaws and annoyances, but the result is worth it!)
I'll debut the first of these new presentations at OTUG on May 15th. I'll be speaking about "Design for Operations". The talk will be about 70% from the last part of Release It, and about 30% original content. OTUG will be giving away a couple of copies of my book, but you have to be there to win!
Finally, I'm working on an article about performance and capacity management. Most capacity planning work is done entirely within Operations, without much involvement from Development. At the same time, most developers don't have a visceral appreciation for how dramatically the application's efficiency can affect the system's overall profitability.
This article will show the relationship between application response time, system capacity, and financial success. I'm hoping to include a simulator app for download that you can use to play with different scenarios to see what a dramatic difference 100ms can make.
Is it an antipattern to have a consulting firm provide both the coach and developers? By providing the developers, the firm is motivated to deliver on the project, with coaching as an adjunct. If, instead, the firm provides just the coach, it will be judged by how well the client adopts the process. These two motives can easily conflict.
Case in point: at a previous client of mine, my employer was charged with completing the project, using a 50-50 mix of contractors and client developers. My employer, a consulting firm, provided several developers experienced with XP and Scrum, as well as an agile coach. The firm was thus charged with two imperatives: first, deliver the project; second, introduce agile methods within the client.
With project success as a requirement, the firm decided to intereview the developers at the outset of the project. The client's developers (rightly) perceived that they were interviewing for their own jobs. This started a negative dynamic that ultimately resulted in 80% attrition among the client's developers.
On a pure coaching engagement, the coach would probably have "made do" with whomever the client provided.
We delivered all the features, basically on time, with very high quality. Financially speaking, it was a success, generating more orders and more revenue per order than its predecessor. It is harder to say that the engagement as a whole was a success, though. Almost all of the developers were contractors, so the client got their product, but very little adoption of agile methods.
Perhaps if the coach and the contract developers had come from different firms, the motivations would not have been as tangled, and more of the client's valuable people would have stayed. The team might not have suffered from the strained, unhealthy environment from the early days of the project.
Then again, perhaps not. The client may have been expecting that level of attrition. Maybe that's just to be expected when you trying to bring a random selection of corporate developers over to agile methods, especially if the methods are decreed from above instead of brought upward by grass-roots. Maybe the dynamic would have existed even with a coach that was totally disinterested in the project outcome.
Since NetInfo Manager is going away under Leopard, we've got a gap in capability. How do you relocate your home directory without the GUI?
There are a few reasons you might want to move your home directory to another volume. For example, you might reinstall your OS frequently. Or, perhaps you just want to keep your data on a bigger disk than the one that came in the machine. In my case, both.
The venerable NetInfo is being replaced entirely with Directory Services. (Try "man 8 DirectoryServices" for more information.) There's a handy command-line tool you can use to interact with the DirectoryServices.
Let's start by opening up a Terminal window. (Applications > Utilities > Terminal) At first, you'll be logged in as yourself, not as root.
Last login: Wed Dec 31 18:00:00 on ttyp0 donk:~ mtnygard$
The first thing is to get out of your home directory, because we're going to delete it in about a minute and a half. Change to the root directory and make yourself into the root user with "sudo".
Last login: Wed Dec 31 18:00:00 on ttyp0 donk:~ mtnygard$ sudo su - Password: donk:~ root#
Next, fire up "dscl", the directory services command line. Without arguments, this gives you an interactive, shell-like environment to explore the directory. It also spews a bunch of help messages. If you give it "localhost", then it quietly assumes you wanted to interact with the directory.
You can list entries, cd around the directory hierarchy, and even create entries or change attributes.
User information is stored under /Local/Users, so we'll cd to that now.
donk:~ root# dscl localhost > cd /Local/Users /Local/Users >
Now, running "ls" will show you all the users that your machine knows. Try it now.
donk:~ root# dscl localhost > cd /Local/Users /Local/Users > ls _amavisd _appowner _appserver _ard _calendar _clamav _cvs _cyrus _eppc _installer _jabber _lp _mailman _mcxalr _mdnsresponder _mysql _pcastagent _pcastserver _postfix _qtss _securityagent _serialnumberd _spotlight _sshd _svn _teamsserver _tokend _unknown _update_sharing _uucp _windowserver _www _xgridagent _xgridcontroller daemon mtnygard nobody root /Local/Users >
Holy crap! Who the hell are all these people?
Well, of course, they aren't people. All the usernames starting with an underscore are application IDs. Root, nobody, and daemon are all part of the OS. Once you eliminate them, there should just be the people you've actually created accounts for. If you see any names you don't recognize at this point, this would be a good time to shut off your network connection.
At this point, you could "cd" directly into the entry for your user. It won't show you anything special; users do not have subnodes in the directory. It would set up your context for future commands, limiting them to just that user. In this case, however, we'll stay at /Local/Users and run "cat" on my username.
/Local/Users > cat mtnygard dsAttrTypeNative:_writers_hint: mtnygard dsAttrTypeNative:_writers_jpegphoto: mtnygard dsAttrTypeNative:_writers_passwd: mtnygard dsAttrTypeNative:_writers_picture: mtnygard dsAttrTypeNative:_writers_realname: mtnygard dsAttrTypeNative:authentication_authority: ;ShadowHash; dsAttrTypeNative:generateduid: 7F6A8EDE-63EC-4A34-9391-031A9C77806D dsAttrTypeNative:gid: 501 dsAttrTypeNative:hint: dsAttrTypeNative:home: /Users/mtnygard dsAttrTypeNative:jpegphoto: ffd8ffe0 00104a46 49460001 01000001 00010000 ffdb0043 00020202 ... 7fffd9 dsAttrTypeNative:name: mtnygard dsAttrTypeNative:passwd: ******** dsAttrTypeNative:picture: /Library/User Pictures/Sports/Tennis.tif dsAttrTypeNative:realname: Michael Nygard dsAttrTypeNative:shell: /bin/bash dsAttrTypeNative:uid: 501 AppleMetaNodeLocation: /Local/Default AuthenticationAuthority: ;ShadowHash; AuthenticationHint: GeneratedUID: 7F6A8EDE-63EC-4A34-9391-031A9C77806D JPEGPhoto: ffd8ffe0 00104a46 49460001 01000001 00010000 ffdb0043 00020202 ... 7fffd9 NFSHomeDirectory: /Users/mtnygard Password: ******** Picture: /Library/User Pictures/Sports/Tennis.tif PrimaryGroupID: 501 RealName: Michael Nygard RecordName: mtnygard RecordType: dsRecTypeStandard:Users UniqueID: 501 UserShell: /bin/bash /Local/Users >
Hmm. Seems like it must mean something. This is listing the values of all the attributes of my user profile. It's what I want, but there's a big pile of noise in the middle. That noise is a textual representation of my profile's JPEG. (I've edited it out of this transcript.) If you scroll up past that, you'll see the attribute of real interest.
The property dsAttrTypeNative:home tells the OS where to find my home directory.
I can change it with dscl's "change" command. The format of change is a little strange because it has to deal with multi-valued properties (as do all of the directory services commands.)
/Local/Users > change mtnygard dsAttrTypeNative:home /Users/mtnygard /Volumes/Data/mtnygard /Local/Users >
The first parameter is the object to change, the second parameter is the attribute to change. The third parameter is the old value that you want to replace (multi-valued list for each attribute, remember.) Finally, the fourth parameter is the new value you want to set.
Not quite done yet, though. I've given the OS a bogus home directory. There's no such directory as /Volumes/Data/mtnygard yet.
To get there, I have to move my directory from under /Users to the new location. I have to do this as root, but I don't want root to end up owning all my personal stuff. Fortunately, there's a "cp" option for that.
donk:~ # cp -Rp /Users/mtnygard /Volumes/Data/
Now, we're almost, almost done. Log off and log back on into your roomy new home directory.
My friend David Hussman once said to me, "The next person that says the word 'POJO' to me is going to get stabbed in the eye with a pen." At the time, I just commiserated about people who follow crowds rather than making their own decisions.
David's not a violent person. He's not prone to fits of violence or even hyperbole. What made this otherwise level-headed coach and guru resort to non-approved uses of a Bic?
This weekend in No Fluff, Just Stuff, I had occasion to contemplate POJOs again. There were many presentations about "me too" web frameworks. These are the latest crop of Java web frameworks that are furiously copying Ruby on Rails features as fast as they can. These invariably make a big deal out of using POJOs for data-mapped entities or for the beans accessed by whatever flavor of page template they use. (See JSF, Seam, WebFlow, Grails, and Tapestry 5 for examples.)
Mainly, I think the infuriating bit is the use of the word "POJO" as if it's a synonym for "good". There's nothing inherently virtuous about plain old Java objects. It's a retronym; a name made up for an old thing to distinguish it from the inferior new replacement.
People only care about POJOs because EJB2 was so unbelievably bad.
Nobody gives a crap about "POROs" (Plain old Ruby objects) because ActiveRecord doesn't suck.
In Release It, I talk about users and the harm they do to our systems. One of the toughest types of user to deal with is the flash mob. A flash mob often results from Attacks of Self-Denial, like when you suddenly offer a $3000 laptop for $300 by mistake.
When a flash mob starts to arrive, you will suddenly see a surge of TCP/IP connection requests at your load-distribution layer. If the mob arrives slowly enough (less than 1,000 connections per second) then the app servers will be hurt the most. For a really fast mob, like when your site hits the top spot on digg.com, you can get way more than 1,000 connections per second. This puts the hurt on your web servers.
As the TCP/IP connection requests arrive, the OS queues them for servicing by the application. As the application gets around to calling "accept" on the server socket, the server's TCP/IP stack sends back the SYN/ACK packet and the connection is established. (There's a third step, but we can skip it for the moment.) At that point, the server hands the established connection off to a worker thread to process the request. Meanwhile, the thread that accepted the connection goes back to accept the next one.
Well, when a flash mob arrives, the connection requests arrive faster than the application can accept and dispatch them. The TCP/IP stack protects itself by limiting the number of pending connection requests, so if the requests arrive faster than the application can accept them, the queue will grow until the stack has to start refusing connection requests. At that point, your server will be returning intermittent errors and you're already failing.
The solution is much easier said than done: accept and dispatch connections faster than they arrive.
Filip Hanik compares some popular open-source servlet containers to see how well they stand up to floods of connection requests. In particular, he demonstrates the value of Tomcat 6's new NIO connector. Thanks to some very careful coding, this connector can accept 4,000 connections in 4 seconds on one server. Ultimately, he gets it to accept 16,000 concurrent connections on a single server. (Not surprisingly, RAM becomes the limiting factor.)
It's not clear that these connections can actually be serviced at that point, but that's a story for another day.