Fearless Bit

Ran across this article just now and wanted to link to it cuz the concept is really fascinating to me!

http://www.jroller.com/nivanov/entry/cloud_based_thread_pool

What’s more important for a SaaS business: making money on every user or having so many users that you can then create features that are only useful when you have these kinds of numbers?

This article on GigaOM really caught my attention because it’s dear to the hearts of the team I work with.  Creating feature after feature may not be what’s needed to grow — but rather create one or two compelling reasons to be on the platform, and then use that network effect to create additional features that benefit large numbers of users.

What makes the company that provides an online service successful?  At a mid-level view, the list is large, so I’ll focus on the items related to this posting:

1. Working smarter, not harder
2. Try, learn, and improve
3. Solve problems that matter to it’s customers, employees, and shareholders

These often mean different things to leadership than they do to an engineer like me.  We share the same high-level goals, but our lenses are naturally different.  Managed correctly, this is a good thing.  However, it’s human nature for these differences to create problems.

When the organization chooses a direction that I know from experience will violate one of the 3 items above, I find myself in a quandary.  Some people tell me to “support the business.”  Others tell me “it’s just a job.”  Personally, I take great pride in my work and feel very personally dissatisfied knowingly choosing (or being forced to choose) the wrong path.  I also want to do what I can to help the business not only succeed, but thrive.  I don’t like to lead a mediocre career.  So what should I do in these situations?

The answer might seem simple: speak up!  You can safely assume that I do.  So what’s wrong?  Where is the message getting lost?  How does an admin/engineer provide cost/benefit or ROI analysis on something as vague as architecture (especially without the resources to test its limits)?

There’s too much to cover in one posting, so here’s an outline of what I’m thinking about for my next posts:

• Trust your senior engineers in their domains of expertise; allow them to make mistakes (see #2)
• Allocate sufficient resources towards becoming and staying proactive, not reactive (#1)
• Complexity kills; follow the KISS principle (#1)
• Don’t re-invent the wheel (#2, #3)
• Enforce processes via consequences (#1, #2, #3)
• Balancing long- and short-term (#1, #2, #3)
• Personal and organizational effectiveness topics (e.g. Stephen Covey’s publications)

This is meant to be a series of interactive posts — which is part of the reason for breaking it up into several digestible chunks.  I hope to absorb and integrate the feedback that I get.  Please share and comment!  Thank you!

Being that I own a really expensive high-performance SSD that’s only 64 GB, I torture myself by watching this technology space and if Corsair’s numbers for this drive hold up (and the drive holds up), the OCZ Vertex may have a real contender.  Engadget’s coverage is here.

• Release It! Review in “Quotes”
  A review of the book Release It! — there are a fair number of people in the comments of this review indicating this book is a must-read.  While I don’t develop software, I’ll probably read it as topics like these speak out at me:
  • Data purging isn’t sexy but needs to be done
  • Temporary fixes are inevitably permanent.  We all know the various forms of “temporary is never temporary” yet we all still do it.  If I could slap someone every time I saw it happened, my hand would be raw
  • Design for capacity; sure, the architects do it but the developers generally don’t get it
  • The cost of (un)reliability
  • Don’t GUI-up an admin interface.  I wish I could have a revision-controlled set of configuration files for production …
  • Logging needs to be desgined for operations; if you log an error, it better be because you need my help as a System Admin!
• Seven Productivity Tips For People That Hate GTD
  GTD = Getting Things Done.  This is an interesting look at some ways to improve your ability to execute.  I took a look through the site because there are a number of related books and articles and, well, I’m all about reading up on how much I don’t know …
• Why McKinsey’s Cloud Report Missed the Mark
  On my April 19th post, I referred to a McKinsey study regarding the real cost of moving your business to the “cloud.”  Joe Weinman at GigaOM points out that this isn’t such a black-and-white issue.  Much like there’s a valid, proven business model for renting cars at 5 times what it costs to own, there are reasons why someone might want to use “the cloud” even if it’s actually more expensive because it’s more flexible (e.g. you can instantly fire up 1,000 servers — try doing that at your own datacenter) which can translate into cheaper overall
• The Big Agile Practices Survey Report (Part 1)
  Did you read Fearless Bytes IV?  If so, you’ll be happy to know that the results are in for The Big Agile Practices Survey!  It should not only get you thinking about whether it matches what you value, but if you’re a manager or leader, it may make sense to see whether aligning yourself to these results or carrying out the same survey internally makes sense

Joe Weinman at GigaOM posts Another Half-Dozen Half-Truths of the Cloud.  I like the first one (Clouds are less secure) — but the real question is, how do we as service providers get the consumer to trust us?

The third item (If cloud services cost more, they shouldn’t be used) refers to this McKinsey study; I debated posting that study here last week because I want more to communicate that picking up and moving some or all of your “shop” to “the cloud” isn’t nearly as easy as some business leaders are inclined to believe.  At first I thought this study would help make my case but it goes on to say that it is more cost-effective for smaller enterprises.  Perhaps it is, but there’s a lot of hidden cost in moving towards it and supporting the processes behind it.

I don’t use Eclipse myself but a number of my colleagues do so I thought I’d share this blog.

• Why the CAPTCHA Approach Is Doomed
  CAPTCHA doesn’t work.  Don’t waste your time on it.  I’ve seen first-hand that humans are often behind the keyboard abusing your system and I’m pretty sure they can get through CAPTCHA.
• The Big List of Agile Practices
  A well-organized list of Agile practices.  Now that this list has been solidified, they’re now running a survey (which I encourage you to participate in) along with immediate survey results.
• Buy vs. Build
  More “please stop re-inventing the wheel” content.  Sometimes its done out of fear of losing control because you’ve been burned by bad libraries in the past but think about it — the time spent on creating your own library and maintaining it over time far outweighs debugging a few bugs here and there with a vendor-provided library or working with the open source community.  You are 1 person with a side-focus of creating this library to enable your product.  They are (usually) multiple people focused on the library you could be using.
• “I don’t have time to test!”
  Don’t ever repeat this phrase with a straight face in front of me   The fourth comment in the article has a nice reference as well.
• http://codezest.com/archive/2008/09/28/development-tools-i-use-amp-recommend.aspx
  I always love lists of favorite developer tools so I can check out what others are using!
• Free Illustrated C# 2008 Ebook
  I haven’t checked this out yet but really wanted to pass this along ASAP!
• 10 Way to Improve Your Code
  Neal Ford presents for an hour at QCon.  It started off a little shaky for me but I watched the whole thing — it’s worth the watch.  Topics include: composed methods, the problem with singletons (and how to solve it) — mixing “static” with “state”, YAGNI, TDD, pair programming, polyglot programming, parallelism, and a good top-10 signs you’ve got a problem.

Picked up this article from Between the Lines today.  I like HP’s servers but recently they seem to be late to the game.

The real question (for me) is: where is the server technology going and is the business I support aligned with this direction?  I don’t see that there is alignment there but fortunately virtualization means lots of memory and the software I support loves memory so HP’s recent upgrade to their DL360 line which maxes out at 144 GB of memory is a welcome one!

In my previous SSD post, I mentioned that the write performance of MLC drives can slow down with use.  It appears that Intel’s latest firmware update for their X25-M series of SSDs has solved this problem and the drive maintains incredible peformance.  PC Perspective reviews the new update with some impressive results.