Choosing the Right Tool

Kyle Mestery of Cisco recently shared a link on Twitter about Marco Arment’s choice to move back to a dual Mac setup. The article started me thinking along two parallel lines:

  1. First, it got me to thinking about my own Mac setup.
  2. Second, I wondered if there was a similar parallel about the choice of tools in data centers today.

Allow me to explain. In his article, Marco talks about how he abandoned his dual Mac setup—in which he was using a 13″ MacBook Air and a Mac Pro—for a single Mac setup using a really beefed up 15″ MacBook Pro. He had hoped to reduce the overhead of managing multiple Macs by consolidating to a single Mac that would bring together the best attributes of both. What he found a year later was that his attempt to use one tool (the 15″ MacBook Pro) really ended up being less beneficial instead of more beneficial. Instead of getting the best of both worlds, he instead inherited the drawbacks of both. To that end, he’s now moving back to a dual Mac setup.

<aside>Just as a point of clarification for those who might be unfamiliar with Apple’s product lines: the MacBook Air is a highly portable ultraslim notebook with limited expandability; the MacBook Pro is Apple’s more expandable and more powerful notebook; and the Mac Pro is a workstation-class desktop computer with oodles of CPU cores and gobs of RAM.</aside>

The first thought process that occurred to me regarded my own Mac setup. I recently purchased a Mac Pro because I needed a computer that could provide more raw compute capacity than my laptop possessed. Reading Marco’s article validated (in a way) my thinking. However, it also challenged me to consider the type and configuration of laptop that I use. I migrated from a 15″ MacBook Pro to a smaller 13″ MacBook Pro last year, but I still insisted on a MacBook Pro with 8GB of RAM instead of a lightweight MacBook Air with only 4GB of RAM. Don’t get me wrong; I’m extremely happy with my MBP. It does lead me to believe, though, that my next laptop choice needs to be a choice that optimizes its role. What do I mean by that? I chose a Mac Pro because I needed CPU power and lots of RAM. The Mac Pro was the right tool for that job. Similarly, when I select my next laptop, I need to prioritize what I need out of a laptop (weight, size, mobility) and choose the right tool for the job. Instead of choosing a laptop based on how expandable it is, perhaps I should be looking at how suited it is for a highly mobile worker.

There’s a second train of thought here as well, and this train of thought pertains to the tools that we, as IT professionals, choose for our data centers. We also need to make sure that we are selecting the right tool for the job. Marco’s experience shows that using a single tool to perform multiple functions doesn’t always work as well as one might think. He made his initial decision in an effort to reduce complexity—an admirable goal, I’d say. I believe that many IT professionals also strive to reduce complexity in their data centers, and many IT professionals probably do that by reducing the number of technologies, products, or vendors in their data center. But are we sacrificing the functionality of our data centers as a result? When we are choosing on the basis of reducing complexity instead of choosing the right tool for the job, what are we losing? What are we giving up? Instead of trying to shoehorn a solution we already have into a role for which it really isn’t suited simply to “reduce complexity,” shouldn’t we focus instead on choosing the right tool for the job? Shouldn’t we select tools that are optimized to perform the function we need them to perform? Yes, we do need standards and guidelines, and I’m not saying that we shouldn’t strive to keep complexity from overwhelming the data center. But what should be our primary driver for tool selection—the reduction of complexity via the re-use of a less-than-optimal tool, or the selection of a tool optimized for the function it needs to perform?

I’d love to hear your thoughts in the comments.

Tags: ,

  1. Mike White’s avatar

    The power of your analogy was predicated on the reader appreciating the stark contrast between a MacBook Air and a MacBook Pro. I suggest only the most radical phoaming phanboi would have the slightest intuition about that difference.

    Try again.

  2. slowe’s avatar

    Mike W, thanks for the implied statement that I’m a “radical phoaming phanboi.” However, your clarification is appreciated; I’ll update the post to help make the distinction clearer.

  3. Nick Weaver’s avatar

    I have done a ton of projects on my Macbook Air (purchased Q3 ’11). I also have a 17″ monster laptop with 16GB of RAM and dual SSD. But I rarely use the latter for several reasons:

    1. The 13″ MBA *feels* just barely bigger than holding or using my first-gen iPad. Once I bought my MBA my iPad started gathering dust.
    2. The 4GB of RAM of the MBA is *just* enough to run one of two Linux VM’s for coding. But I usually use the MBA directly as a *nix machine as opposed to not being able to use the host laptop when it was Windows in the past. Being able to run MongoDB via brew on the MBA for a project saves needing a VM.
    3. The MBA battery is awesome
    4. Workflow with Lion + Fusion is fast

    Now that the new upgrade options are out I will look for my current pain points in a new device. Mainly:

    1. Need more RAM for more VM’s. Not having 8GB makes lab size limited. Having 16GB would be awesome.

    I have looked at the new retina-based MBP. But I am not willing to give up the easy of size with my 13″ MBA to go 15″ just for memory. And my 27″ Thunderbolt monitor is what I will end up using anyways when I am home over the retina display.

    So I am likely to simply upgrade to an 8GB MBA with newer proc and the bigger 512GB SSD. This basically keeps all the size and portability convenience I had with the ability to run a slightly larger laptop lab for development.

    But that is just for me. I should mention I have a beast workstation that I use for my large lab environments should the laptop not work. But I prefer mobile first over the home lab.

  4. Jivetolkein’s avatar

    Can I be the first to offer the catch all IT answer – ‘It depends’ :-)

    Generally speaking IME the suboptimal but standard tool gets the nod, usually by the holders of the purse strings.. a simpler environment is one thats easier to hire for, replace staff and outsource.

    It makes sense really if you can live with the 5% loss in whatever metric you choose I guess.

  5. G’s avatar


    Interesting read, I had seen Marco’s write up, both interesting views.

    The MP are a great work horse, it was a shame about the upgrade path, but there must be a reason for it, at least it was not discontinued.

    Your article doesn’t mention what spec you went for? I think probably the Dual CPU would be better suited, than the single hex due to your VMware work.

    Additionally, are you running ESXi within Fusion, trying to replicate multiple hosts? As that would be a great solution for testing?

    Any input would be great.

    Many thanks,


  6. slowe’s avatar

    Nick, thanks for your comment. It’s interesting that you actually seem to be moving in the direction that Marco was originally moving—away from multiple machines toward a single machine (in your case, a MacBook Air). Not that there’s anything wrong with that, I simply find it interesting. Thanks!

    Forbes, pretty impressive! I’ve also heard that the newer 2012 MacBook Pros can run a total of *four* displays. Wow.

    Jivetolkien, thanks for your comment. I’m glad the comments are moving more toward the second part of my post—that was where I really wanted the discussion! Do you suppose that IT organizations really quantify (or even consider) the potential loss of functionality, and evaluate to determine if the loss of functionality is acceptable? I agree that a minimal loss in functionality is generally acceptable (assuming that the loss isn’t in an area that is business critical or negatively impacts the business bottom line), but it seems that many IT orgs are so driven toward simplicity that they aren’t even considering the impact. Or am I mistaken? I’d love to hear your thoughts.

    G, thanks for your comment. I’m not currently using the Mac Pro for a lot of VMs; I’m using the cores (and the RAM) for video rendering. However, I do anticipate running more VMs. I don’t know how much nested ESXi I will do, given some of my self-imposed projects for this year.

  7. Paul C’s avatar

    I can’t count the number of times a vendor has offered me a product that does everything under the sun for a 6 digit sum. Not to mention another 6 digit sum for managed services to actually make that product do something useful in your environment.

    Products that dazzle managers with graphs and kpis and promises of detailed statistics yet fill the rest of us with dread as a CIO decides that we desperately need that product,
    yet ‘we have a smart bunch of guys’ so we don’t need managed services … we’ll work it out.

    months pass, even a year, and that product is sitting in service somewhere doing a bare minimum and mostly useless job, while in the background somebody has thrown together cobber/puppet for provising, somebody else has written a set of scripts to connect to remote console via ssh to collect hardware information into a database, somebody else has run up nagios and mrtg and they all hum along doing exactly what they were intended to do.

    The product comes up for maintenance renewel, and you discover that the ‘discount’ they gave you was based on a highly inflated maintenance renewel and it gets declared a failure that the team didn’t unlock the promise of the product and it is turned off forever.

  8. G’s avatar


    I see, yes, for video rendering they are superb, in fact, the hex is better for that.

    Would be interesting to see how you personally find the MacPro as an ESXi lab tool, using Fusion.

    I think Fusion is a great tool but no where near as good as Workstation which is a shame.

    32/64gb of RAM is relatively cheap and using some SSD’s will really improve the performance on the VM’s as you know :)

    Shame these toys are so expensive, but when you compare the Dual Processor versions next to the HP Z series or the Dell, they aren’t too bad on the price, specially as the quality build is great.

    Hope all is well,



  9. Dan’s avatar

    I liked how you set up your article so that the thesis statement of “what should be our primary driver for tool selection—the reduction of complexity via the re-use of a less-than-optimal tool, or the selection of a tool optimized for the function it needs to perform?” I helped me picture what you were trying to say.
    Good insight.

  10. Mike’s avatar

    Good post Scott. I’m in a similiar quandary. I have new laptop on its way – had a choice between dell, mbp13″ or air. Would have liked the option to mac pro tower but that sadly wasn’t an option. Thinking of ebaying my “old” MBP13″ to put it towards a mac pro tower… and then using dropbox/box/skydrive/octopus to keep the data in synch between the two.

    Carmel’s suggested i mothball the old MBP13″ as “backup” laptop should the new one go pop unexpectedly…


Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>