Category: Editorial

Would you do this? Combine two calls, like an INSERT and then a SELECT, into the same batch sent from a ADO.NET client? It’s a proposal from Visual Studio Magazine to combine code into one batch instead instead of making two query calls (and two round trips).

It’s an interesting idea, and for those programmers that are older, we probably unconsciously think about this all the time. We still remember when bandwidth was measured in kbps and processing power in MHz, and we often consider those resources to be valuable. We try not to waste them, precisely because they impact performance. Perhaps that’s why we recoil at some of the ORM frameworks and other programming structures that make development easier but are tremendously chatty in how they communicate with our servers.

I know that programmer salaries are high, and that labor is one of the more expensive parts of building software. However rather than building inefficient applications that will create customer complaints, shouldn’t we be educating our developers more to ensure they write efficient applications? They can still use a framework or template of some sort, but they should do so in an efficient way. Along with training, using code reviews and keeping everyone on a central VCS where all code is visible can help spread knowledge throughout your staff.

I realize that many managers are loathe to invest in developers’ knowledge when those developers might leave your company for another position, but imagine that a manager elsewhere is investing in his developer. It’s entirely possible those other developers then come to work for you. Ultimately we all benefit if we work to spread and build better software development practices throughout the industry.

Steve Jones

The Voice of the DBA Podcast

Listen to the MP3 Audio ( 2.1MB) podcast or subscribe to the feed at iTunes and LibSyn.

Recently I was practicing for the keynote talk at the 2014 SQL in the City events. There were three of us in the “Ship Often, Ship Safe” talk, and we’d been practicing remotely. For the most part that meant a Skype call where we could work on timing and a Join.me share of a remote desktop to work on demos. The whole delivery became smoother and smoother as we got closer to the event, and I’m glad we spent a few weeks practicing (that’s important).

However, we had an issue when we all arrived in London to practice. The first few run throughs, using a live laptop instead of a remote session, went much slower and less smoothly. In fact, the entire timing of the keynote was a problem because certain tasks were taking entirely too long. We were starting to think that we’d have to rewrite the talk in places, or take screen shots and show those.

One of the presenters then suggested that we find a working network connection and retry our demos. It sounded a bit silly, but we relocated to a hotel room and connected to the network. Sure enough, everything ran faster.

It turns out that there were various services, our own products, IE, and a few more that all had a few delays built in to check something across the network. Together, those delays added up to substantial time lags for various processes.

While I think we’ve come a long way towards building amazing connectivity at high speeds, there are still plenty of times, including common tasks taking place in server rooms, where we don’t have a live connection to the Internet. I would hope that people would consider that the network might not be live and allow their services to continue to function if Internet access is not a core requirement. I suspect far, far too many developers still do not think of this.

I don’t know that we’ll get better written software to allow for a lack of Internet access, but if you find things running slow, you might try enabling an Internet connection and seeing if your performance improves.

Steve Jones

The Voice of the DBA Podcast

Listen to the MP3 Audio ( 2.7MB) podcast or subscribe to the feed at iTunes and LibSyn.