Archive for March 2009
Monday Morning Head-Start [Software Business Links]
Software related links to start your week!
1. Email – Proving you are you – DKIM
2. Building good teams – a paradox
4. EU rejects French anti-p2p stance
Find more Software Business information OMS SafeHarbor.
Monday Morning Head-Start [Software Business Links]
Software related links to start your week!
1. Bug Tracking and Distributed Version Control Systems (“In the Weeds” Warning Applies)
2. 10 Techniques Management Should Adopt
3. Balancing Customer Service and Satisfaction
Find more Software Business information OMS SafeHarbor.
Monday Morning Head-Start [Software Business Links]
Software related links to start your week!
3. DKIM and signing emails to stop spam
Find more Software Business information OMS SafeHarbor.
Download Completion Rate Data – What CDNs don’t show you
Earlier this year we said we’d publish a view of this data, and here it is.
The plot below tells you this; the bigger your files, the harder it is to deliver them (and a little bit more, if you read further on).

This is a pseudo-random sampling of a few thousand downloads by users around the world. Piecing together logs and receipts from different networks and different download clients is not easy. But, the “complete” data is very tight. Any oversight is in the downloads that were grouped into the “incomplete” bucket. That is to say, this is the worst-case view of completion rate; its likely, that more attempts completed than we report here. At OMS we expand our reporting to include the concept of a completed “transaction” which may include many files, and/or multiple attempts, as our focus in on completing the business transaction, of which a single file download is only a component.
One of the first questions asked in any discussion of content delivery or software distribution is about completion rates. Since many CDNs tout their ability to improve download completion rates, thats not unreasonable to expect. My issue has always been the “versus what?” benchmark question that rarely gets addressed in that same discussion. Maybe this data point can be a reference for you.
Why CDNs won’t show you this, and why we almost didn’t either
After weeks of trying to figure out the best way to present this data, we finally settle on a very simplest format. But, we came close to not exposing this info at all.
The risk in this little plot?
1. You may not accept that 90/100 first attempts to download a CD image is pretty-good.
2. You may not accept that we/they have almost no control over those 10 incomplete first attempts. For example; how many out of those 100 people stopped their download on purpose, or lost the wireless connection from their neighbors apartment?
The ability to reliably connect and download from the Internet has reached the same level of expectation has getting a dial-tone when you pick up the phone. For most of us that means, always there, always fast. Even when you do get an error in your normal use of a browser, its likely that you just refresh or try again without even thinking about it. But, if you start a download for a file you paid for and really need, and something happens, look out! (We talked about the Top 10 reasons downloads fail in an earlier blog post).
That’s why companies that talk about downloads don’t like to talk about this data, it confronts your expectations with messy reality.
We hope this was helpful. Comments welcome.
Find more Software Business information OMS SafeHarbor.
Monday Morning Head-Start [Software Business Links]
Software related links to start your week!
1. Is music licensing a lesson for software?
2. Release management or build management in a nutshell. A fantastic explanation.
Find more Software Business information OMS SafeHarbor.
