Kirk Evans Blog

.NET From a Markup Perspective

The PDC That Took Weeks to Digest

The PDC That Took Weeks to Digest

  • Comments 3

I was there:  PDC 2008, when Windows Azure was announced.  Reactions in the crowd, admittedly, ranged from "great, not even the keynote presenter can pronounce it consistently" to "I expected more pizzazz from the Don and Chris keynote."

It's hard to picture, at first, how any of this cloud stuff is relevant.  Do I really want SQL in a cloud?  And what was that service bus thing that I didn't quite grok?

Let it stew.

Reactions in IM and personal conversations recently have ranged from "holy smokes, Microsoft just shut up everyone who thought they were irrelevant" to "this is the absolute coolest thing Microsoft has done since .NET". 

I can imagine that in coming months reactions will be much more like "d00d!  Holy cow, this is awesum.  Can you get me a developer key?"

Let it bubble.

I had pictures of my kids on my USB external hard drives.  One of them failed, and I lost multiple years of pictures of my kids due to a failed hard drive.  A hard lesson to learn, don't rely on a hard drive for long-term durable storage, but only after the fact.  What happens in a fire, one bad strike of lightning?  Uploading those pictures onto Flickr or my Live Space is now a no-brainer... I already keep some of my data in the cloud, and nothing's more personally identifiable than pictures.

I don't send hand-written letters anymore, I send direct messages on Twitter or message on Facebook.  I post pictures of my family, have been tagged in lots of pictures (some more embarrassing than others), and have come to learn that I have lots of personally identifiable information in the cloud. It's about trust, who I trust to keep that information. 

Let it simmer.

Truthfully, as a consumer, I really don't care how my favorite web sites are implemented.  There's not one place that holds all of my data, there's lots of different places.  I control where that data lives.  I don't care if it sits in a data center in Texas or Iowa or wherever.  I don't care if Twitter actually fixes its problems and starts using SQL Data Services, I just care that my status updates appear.  I don't care if Facebook uses Microsoft's internet service bus to deliver messages to my friends, as long as the UI looks nice and is usable.

I don't quite grok all of the potential of Live Mesh, but I know I am getting a hell of a lot of value out of the current implementation.  I push photos from my trip into a folder on my machine while traveling in Washington, and next thing you know my wife is seeing updated images in her Vista Sidebar picture viewer gadget while she is at home in Texas. 

As a consumer, I am faced with going to Fry's and buying a rack with redundant storage for my home to protect my pictures of my kids.  Or, I can upload them to Facebook, Live, and Flickr.  I can host on premises and pay an arseload for the hassle of failing hardware, or I can host in the cloud.  Live Mesh does a lot of that for me these days, and I know that there will be other parts of the platform integrated in ways I cannot conceive while typing right now.

As a business owner, I would quickly come to the same realization.  I can host in my data center and pay for the hardware, racks, power, cooling, backups, and staff to manage it all (including wages and benefits), or I can push my non-real-time, non-mission-critical-yet-important-to-my-operations data to the cloud.  I can build an integrated enterprise that will, inevitably, be integrated into the cloud.  Or I can piecemeal something together that ends up costing more.  I am betting that enough details were announced at PDC that any casual observer can see what's coming.

As a developer, I can make the case to my boss that the solution I am proposing will require 5 new servers requiring terrabytes of storage projected over the next few years, and acknowledge that virtualizing the solution will have diminished returns because the solution is compute bound.  Or, I can provision to the cloud and then dynamically add more compute nodes to accomodate load, then decrease nodes during off-peak times.  I can propose a solution where we pay for what we need instead of paying for a rack of servers averaging 10% CPU utilization.

Let it brew.  These things take time.

Page 1 of 1 (3 items)
Leave a Comment
  • Please add 8 and 3 and type the answer here:
  • Post
Translate This Page
Search
Archive
Archives