Common Application Workloads and Scenarios for Microsoft Azure

Have you heard about "The Cloud"?

Just kidding… I’m pretty sure you have.  But have you thought about what you are going to do with the cloud?

When you have a physical infrastructure that you've been building and maintaining for years, it's not hard to see how you might not realize how cloud computing can help your business.  The cheapest storage and compute you can get is the stuff you already own, right?

Let's take a look at some common cloud computing patterns and scenarios where they might be useful.  With the end of life of Windows Server 2003, you might be taking a close look at the functions and applications in your data center.  Now is the time to modernize and mature the systems and tools you use to keep your company (and your skill set) working .

On and Off Workloads

These are workloads where you have resources (like storage and compute) sitting idle for long periods of time.  You made these purchases for specific reasons, for a planned growth of a file server, maybe for a development/test lab or perhaps to host "cold" VMs for disaster recovery.  Ultimately these resources go under used and eventually the hardware becomes outdated.  It becomes harder to justify the costs of keeping it up to date when it sits inactive for so long.  Or maybe you have opposite problem - you don't have the capital budget to build a lab or recovery site and end up using desktop grade equipment to test.

Growing Fast

Successful services need to be able to grow and scale.  When IT can't provision physical hardware fast enough, it becomes challenging to keep up with regular growth.  Do you expand your data center? Keep trying to reduce the footprint of each server?  Maybe it's a storage issue - it doesn't seem to matter how many policies you have in place, data grows.  Users of your systems like to save, save, save and demand larger mailboxes, home folders and databases.

Unpredictable (and Predictable!) Demand

Put those two "growing fast" and "on/off workload" problems together and it all comes down to supply vs. demand.  Unexpected peaks in service like the sudden popularity of your product or something less amazing, like a runaway job or process can cripple your infrastructure.  Even if you have predictable demands (like seasonal shopping spikes or month-end processes) it can be hard to balance the cost of wasted resources vs the cost of not meeting the demand when it comes. And even if you do "balance" it, the average usually means there is a lot of time when nobody is happy.

Enter the Cloud

With Azure, you only pay for what you use, when you use it, making it a viable option for on-demand lab environments, disaster recovery testing, batch workloads and scaling or bursting to the cloud when needed.

Look to the cloud as a solution for off-site backups, cloud-integrated storage and for pilot-to-production role outs.  I'm pretty sure I'm not the only one who piloted something on an old machine under my desk, only to find out a few months later that my "customers" (end users) liked it so much that it's now working to capacity and would be a serious pain point for many if it failed. 

By creating your own virtual network and connecting to your physical LAN with a site-to-site VPN, you've  created an ever expandable, turn-it-off, turn-it-on data center that doesn't require you to touch racks, tie cables or think about the ratings of air handlers or the weight of UPS gear.  Turn on point-to-site VPN and it can double as a disaster recovery site that workers can connect to remotely.

I'm not asking you to dump your data center.  The cheapest compute and storage you have IS the stuff that you already own.  But by considering each of your services and applications, you can decide which ones are better off in-house (and take full advantage of the resources you own) and which ones can thrive in the cloud.  

Want to learn more?  Make sure to follow the rest of the blog series and check out the Microsoft Virtual Academy for videos and lectures.  If you are just getting started with Windows Azure, I suggest the JumpStart for IaaS.  Also, this blog post is part of a greater series of posts, check out the full series at http://aka.ms/ModernCloud