Monthly Archives: March 2014

Not Your Father’s Cloud: Microsoft Azure Traffic Manager Explained

If you need to understand the format of this post, take a look at my introduction to the series.

Like A Boss

Traffic Manager is the grand-pappy of load balancing in Microsoft Azure (though strictly speaking it is not actually a load balancing solution).  Unlike the transparent load balancing built into Cloud Services which provide intra-Region (i.e. Japan) balancing, Traffic Manager allows you to route traffic across Regions (i.e. Japan and US) depending on service availability.  Traffic Manager also has least-latency routing support so that you can deploy your application in multiple Regions and provide great user experiences by routing a user to their nearest instance.

Goes Well With

  • Applications with a global deployment footprint where consumers can be routed to a nearby instance for better performance.
  • Applications that require load balancing beyond simple Round Robin.
  • Would be familiar to advanced load balancing used in many hosting facilities.

Open Other End

  • Simple intra-Region services that can leverage standard Cloud Service load balancing don’t really need to use Traffic Manager.
  • Split services where you are running a hybrid setup of some privately-hosted and some cloud-hosted endpoints aren’t supported.

Contents May Be Hot

  • Traffic Manager utilises only HTTP probes for health checks of balanced applications.  If you need UDP it won’t work.
  • Your application has to return a 200 HTTP response code to be considered healthy.  Anything else will not work.

Don’t Take My Word For It

Tagged , , ,

Not Your Father’s Cloud: Microsoft Azure Service Bus Explained

If you need to understand the format of this post, take a look at my introduction to the series.

Like A Boss

Microsoft Azure Service Bus is a PaaS offering that allows application developers to connect components of larger systems to one another in a loosely coupled fashion. Loose coupling facilitates scalability of systems through asynchronous messaging (send a message, don’t wait, receive a response at a later time) and Service Bus’ messaging architecture also enables scalability via publish/subscribe messaging (send one message, lots of subscribers can receive it).  Through use of Service Bus Relay it is also possible to integrate on-premise systems with Microsoft Azure without the need to create Virtual Private Networking (VPN) connections.  The most recent addition, Notification Hubs, allows push notifications to large numbers of mobiles devices to be sent by way of a single call.  As with many other Microsoft Azure services there are SDKs in a range of languages including .Net, Java, Node.js, PHP, Python and Ruby.

Goes Well With

  • Custom-built Cloud applications that require inter-component messaging.
  • Integration of on-premise applications with Cloud applications (Service Bus Relay).
  • Mobile solutions that could target large subscriber bases (you know, like the Olympics!):

Open Other End

  • Legacy systems that do not support an asynchronous integration pattern.
  • Off-the-shelf software that has an API that cannot be modified or proxied to support Service Bus Relay.
  • If you’re likely to have a queue exceeding 5GB you will need to use Queue Storage Service instead.

Contents May Be Hot

  • Quotas on some aspects of Service Bus may mean you need to solve larger scale-out scenarios with multiple queues, topics or subscriptions.
  • Make sure you design your solutions well.  Null messages to empty queues or subscriptions still incur a cost.

Don’t Take My Word For It

Let the man behind the service, Clemens Vasters, give you an introduction to Service Bus.

(originally from: http://www.windowsazure.com/en-us/documentation/services/service-bus/)

Tagged , , , , ,