CODEWITHSUNDEEP
azure
terfdaq
terfdaq

Reputation: 100

Downtime experienced swapping instances of different sizes

In production I am running 3 small instances of my web role. In staging I have 3 extra small instances of that web role. When I swapped them out my site experienced about 2 minutes of downtime which I wasn't expecting.

According to the article Update, Upgrade and VIP-Swap for Windows Azure Service–What are the Differences? this shouldn't have occurred. Why would downtime be experienced during the swap and is there a way to swap roles of different instance sizes without experiencing downtime?

Upvotes: 1

Views: 918

Answers (2)

Igorek
Igorek

Reputation: 15850

Adding to what Michael has mentioned - there shouldn't have been any downtime. However, if you're swapping into xtra-small instances that are "cold" and just have been deployed into, IIS will need to recompile the sites and bring them online. Xtra-small instances are notoriously slow.

Thus: Your downtime could have been caused by the compilation/caching/app-start/etc.

Upvotes: 2

mcollier
mcollier

Reputation: 3719

While you can change the VM sizes now as part of an in-place update, you should expect some instance downtime. The Windows Azure fabric controller will walk the upgrade domains, taking 1 upgrade domain down to change the VM sizes, and then moving on to the next. If you have 2 instances, you shouldn't have noticed an entire outage (just 1 machine going down at a time).

You can see some further details at http://msdn.microsoft.com/en-us/library/windowsazure/hh472157.aspx.

Upvotes: 3

Related Questions