scalability and redundancy in Azure

Lior Tomer 60

Hi, I have Cloud service (extended support) The instance I have is Standard_A4_v2 I want to ensure two things scalability and redundancy : Regarding scalability, I have the option to scale and set a rule that once the rule occurs, the scale will be performed automatically. Is this true? If so, how long does it take for scalability to happen? If, for example, I add another instance that will be automatically added as it were, will the price be doubled? Regarding redundancy, how exactly I do this if by chance the service goes down ? there is a option after i buy the services to take care for redundancy ?

kobulloc-MSFT 25,736 Reputation points Microsoft Employee

2023-10-30T16:41:12.1733333+00:00
Hello, @Lior Tomer !

Scalability: Azure Cloud Services extended support uses Azure Autoscale. Once the service detects the conditions for scaling, resource deployment begins. The two variables here are what time grain is used (default seems to be using metrics aggregated every minute) and how long the resources take to deploy. I believe you are charged for active resources however I'd need to look closer into pricing.

Redundancy: There doesn't appear to be much in the documentation that explicitly describes how regional resiliency is achieved however usually this means duplicate sets of resources. I'll take a closer look at this.

Lior Tomer 60

Hi, is there anything new? There should be some explanation for the situation of an instance falling and the definition of another instance in another availability zone. If I set up two instances that will be active, will this give me a solution for both

kobulloc-MSFT 25,736 Reputation points Microsoft Employee

2023-10-31T23:20:26.3133333+00:00

Hello, @Lior Tomer !

I've reached out to the content author for additional clarification and will give you an update as soon as I know more.
kobulloc-MSFT 25,736 Reputation points Microsoft Employee

2023-11-03T17:38:08.3666667+00:00

Hello, @Lior Tomer !

I hope this has been helpful! Your feedback is important so please take a moment to accept answers.

If you still have questions, please let us know what is needed in the comments so the question can be answered. Thank you for helping to improve Microsoft Q&A!

1 answer

kobulloc-MSFT 25,736 Reputation points Microsoft Employee

2023-11-03T04:59:02.2966667+00:00
Hello, @Lior Tomer !

Can you tell me more about scaling and redundancy with Azure Cloud Services (extended support)?

I had a conversation with the Cloud Services team and was able to get some more clarification on both scaling and redundancy:

Scaling: Once the rule is initiated and scaling starts, there's no set time for when scaling will complete as the autoscaling operation is dependent on multiple factors (primarily having available capacity within that cluster, zone, and region). We are looking at some scaling operations telemetry to provide a better time estimate and I'll update this when that is provided. Cloud services pricing is based on instances, so additional instances would result in additional costs.

Redundancy: Legacy architecture was ASM based and had cross-regional dependencies which meant that an outage in one region could affect resources in other regions. Cloud Services (extended support) uses a modern, ARM based stack offering which means that deployments and all associated resources are present within a single region. This improves resiliency as there are reduced points of failure.
If you wanted to plan for redundancy in other regions in case there is an outage in your primary region, this would need to be deployed prior to the outage and the additional instances would result in additional costs.

I hope this has been helpful! Your feedback is important so please take a moment to accept answers.

If you still have questions, please let us know what is needed in the comments so the question can be answered. Thank you for helping to improve Microsoft Q&A!
Please sign in to rate this answer.
Lior Tomer 60 Reputation points

2023-11-03T20:21:59.35+00:00

I want to plan redundancy in the same region how can i do it ? for example creating more instance in different availability zone will promise redundancy this is the way to ensure the redundancy ?

kobulloc-MSFT 25,736 Reputation points Microsoft Employee

2023-11-03T20:31:28.8766667+00:00

Hello, @Lior Tomer !

In theory, a failure that does not affect the rest of the region (example: a VM disk that reaches end of life) would trigger autoscaling to ensure that the targets are met. You could deploy additional resources in a different availability zone (and this may not be a terrible idea if you are going to split the resources anyway) but I'm not sure how many scenarios that would safeguard against that are not already covered by autoscaling and regional resiliency (perhaps a power outage on a dependent resource in one availability zone).

Lior Tomer 60 Reputation points

2023-11-04T10:15:01.1166667+00:00

So basically, if I don't plan to deploy additional resources, wouldn't you suggest that I deal with Redundancy ? And if so I didn't exactly understand what I should do in action.

kobulloc-MSFT 25,736 Reputation points Microsoft Employee

2023-11-04T17:56:23.4333333+00:00

There is built in local redundancy for a variety of issues that is provided with autoscaling. If there was a power outage on a rack, if hardware reached end of life, or if a network cable went out, autoscaling would recover by adjusting instances to match the target within that region.

If there was a catastrophic failure that impacted an entire region, which are exceedingly rare and usually short lived, then you would need to plan for that in advance in this case with additional resources in another region. One strategy in this case might be to split the workload between regions and scale up in the event of an outage.

Lior Tomer 60 Reputation points

2023-11-05T07:14:05.1333333+00:00

But the autoscaling works according to a rule, is there a rule that allows switching machines in the cases you mentioned: power outage on a rack, if hardware reached end of life, or if a network cable went out. or is done automatically without rule ?

kobulloc-MSFT 25,736 Reputation points Microsoft Employee

2023-11-13T21:29:28.3366667+00:00

You would need to use an autoscaling rule for those scenarios. The documentation describes two ways of approaching this:

https://video2.skills-academy.com/en-us/azure/cloud-services-extended-support/configure-scaling

You can ensure that there is always a specific instance count (which automatically maintains the number of instances you want) or a specific metric threshold (using CPU as an example, if an instance went down it would but additional strain on the remaining instances, triggering autoscale to compensate until the scale condition was met).
Sign in to comment

Share via

scalability and redundancy in Azure

1 answer