Azure-docs: Application Gateway v2 Pricing
Please clarify the pricing model for the v2 Application Gateway. In the Azure Portal I can set Capacity type, Minimum instances, and Maximum instances.
In this article there are references to Fixed price and Capacity Unit price. Can a better definition of these be provided? Which of these corresponds to the minimum and maximum instances set in the Portal? I'm thinking Capacity Unit? But then the pricing example here is confusing me.
"Fixed price = 744(hours) * $0.20 = $148.8
Capacity units = 744 (hours) * 10 capacity unit per instance * 5 instances * $0.008 per capacity unit hour = $297.6"
Is Fixed price per Application Gateway resource in Azure?
What is the pricing per instance?
Where do you set capacity units per instance, or is this automatic? If it's automatic, what is the need for having multiple instances?
Document Details
⚠ Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.
- ID: faa5b8a2-11a1-6a2d-07cb-ffcf39643ed4
- Version Independent ID: 298a134e-fb27-f16d-bff2-d43eb0f766fd
- Content: Autoscaling and Zone-redundant Application Gateway v2
- Content Source: articles/application-gateway/application-gateway-autoscaling-zone-redundant.md
- Service: application-gateway
- GitHub Login: @vhorne
- Microsoft Alias: victorh
JoeRistine
All 27 comments
@JoeRistine Thanks for the feedback!
Is Fixed price per Application Gateway resource in Azure?
The fixed price is independent of scale count, and is a flat cost associated to an App Gateway. If you have 2 App Gateway V2s running for a month with different instance counts, the fixed price will be the same.
What is the pricing per instance?
The Capacity Unit price is where the variable pricing comes in, and where the instance price is calculated. Capacity units per instance is usually at 10 per instance, but can change depending upon how you are utilizing the app gateway. Because the Capacity Units are partially dependent on the load, it can be difficult to estimate an exact price for an app gateway until you know how App Gateway will handle your applications load.
Where do you set capacity units per instance, or is this automatic? If it's automatic, what is the need for having multiple instances?
It is possible to both Set the number of instances, as well as have App Gateway automatically scale the number of instances. This doc is designed to cover both scenarios.
To set this number, you can use the "Configuration" area of the portal:
Please let me know if you have any other questions!
TravisCragg-MSFT
on 19 Jun 2019
@JoeRistine We will now proceed to close this thread. If there are further questions regarding this matter, please tag me in your reply. We will gladly continue the discussion and we will reopen the issue.
TravisCragg-MSFT
on 25 Jun 2019
@TravisCragg-MSFT Now that billing has started for Application Gateway v2, I find that this article does not explain the pricing model well at all. The article gives the impression that capacity unit billing is consumption based, which is partly true. It fails to mention that 10 capacity units is the minimum amount that is billed per instance. Also, the minimum instance count allowed via the Azure portal is 2 (while this can be set to 1 via PowerShell - this might be helpful to include in this article for non-production environments).
The minimum price for an Application Gateway WAF_v2 instance created in the Portal would be as follows (East US pricing).
Fixed price - $0.36 per gateway-hour * 730 hours = $262.80
Consumption price - Minimum of 2 instances allowed by the Portal * minimum of 10 capacity units per instance - $0.0144 per capacity unit-hour * 20 capacity units * 730 hours = $210.24
Total minimum monthly cost = $473.04
One non-production environment I've implemented averages 0.42 Current Capacity Units since billing has started. It is still billed at 20 capacity units. Consumption can be slightly reduced by $105.12/month by setting the instance count to 1 via PowerShell.
Both Example 2 and Example 3 seem incorrect to me. How is it possible to get 4 capacity units in Example 2 or 8 capacity units in Example 3? Even at the bare minimum of 1 instance, minimum capacity units would be 10.
Example 2
An Application Gateway standard_v2 is provisioned for a month and during this time it receives 25 new SSL connections/sec, average of 8.88-Mbps data transfer. Assuming connections are short lived, your price would be:
Fixed price = 744(hours) * $0.20 = $148.8
Capacity unit price = 744(hours) * Max (25/50 compute unit for connections/sec, 8.88/2.22 capacity unit for throughput) * $0.008 = 744 * 4 * 0.008 = $23.81
Total price = $148.8+23.81 = $172.61
Example 3
An Application Gateway WAF_v2 is provisioned for a month. During this time, it receives 25 new SSL connections/sec, average of 8.88-Mbps data transfer and does 80 request per second. Assuming connections are short lived, and that compute unit calculation for the application supports 10 RPS per compute unit, your price would be:
Fixed price = 744(hours) * $0.36 = $267.84
Capacity unit price = 744(hours) * Max (compute unit Max(25/50 for connections/sec, 80/10 WAF RPS), 8.88/2.22 capacity unit for throughput) * $0.0144 = 744 * 8 * 0.0144 = $85.71
Total price = $267.84 + $85.71 = $353.55
Either the examples provided are incorrect or Azure billing is incorrect. The consumption details for the non-production instance I mentioned are below. Capacity units used for this instance averaged 0.42 over the time period below, but is billed at 20 capacity units.
$ 0.86 | Application Gateway WAF v2 - Capacity Units - US East | 60.00 | 7/3/2019 0:00 | 7/3/2019 23:59
$ 6.91 | Application Gateway WAF v2 - Capacity Units - US East | 480.00 | 7/4/2019 0:00 | 7/4/2019 23:59
$ 6.91 | Application Gateway WAF v2 - Capacity Units - US East | 480.00 | 7/5/2019 0:00 | 7/5/2019 23:59
$ 6.91 | Application Gateway WAF v2 - Capacity Units - US East | 480.00 | 7/6/2019 0:00 | 7/6/2019 23:59
$ 6.91 | Application Gateway WAF v2 - Capacity Units - US East | 480.00 | 7/7/2019 0:00 | 7/7/2019 23:59
$ 1.08 | Application Gateway WAF v2 - Fixed Cost - US East | 3.00 | 7/3/2019 0:00 | 7/3/2019 23:59
$ 8.64 | Application Gateway WAF v2 - Fixed Cost - US East | 24.00 | 7/4/2019 0:00 | 7/4/2019 23:59
$ 8.64 | Application Gateway WAF v2 - Fixed Cost - US East | 24.00 | 7/5/2019 0:00 | 7/5/2019 23:59
$ 8.64 | Application Gateway WAF v2 - Fixed Cost - US East | 24.00 | 7/6/2019 0:00 | 7/6/2019 23:59
$ 8.64 | Application Gateway WAF v2 - Fixed Cost - US East | 24.00 | 7/7/2019 0:00 | 7/7/2019 23:59
$ 0.00 | Bandwidth - Data Transfer Out - Zone 1 | 0.01 | 7/3/2019 0:00 | 7/3/2019 23:59
$ 0.00 | Bandwidth - Data Transfer Out - Zone 1 | 0.07 | 7/4/2019 0:00 | 7/4/2019 23:59
$ 0.01 | Bandwidth - Data Transfer Out - Zone 1 | 0.09 | 7/5/2019 0:00 | 7/5/2019 23:59
$ 0.00 | Bandwidth - Data Transfer Out - Zone 1 | 0.05 | 7/6/2019 0:00 | 7/6/2019 23:59
$ 0.00 | Bandwidth - Data Transfer Out - Zone 1 | 0.05 | 7/7/2019 0:00 | 7/7/2019 23:59
@JoeRistine If you are being charged different from how this doc describes, your best bet is to create a support request for them to validate the charges and rates you received. If you do not have a support plan, please let me know!
TravisCragg-MSFT
on 8 Jul 2019
I just posted about the same issue. I've had a Support Case open for two weeks, and getting no answers. But finally figured this out today. Here's what I posted:
The v2 pricing information is misleading. I've spent two weeks trying to figure out why our totally unused Preview V2's WAF's with Auto-Scaling Min 2 instances cost $482.11, and finally figured this out....
The pricing information above claims that CU's are variable cost, based on usage. Turns out they are not. A hard 10 CU's per 'instance' is charged, regardless of usage.
For Auto-Scaling deployments, a minimum of 2 instances is required, so 20 CU's / hour. For Manual, 1 instance is required (10 CU's/hour).
So the minimum price for WAF V2 is $482.11 with Auto-Scaling, and $374.98 for a 1-instance Manual scaling deployment. Similar calculations can be done for Standard_V2 ($267.84 with auto-scaling, $208.32 manual scaling).
The detail about the 10 Cu's / instance being a fixed cost is mentioned well below the Pricing section, in the "Autoscaling" section. The "Pricing" section should be revised to remove "variable cost" text such as:
Remove: "With the v2 SKU, the pricing model is driven by consumption and is no longer attached to instance counts or sizes."
Remove: "There are three dimensions to capacity unit - compute unit, persistent connections, and throughput." I had zero of all of these and got charged a full 20 CU's / instance.
Remove: Might as well just remove all of the pricing examples...because the CU cost is charged as 10x #'instances' that are running...
LenCardinal
on 27 Aug 2019
Bottom line, Examples 2 and 3 are wrong. The minimum a V2 standard would be billed at is $208.32 (manual mode, 1 instance), the minimum a V2 WAF would be $374.98 (manual mode, 1 instance). With Auto-scaling, the minimums are higher: $267.84 for Standard; $482.11 for WAF.
LenCardinal
on 27 Aug 2019
Agreed. The pricing model is misleading, and the 2nd and 3rd examples in the article aren't possible.
I'd also suggest that it be allowed for users to configure an AGW with 1 instance via the portal. Why are we only allowed to do so via manual configuration (PowerShell)?
JoeRistine
on 27 Aug 2019
Just opened a support case, we'll see what they say.
JoeRistine
on 27 Aug 2019
I have had a Support Case open for just short of two weeks. There was no movement. I just responded to them about these findings. Let me know what your support contacts says.
I would be curious to know if, in Auto-Scaling mode, when the instance count automatically scales from 2->3 instances, would the third instance be billed based on usage, or just as a flat 10 extra CU's like the first two instances. Another way to put it: When - if ever - does usage-based billing kick in?
Finally, Joe, can one create an Auto-Scale V2 with 1 instance using PowerShell? You can do that in the Portal for a Manual Scale AG..
LenCardinal
on 27 Aug 2019
I'm running two instances with auto-scaling up to 4 in a production environment. The max it's reached is about 5 CU's and we've been billed at 20 CU's since billing started in July. Would be interested to know when usage-based billing kicks in too.
I'm not able to configure my AGW to 1 instance via the portal, but am via PowerShell. I haven't tried auto-scaling at 1 minimum instance via PowerShell. The instance in the screenshot below is actually set to 1 instance, but resets to 2 in the portal.
JoeRistine
on 27 Aug 2019
When creating a new instance:
JoeRistine
on 27 Aug 2019
@joeRistine @lenCardinal can you post your Support Request # (SR#) here as well? I would love to see what they say.
I am also going to re-open this issue and assign it to the doc author to re-evaluate the pricing descriptions & examples and make sure they are correct.
TravisCragg-MSFT
on 4 Sep 2019
Sure, it's 119082724004940 @TravisCragg-MSFT
JoeRistine
on 4 Sep 2019
Mine was: 119081624004021
I did not receive any response from Support on this case. Eventually, I just figured out that V2 AG’s are billed at a minimum of 20 CU’s per hour (explained in my earlier posts). Support did not refute it.
As for the documentation (assuming I’m understanding this correctly, based on Joe’s, others’, and my experiences), the Pricing section of the doc should be updated to state: (a) a full 10 CU’s are billed for each of the “minimum instances” – regardless of usage – and (b) The minimum # instances per AG is 2. So a minimum of 20 CU’s (10x2) are going to be billed for every AG. The documentation examples would be updated accordingly.
Sent from Mailhttps://go.microsoft.com/fwlink/?LinkId=550986 for Windows 10
From: Joe Ristine notifications@github.com
Sent: Wednesday, September 4, 2019 4:00:49 PM
To: MicrosoftDocs/azure-docs azure-docs@noreply.github.com
Cc: Len Cardinal lencardinal@hotmail.com; Mention mention@noreply.github.com
Subject: Re: [MicrosoftDocs/azure-docs] Application Gateway v2 Pricing (#33601)
Sure, it's 119082724004940 @TravisCragg-MSFThttps://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FTravisCragg-MSFT&data=02%7C01%7C%7C57292da86eae4ac6120908d731729596%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637032240505948862&sdata=QqfOw6XK8ZKzGsWB8kuvL%2FW%2FjNUF4ZohwPpRa%2F4ONr4%3D&reserved=0
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHubhttps://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FMicrosoftDocs%2Fazure-docs%2Fissues%2F33601%3Femail_source%3Dnotifications%26email_token%3DADRDDTQKOVPH56C5RU2FF3DQIAHXDA5CNFSM4HZLGVQ2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD54ZHNY%23issuecomment-528061367&data=02%7C01%7C%7C57292da86eae4ac6120908d731729596%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637032240505958867&sdata=I3%2Bf2WmhDi9GtT%2BEWmoDZSU2iwTur8g%2FRe0t8xThfbc%3D&reserved=0, or mute the threadhttps://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FADRDDTV6NSFAZNF5DH5V73DQIAHXDANCNFSM4HZLGVQQ&data=02%7C01%7C%7C57292da86eae4ac6120908d731729596%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637032240505968872&sdata=LBFVuC8bp0VlvKsMHZPP8PrmOgCvt3sfa5ONd9475eI%3D&reserved=0.
LenCardinal
on 5 Sep 2019
Response from Microsoft:
Hello Joe,
Thank you for your patience.
I have investigated this further with the Product Group for Application Gateway and have a confirmation that Capacity Units will be billed as
10 * No of Instances. Hence there is a minimum charge of 10 CUs per instance.We have currently documented the above here:
https://docs.microsoft.com/en-us/azure/application-gateway/application-gateway-autoscaling-zone-redundant#pricingRelevant fragments:
Capacity Unit price - This is consumption-based cost that is charged
in addition to the fixed cost. Capacity unit charge is also computed hourly or partial hourly.
NoteEach instance can currently support approximately 10 capacity units
But, we have also acknowledged the fact that there is little ambiguity in the documentation and
we could improve the communication of this on our public facing documentation.My PG is discussing this internally and will move the change into production in next 2-3 weeks (Expected ETA).
Do let me know if you have any further questions or concerns on this case.
JoeRistine
on 9 Sep 2019
Thanks, Joe,
It looks to me like they are still missing the point. I guess let's wait and see.
LenCardinal
on 11 Sep 2019
Based on their docs here for V2_standards
https://azure.microsoft.com/en-us/pricing/details/application-gateway/
|||
|-|-|
|Fixed | $0.20 per gateway-hour |
|Capacity Unit1 | $0.008 per capacity unit-hour |
I spun up a new V2_standard AG and left it inactive for 2 days. Each day I had a $8.20 cost. Based on the info about the 20 CUs default per instance, this still doesn't make sense to me.
24h * $0.2 = $4.80
20CU * 24h * $0.008 = $3.84
That brings the total to $8.64, which is $0.44 higher than actual. Now I'm guessing because I'm using a corporate account, there might be discounts, but it's hard to actually plan what the cost would be w/o some detailed documentation on how AG pricing is calculated. It would really be nice to get some clarity around this.
datfinesoul
on 2 Oct 2019
(dat: I'm guessing that the price difference is a discount.)
It looks like there has been some progress here. The Azure Portal now lets you set minimum instances=0 for AutoScale deployments. And the detailed pricing explanation at https://docs.microsoft.com/en-us/azure/application-gateway/application-gateway-autoscaling-zone-redundant has been updated to mention the '0' option for minimum instances. Such a configuration should result in CU pricing based purely on usage.
Details: Back when I and others were carping about CU charges for no usage, the detailed pricing information at https://docs.microsoft.com/en-us/azure/application-gateway/application-gateway-autoscaling-zone-redundant#pricing did not mention a 'minimum CU's.'. Lower in the page, however, the section https://docs.microsoft.com/en-us/azure/application-gateway/application-gateway-autoscaling-zone-redundant#scaling-application-gateway-and-waf-v2 mentioned that 'reserved instances' were charged at 'full capacity' (10 CU's). 'Reserved' being the 'minimum instances' in an AutoScale deployment, or the fixed number of instances in a Manual deployment. On top of that, the Azure Portal forced a minimum of two instances* (at least with AutoScale), which resulted in 20 CU's per month minimum.
Now, the Azure Portal allows configuration of zero minimum instances for AutoScale. And the section https://docs.microsoft.com/en-us/azure/application-gateway/application-gateway-autoscaling-zone-redundant#scaling-application-gateway-and-waf-v2 mentions the 0 minimum instance option.**
So....now, you should be able to create an Auto-Scaling AG with Minimuminstances='0', and have CU's be calculated solely based on usage. That is great!
- A few months ago, @JoeRistine was able to use PowerShell to set min instances to '1' (though the Portal marked that as invalid). POSH or some other automation might have allowed AutoScale deployments with zero minimum instances...avoiding the minimum CU charges.
** Would be interesting & important to know if a deployment with '0' minimum instances potentially results in a noticeable 'warm-up' problem, when traffic starts up after a period of inactivity.
LenCardinal
on 2 Oct 2019
Shorter answer: the following should result in CU charges based purely on usage:
1) Use Auto-Scaling configurations
2) Set "Minimum autoscale instances" to '0':
LenCardinal
on 2 Oct 2019
@LenCardinal That's very helpful information, thank you! On an inactive AG that is a ~42% savings over a 2 instance manual, so that's great. Would still be nice to have all this clearly added to the price calculators @TravisCragg-MSFT. I'll run some tests with this setup today.
datfinesoul
on 2 Oct 2019
@JoeRistine @LenCardinal @datfinesoul Thank you all for your additional follow ups on this issue!
@vhorne currently has the action to update the pricing in the doc.
TravisCragg-MSFT
on 2 Oct 2019
Am I understanding the pricing correctly, in that the v2 standard sku fixed price is about 10 times more expensive than the price for a v1 small standard sku? That's crazy!
We are currently setting up our test and stage environment, and the plan was to use an application gateway in front of each of our four VM scale sets (each scale set is a completely different service, so we can't join them). That would mean a total of 8 application gateways in non-production. That's $1400 per month in fixed costs! That is a massive increase in costs compared to the v1 sku.
Is there really no v1 version suitable for development/testing environments, with virtually no traffic (3 person team, only traffic is when we actively test something)? We are willing to reduce the "power" of each application gateway, not the number of gateways used. And we don't want a vastly different setup compared to production (meaning: we want v1 in both prod and non-prod, or v2 in both).
batanun
on 28 Oct 2019
You can probably get away with 1 for non-production using VNet peering.
Sent from my iPhone
From: batanun notifications@github.com
Sent: Monday, October 28, 2019 12:45:55 PM
To: MicrosoftDocs/azure-docs azure-docs@noreply.github.com
Cc: Joe Ristine Joe@Ristine.me; Mention mention@noreply.github.com
Subject: Re: [MicrosoftDocs/azure-docs] Application Gateway v2 Pricing (#33601)
Am I understanding the pricing correctly, in that the v2 standard sku fixed price is about 10 times more expensive than the price for a v1 small standard sku? That's crazy!
We are currently setting up our test and stage environment, and the plan was to use an application gateway in front of each of our four VM scale sets (each scale set is a completely different service, so we can't join them). That would mean a total of 8 application gateways in non-production. That's $1400 per month in fixed costs! That is a massive increase in costs compared to the v1 sku.
Is there really no v1 version suitable for development/testing environments, with virtually no traffic (3 person team, only traffic is when we actively test something)? We are willing to reduce the "power" of each application gateway, not the number of gateways used. And we don't want a vastly different setup compared to production (meaning: we want v1 in both prod and non-prod, or v2 in both).
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHubhttps://github.com/MicrosoftDocs/azure-docs/issues/33601?email_source=notifications&email_token=AKYVNMNLXELIAR4ME7OOSJ3QQ4JMHA5CNFSM4HZLGVQ2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOECNSH5Y#issuecomment-547038199, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AKYVNMLFANCTJBX5OEOI4FTQQ4JMHANCNFSM4HZLGVQQ.
JoeRistine
on 28 Oct 2019
Am I understanding the pricing correctly, in that the v2 standard sku fixed price is about 10 times more expensive than the price for a v1 small standard sku? That's crazy!
We are currently setting up our test and stage environment, and the plan was to use an application gateway in front of each of our four VM scale sets (each scale set is a completely different service, so we can't join them). That would mean a total of 8 application gateways in non-production. That's $1400 per month in fixed costs! That is a massive increase in costs compared to the v1 sku.
Is there really no v1 version suitable for development/testing environments, with virtually no traffic (3 person team, only traffic is when we actively test something)? We are willing to reduce the "power" of each application gateway, not the number of gateways used. And we don't want a vastly different setup compared to production (meaning: we want v1 in both prod and non-prod, or v2 in both).
Yes - I just found this out the hard way. I selected V2 due to the Azure documentation stating that V1 was being phased out not realising it is more than 10X the cost. Not sure if that changes, but I am now trying to get V1 APPGW to work for my solution. There is clearly a difference between V2 and V1 because having a lot of issues with the exact same configuration.
graham83
on 13 Nov 2019
vhorne
on 20 Feb 2020
Hi all,
I'm confused about the v2 pricing model. I understand there is a fixed price for using an APGW v2 and a price related to CUs usage. It does not mention how APGW distributed on Availability Zones are priced. So if I deploy an APGW on 3 AZ, will I be charge for 3x the fixed price? Or if I have 10 CUs configured, will I be charge 3x 10CUs price? Or anything else?
Thanks.
crobin1
on 21 Jul 2020
Hi @TravisCragg-MSFT I'm looking at AppGW and find it very confusing as well. Here's an example:
Pricing page https://azure.microsoft.com/en-us/pricing/details/application-gateway/
It mentions _Capacity Unit_, with a per-capacity unit hour charge.Pricing on MSDN docs https://docs.microsoft.com/en-au/azure/application-gateway/application-gateway-autoscaling-zone-redundant mentions _Capacity Unit price_
When in the azure portal, it talks about scaling using _Scale units_ which is not defined anywhere
Confirming a "scale unit" is 1x VM and has 10x Capacity Units?
Scaling from 2 to 3 _Scale units_ shows an increase from _20 Capacity Units_ to _23.33 Capacity Units_, hows that work? It would make sense if this was 20 -> 30 (1 scale unit, which is really just a VM, is 10 CUs) so its not clear to me the relation between Scale unit and Capacity units (which supports a certain workload + cost).
- The sentence _Please note that even if you configure autoscaling with zero minimum instances the service will still be highly available, which is always included with the fixed price_ contradicts _However, creating a new instance can take some time (around six or seven minutes). Hence, if you do not want to cope with this downtime you can configure a minimum instance count of 2, ideally with Availability Zone support_ --- I'm assuming the "fixed cost" is really 1 VM instance, so if it goes down AppGW is offline until a new instance is made, hence the suggestion for 2 scale units/20 CUs? https://docs.microsoft.com/en-au/azure/application-gateway/application-gateway-autoscaling-zone-redundant
Plasma
on 17 Aug 2020
Related issues
jebeld17
·
3Comments
spottedmahn
·
3Comments
behnam89
·
3Comments
varma31
·
3Comments
Agazoth
·
3Comments
Most helpful comment
@TravisCragg-MSFT Now that billing has started for Application Gateway v2, I find that this article does not explain the pricing model well at all. The article gives the impression that capacity unit billing is consumption based, which is partly true. It fails to mention that 10 capacity units is the minimum amount that is billed per instance. Also, the minimum instance count allowed via the Azure portal is 2 (while this can be set to 1 via PowerShell - this might be helpful to include in this article for non-production environments).
The minimum price for an Application Gateway WAF_v2 instance created in the Portal would be as follows (East US pricing).
Fixed price - $0.36 per gateway-hour * 730 hours = $262.80
Consumption price - Minimum of 2 instances allowed by the Portal * minimum of 10 capacity units per instance - $0.0144 per capacity unit-hour * 20 capacity units * 730 hours = $210.24
Total minimum monthly cost = $473.04
One non-production environment I've implemented averages 0.42 Current Capacity Units since billing has started. It is still billed at 20 capacity units. Consumption can be slightly reduced by $105.12/month by setting the instance count to 1 via PowerShell.
Both Example 2 and Example 3 seem incorrect to me. How is it possible to get 4 capacity units in Example 2 or 8 capacity units in Example 3? Even at the bare minimum of 1 instance, minimum capacity units would be 10.
Either the examples provided are incorrect or Azure billing is incorrect. The consumption details for the non-production instance I mentioned are below. Capacity units used for this instance averaged 0.42 over the time period below, but is billed at 20 capacity units.