cyclenerd / google-cloud-compute-machine-types Goto Github PK

Is your feature request related to a problem? Please describe.

I want to compare E2 vs N2D preemptible instances across a few regions using your tool and I can't because it doesn't include preemptible prices at all.

Describe the solution you'd like

Add a column within the "Costs" group with the preemptible price.

Describe alternatives you've considered

Using an alternative solution - https://gcpinstances.doit-intl.com - but it has its own issue with no guarantee of data being up to date (see doitintl/gcpinstances.info#34 for more info).

It would be great if on the web page for a particular machine type there was an option to sort by spot instance price (this sort order is often suprisingly different!).

Describe the bug
On page https://gcloud-compute.com/regions.html the column headers for T2D and T2A are mixed up.
Order of column headers is T2A then T2D. But order of data columns is T2D then T2A.

# http -p b https://gcloud-compute.com/regions.html | grep -E 'T2(D|A)' | head -4
<th scope="col">T2A</th>
<th scope="col">T2D</th>
<td title="Instance Picker with filter on machine type series T2D in region asia-east1">
<td class="table-danger" title="Instance Picker with filter on machine type series T2A in region asia-east1">

Is your feature request related to a problem? Please describe.

Many Google Cloud VM shapes require a quantity of local SSD. For example, a2-ultragpu, a3, z3. The shape costs total appear to not include costs for local ssd amounts. The user may be unaware of this and interpret the data as being inclusive of local SSD costs.

Describe the solution you'd like
Machine shape description should be inclusive of any charges that relate to local VM hardware, including local SSD SKUs.

Describe alternatives you've considered
Add descriptive text indicating local SSD charges are required but not included (deep link to Storage page if needed.)

Additional context

An example is z3-highmem-88, where the 36TB of local ssd cannot be adjusted (mandatory for shape) and accounts for about 30% of the total shape cost ($9,490.01/month with SSD, vs $6,610.01/month without SSD being the cost the app shows.)

See also: Cyclenerd/google-cloud-pricing-cost-calculator#21

If official then delete from clean_up.sql

ERROR: Costs per hour missing for machine type 'm3-megamem-128' in region 'us-central1'.
ERROR: Costs per hour missing for machine type 'm3-megamem-128' in region 'us-west4'.
ERROR: Costs per hour missing for machine type 'm3-megamem-64' in region 'us-central1'.
ERROR: Costs per hour missing for machine type 'm3-megamem-64' in region 'us-west4'.
ERROR: Costs per hour missing for machine type 'm3-ultramem-128' in region 'us-central1'.
ERROR: Costs per hour missing for machine type 'm3-ultramem-128' in region 'us-west4'.
ERROR: Costs per hour missing for machine type 'm3-ultramem-32' in region 'us-central1'.
ERROR: Costs per hour missing for machine type 'm3-ultramem-32' in region 'us-west4'.
ERROR: Costs per hour missing for machine type 'm3-ultramem-64' in region 'us-central1'.
ERROR: Costs per hour missing for machine type 'm3-ultramem-64' in region 'us-west4'.

Today the Compute Engine API has returned machine types in europe-southwest1. The region will be the new location in Madrid. Region and Zone as well as Billing API do not know the region yet. I have therefore removed them manually. When the region is 100% ready it must be removed from instances/clean_up.sql.

Today the Compute Engine API has returned machine types in us-east5. The region will be the new location in Columbus. Region and Zone as well as Billing API do not know the region yet. I have therefore removed them manually. When the region is 100% ready it must be removed from instances/clean_up.sql.

The metrics already contain a value "benchmark". We found a metric "benchmark points per $" really useful while evaluating which instance type to pick.

Do you think it would be easy to add such a (virtual) metric?

Thank you very much for your awesome project! It really makes our life easier! 😍

Describe the bug
c2d family show 32 Gbps network limit for all shapes, but the 2 and 4 vCPU shapes should be 10, and the 8 vCPU shape should be 16. See https://cloud.google.com/compute/docs/compute-optimized-machines#c2d_machine_types

See also: Cyclenerd/google-cloud-pricing-cost-calculator#31

If official then delete from clean_up.sql.

Include carbon data in region overview and Instance Picker.

Show 🍃 Low CO2 icon.

Source:
https://cloud.google.com/sustainability/region-carbon#data

Is your feature request related to a problem? Please describe.
For some use cases it's good to know exact cluster geoposition. Geoposition of region (==city center) is not good for that, because it could be very far from cluster geopisition. For example,
50.1109, 8.6821 - frankfurt center
50.122500, 8.974167 - cluster position
Error is about 21km.

Describe the solution you'd like
Provide additionaly cluster geoposition
https://en.wikipedia.org/wiki/Google_data_centers

Describe alternatives you've considered

Additional context

Hi!

Thanks for a very promising tool! I was using https://gcpinstances.doit-intl.com till now but your solution seems to be even better!

However you seem to be missing C2D benchmarks results.

(I have just added them to gcpinstances.doit-intl.com in doitintl/gcpinstances.info#29 as they were missing it too, so perhaps you can reuse that PR to add it to your app. Or just take it from the source, of course. :))

Today the Compute Engine API has returned machine types in europe-west8. The region will be the new location in Milan. Region and Zone as well as Billing API do not know the region yet. I have therefore removed them manually. When the region is 100% ready it must be removed from instances/clean_up.sql.

When the region us-south1 (Dallas) is 100% ready it must be removed from instances/clean_up.sql.

Today the Compute Engine API has returned machine types in europe-west9. The region will be the new location in Paris. Region and Zone as well as Billing API do not know the region yet. I have therefore removed them manually. When the region is 100% ready it must be removed from instances/clean_up.sql.

Describe the bug
In different places different measurement units are used to describe memory, but the same amount are used.

To Reproduce

1. Look at here - GB. https://gcloud-compute.com/instances.html@
2. Dowload csv, find column called memoryGiB

What is true? GB or GiB?

Expected behavior
Consistent measurement unit