search

MicrosoftDocs / azure-docs

Open source documentation of Microsoft Azure
https://docs.microsoft.com/azure
Creative Commons Attribution 4.0 International
10.3k stars 21.48k forks source link

How many tps with the recommended settings? #44000

Closed ogarules closed 4 years ago

ogarules commented 4 years ago

How many tps with the recommended settings?

Document Details

⚠ Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.

AshokPeddakotla-MSFT commented 4 years ago

@ogarules Thanks for the feedback! We are currently investigating and will update you shortly.

ram-msft commented 4 years ago

@ogarules Could you please add more details about your use case that you are trying to solve. Containers do not cap transactions per second (TPS) and can be made to scale both up and out to handle demand if you provide the necessary hardware resources. You are billed for the calls that come into the group of containers holistically also these container endpoints can live “closer” to the associated business logic; lowering latency even further. Please follow the below document for features and benefits and supported pricing tier.

image https://docs.microsoft.com/en-us/azure/cognitive-services/cognitive-services-container-support#features-and-benefits

ogarules commented 4 years ago

@ogarules Containers do not cap transactions per second (TPS) and can be made to scale both up and out to handle demand if you provide the necessary hardware resources. Please follow the below document for features and benefits. https://docs.microsoft.com/en-us/azure/cognitive-services/cognitive-services-container-support#features-and-benefits

Hi

My question was more oriented to: with the recommended configuration how many concurrent analysis can be made? wich is use full so I can calculate the infraestructure necessary to acommodate my business case wich is directly relate to the budget I'll need, based on the concurrent analysis i'll be making, for example, the text analytics container page states the concurrent analysis that can be made per second, with the recommended settings, and I can use that number to calculate the size of the infraestructure I'll be needing for my kubernetes cluster with that, is it possible you can share that number for this containers?

IEvangelist commented 4 years ago

Hi @ogarules,

The Speech service is not measured by transactions per second (TPS), so it is not shown in the docs for Speech. This is intentional and is not changing. In terms of a recommended configuration, I suggest starting at the recommended resources instead of the minimum, then with your own benchmarking and testing - determine if more resources would help with throughput. Unfortunately, we're not able to help out much with this from the perspective of TPS. We are currently working on providing guidance for throughput capacity with the Speech service, and that will apply to this question.

please-close