CODEWITHSUNDEEP
google-app-enginegoogle-cloud-platform
dendog
dendog

Reputation: 3338

Latency between services in google app engine

I would like to break out my app hosted on Google app engine into various services, and use other runtimes.

The purpose is for performance, the information I cannot find is what is the latency between different services within a single google app engine application. To clarify I am interested in the communication latency as using HTTP.

Also what would be the latency between GAE and another Google service such as functions or Run?

Upvotes: 2

Views: 760

Answers (2)

Travis Webb
Travis Webb

Reputation: 15018

You can expect roundtrip network latency within a particular region to be less than 1ms 95% of the time. Roundtrip network latency within a single datacenter can be < 500μs. Inter-region network latency depends on their geographic distance, and whether you're using Standard or Premium Tier networking.

The only way to measure your latency in practice is to measure your latency in practice. Tools such as Trace can help you profile your App Engine applications for latency.

Network latency is like gravity: it serves as a baseline for what latencies are theoretically possible between system components, but actual latency observed in practice will depend on your application, where your resources are deployed, and which GCP services you consume. For example, network latency does not include the cost of serializing/deserializing HTTP requests coming off the wire, or the execution time of your code.

In my gravity metaphor, think of this as "wind resistance". If you asked "how fast does a box fall?", knowing the gravitational constant will only tell me the maximum speed it can fall in the absence of all normal worldly constraints. In order to answer your question precisely, I'd need to know the size, shape, weight, and texture of the box, the air temperature and speed/direction, the height you're dropping it from, as well as what planet you're on and what your atmosphere is made out of. So, a simple question like this often has no simple answer.

Upvotes: 7

Harif Velarde
Harif Velarde

Reputation: 753

The latency you can face between different instances depends from different situations, for example, the zone where the app engine instances are located or the task they are executing, for example, is not the same time for one 'hello world' application than other more complex. Also, I recommend you take a look on the official documentation related how the requests are handled by App Engine (in this case standard environment) [1], as per is mentioned on this document, the request duration for App Engine Standard is maximum 60 seconds. You can also check this post as a reference [2].

[1] https://cloud.google.com/appengine/docs/standard/python/how-requests-are-handled

[2] https://cloud.google.com/blog/products/gcp/best-practices-for-app-engine-startup-time-google-cloud-performance-atlas

Upvotes: 0

Related Questions