Why increase latency?

We are scaling the number of users in our GAE java application. We now have three triples of active users that we had a few days ago.

We noticed that the average latency represented by the GAE bar has almost doubled. There were no code changes.

What can cause latency for each request to increase the number of users? Presumably, in a well-designed app, appengine will scale up the number of instances so that our latency is kept to a minimum. I have not changed any of the set wait delay and similar settings.

What are the most common design flaws when implementing an appengine application that can cause latency as usage rises?

Data Warehouse Conflict? The data we store is in separate entity groups for each user, so this seems like a low probability .....


source to share

2 answers

You need to provide additional information about your application, for example, write to the data store. If many clients write to the datastore, you will create a plug if you are using a single object.

What you need to do is measure each part for how long it takes and compare the results between the two loads.



There seems to be a lag issue when setting "threadsafe" to "yes". Maybe you will click the same issue as shown here: http://code.google.com/p/googleappengine/issues/detail?id=6323

Update: By default, each web server only processes one request at a time. If you mark your application as thread safe, App Engine can send multiple requests to each web server in parallel. To do this, simply add a true appengine-web.xml element as described in the Using Parallel Queries section.



All Articles