High Replication latencies and it's related costs.

[thecheatah]

Hi,

I am noticing that the latency on my HR based application went up
significantly. My app does a simple read and a write. It is taking
over 1 second to do this operation. In M/S the this operation was <
100ms

Even under multi threading, my machines are not processing as many
requests as my M/S app could handle. My costs have gone up by 50%
after having switched from M/S. Is this normal?

Last night a single machine handled about 6-7 requests/second. Today
it has dropped to 4. Nothing has changed. The request are generated by
devices and thus are very consistent.

It seems that the latency on the HR datastore is causing my cost to
increase. I have no control over this latency. Is this something I can
get money back for?

Ravneet

[Gopal Patel]

if request/second dropped, nothing could be done. latency has nothing to do with request/second. if nothing else, it will increase the instance count, google have no control over request/second. 

--
You received this message because you are subscribed to the Google Groups "Google App Engine" group.
To post to this group, send email to google-a...@googlegroups.com.
To unsubscribe from this group, send email to google-appengi...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.

[andidol]

Have the same problem. HR datastore is very slow since yesterday..

[Ikai Lan (Google)]

Can you post the code about what you are doing? "Simple read and write" isn't descriptive enough.

There have been issues with High Replication, but these are/have been addressed.

--

Ikai Lan 
Developer Programs Engineer, Google App Engine

plus.ikailan.com

[thecheatah]

The over all requests/second have been consistent from day to day. I
am talking about the ability of a single machine's ability to handle
requests/second.

Ravneet