diff options
author | drebs <drebs@riseup.net> | 2017-11-02 09:56:28 -0200 |
---|---|---|
committer | drebs <drebs@riseup.net> | 2017-11-02 09:56:28 -0200 |
commit | 2d72b4e044bb3a0fe4e6de70d8470a94e85e7754 (patch) | |
tree | 6d78e4cc8126fa6846e0a27525012811c16c760b /server-scalability/README.txt | |
parent | b36ebc87f449af5ad7a677350a3ccfc7d26cfb1e (diff) |
add blobs server scalability tests results
Diffstat (limited to 'server-scalability/README.txt')
-rw-r--r-- | server-scalability/README.txt | 38 |
1 files changed, 38 insertions, 0 deletions
diff --git a/server-scalability/README.txt b/server-scalability/README.txt new file mode 100644 index 0000000..e514cb8 --- /dev/null +++ b/server-scalability/README.txt @@ -0,0 +1,38 @@ +Blobs Server Scalability Tests Results +====================================== + +These are results obtained by running the blobs scalability tests that can be +found in: + + https://0xacab.org/leap/soledad/tree/master/scripts/scalability + +Overview +-------- + +The blobs server scalability tests aim to find the maximum amount of parallel +blobs upload and download requests that the server is able to respond to. We +use FunkLoad[1] to launch an increasing number of parallel upload/download +requests and to create reports with a summary of the results. The metrics +obtained are: + +- The number of Successful Tests Per Second given the number of Concurrent + Users. + +- The number of Successful Pages Per Second given the number of Concurrent + Users. A Page in the context of FunkLoad is a request to the server. For + tests that have only one Page request, this number is the same as the above. + +- The total number of Requests Per Second (successful or not) given the number + of Concurrent Users. + +- Graphs with the associated response times. + +[1] https://funkload.nuxeo.org/ + + +Scenarios +--------- + +The size of the uploaded/downloaded blobs and the client and server bandwidth +are parameters that highly influence the results of these tests. See the file +results/README.txt for information on the scenario used for each test. |