ABCDEFGHIJKLMNOPQRSTUVWXYZAAABACADAEAFAGAHAIAJAKALAMANAO
1
Server config:No upper bound on memory limitKey size:
Not fixed size
Value size:1KTest duration:30 seconds
2
3
Load (#clients N)110100100010000
4
Latency (In ms)MemoryLatency (In ms)MemoryLatency (In ms)MemoryLatency (In ms)MemoryLatency (In ms)Memory
5
Keys (M)Average latency (ms)95%ile (ms)99%ile (ms)RPSmin (ms)max (ms)Peak memory usage (KB)Average latency (ms)95%ile (ms)99%ile (ms)RPSmin (ms)max (ms)Peak memory usage (KB)Average latency (ms)95%ile (ms)99%ile (ms)RPSmin (ms)max (ms)Peak memory usage (KB)Average latency (ms)95%ile (ms)99%ile (ms)RPSmin (ms)max (ms)Peak memory usage (KB)Average latency (ms)95%ile (ms)99%ile (ms)RPSmin (ms)max (ms)Peak memory usage (KB)
6
50.21330.5033036325.0463844.608535508813.1514024049.503404070721188.3963007100136.507472496740410.0932006300386.7011036479484
7
101.6919190.50193546765.84711005.1010736743613.8616027051.203274116361238.7476008600155.109025498936255.5718002900123.6019550463000
8
1001.3619190.50193626006.12881005.1010037215215.4715033049.304574098761198.9374008600164.409261496352579.38340010000237.3018733487480
9
10001.2918180.60183730165.5677885.5011338970411.821402104902503265161045.6765007400149083335018961049.9680011000377.3014274477672
10
100001.5722220.60224074248.811201405.2014138970410.7913020049.802394224201032.7867007900161.8085125226561169.19810013000396.6015769539040
11
12
13
14
15
16
17
Description:Perform lookup of a public key with N number of parallel clients, when the server is loaded with M number of keys. The observation are made with clients and server running in the same machine. Hence the latency numbers does not have any network overhead. The latency numbers would change when the server and clients have network related latency.
18
19
20
Legend:
21
Average latency (ms)This refers to the average response time for the requests, measured in milliseconds. It represents the arithmetic mean of all the response times recorded during the test for a specific task or endpoint.
22
95%ile (ms)This is the response time at the 95th percentile. It indicates that 95% of the requests had a response time equal to or lower than this value.
23
99%ile (ms)This represents the response time at the 99th percentile. It indicates that 99% of the requests had a response time equal to or lower than this value.
24
RPSThis represents the current requests per second being generated by the test at the time of measurement.
25
min (ms)This is the minimum response time observed during the test, measured in milliseconds.
26
max (ms)This is the maximum response time observed during the test, measured in milliseconds.
27
Peak memory usage (KB)Peak memory refers to the highest amount of memory that was consumed by the load testing process during its execution
28
29
30
31
Inferences:
32
Average Latency vs. Load Clients:
33
As the number of load clients increases:

1. For each key count, there's a general trend of an increase in average latency.

2. This increase is more pronounced when the load clients are higher, especially noticeable when comparing 1 client to 1000 clients.
34
Average Latency vs. Key Count:
35
For each load client count:

1. There's a slight increase in average latency as the number of keys increases.

2. However, this increase is not consistent across all load client counts and may vary based on system load and resource availability.
36
37
38
Peak Memory Usage vs. Load Clients:
39
Peak memory usage tends to vary based on load clients:

1. With 1 client, peak memory usage ranges from 303632 KB to 407424 KB across different key counts.

2. With 1000 clients, peak memory usage ranges from 496740 KB to 522656 KB across different key counts.

3. There's a general increase in peak memory usage as the number of load clients increases, indicating higher resource utilization under heavier loads.
40
41
42
43
44
45
46
RPS vs. Load Clients:
47
1. RPS tends to increase as the number of load clients increases

2. With 1 client, RPS ranges from 0.5 to 0.6 across different key counts.

3. With 1000 clients, RPS ranges from 136.5 to 161.8 across different key counts.

4. The increase in RPS is not linear with the increase in load clients, indicating potential saturation or system-level limitations.
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100