Service for keeping track of the last price for financial instruments. Producers will use the service to publish prices and consumers will use it to obtain them.
This section describes domain objects.
The price data consists of records with the following fields:
id: a string field to indicate which instrument this price refers to.asOf: a date time field to indicate when the price was determined.payload: the price data itself, which is a flexible data structure.
Producers should be able to provide prices in batch runs. The sequence of uploading a batch run is as follows:
- The producer indicates that a batch run is started
- The producer uploads the records in the batch run in multiple chunks of 1000 records.
- The producer completes or cancels the
batch run.
Batch run is a session of uploading chunks.
Consumers can request the last price record for a given id. The last value is determined by the asOf time, as set by the producer. On completion, all prices in a batch run should be made available at the same time. Batch runs which are cancelled can be discarded.
The service should be resilient against producers which call the service methods in an incorrect order, or clients which call the service while a batch is being processed.
System implements simple client/server protocol based on Aeron. The protocol was discussed here: aeron-io/aeron#652
This implementation also introduce concept of independent local clients which can be used in separate threads.
When client process is started it creates Gateway (AbstractServiceGateway) to access service.
Gateway manages connection to server as pair of Aeron Piblication and Subscription.
Clients (see AbstractServiceClient) sends requests to server writing them through the Gateway's publication (Concurrent Publication is used).
Gateway runs single read cycle (see com.xxx.core.client.AbstractServiceGateway.ReadCycleTask) in separate thread to read responces from the server (Aeron Subscription).
When ReadCycleTask receives a message from server and redirects it to right local client writing to client's OneToOneRingBuffer.
Server use dedicated channel for all incomming connections (at the moment admin messages are also running there).
Server works in single thread mode. Main loop recieves messages from the Aeron Subscription, react on them (business logic or administration) and replies to clients.
All application level messages have header with the following fields:
- connectionId - client sets it in each request, using this number server can define to which send message back. Generated by server.
- clientId - used to redirect responses from main Gateway's read thread to the right client object (I know may be it's confusing name, could not create better)
- correlationId - unique id per request/response pair generated by client.
All components are trying to avoid memory allocations during the whole message processing path.
All messages and internal structures use Flyweight pattern wraping direct buffers.
Server does not save all records client sends in chunks. Instead of it updates state of the batch run.
Batch run (com.xxx.service.lastprice.BatchRun) presents aggregation view of all previous successful chunk uploading operations.
BatchRun objects are managed by repository which tracks last batch usage time, it allows to delete abandoned batch runs and return them to the repository's pool.
If client decides to cancel batch run it is just returned to the repository pool and can be reused.
If batch is completed it is merging to the similar structure MarketState. All read operations will be done from market state.
Process of chunk uploading uses additional operational batch run structure (com.xxx.service.lastprice.LastPriceServiceHandler.operationalBatchRun) to ensure all chunk records will be writen or none of them. When operational batch run is filled it is merged to the right batch run.
Examples of client/service communications are presented in samples module as performance tests.
Tests assume clients and server will be run on the same machine.
To run a perfomance test three steps are needed to complete:
- Run external Aeron Media Driver on the host. Package has low latency media driver runner implementation
LowLatencyMediaDriver - Start the service running
LastPriceServiceRunner - Start one of the performance test you need to check results for
If service and clients are on different hosts, then each host should run own media driver process.
Several scenarious were implemented to test system perfomance in terms of request/responce latency.
This test is implemeted in StartCancelBatchRunLatencyTest class.
Test measures latency of two conséquent operations: creation and cancelation of batches.
Results in the table (results are in microseconds):
Value Percentile TotalCount 1/(1-Percentile)
32.543 0.000000000000 1 1.00
43.039 0.100000000000 10047 1.11
46.015 0.200000000000 20015 1.25
48.095 0.300000000000 30243 1.43
49.439 0.400000000000 40066 1.67
50.655 0.500000000000 50230 2.00
51.327 0.550000000000 55055 2.22
52.159 0.600000000000 60155 2.50
53.023 0.650000000000 65015 2.86
54.527 0.700000000000 70041 3.33
56.575 0.750000000000 75051 4.00
57.823 0.775000000000 77502 4.44
59.359 0.800000000000 80023 5.00
61.215 0.825000000000 82523 5.71
63.391 0.850000000000 85003 6.67
66.239 0.875000000000 87502 8.00
68.223 0.887500000000 88784 8.89
70.271 0.900000000000 90017 10.00
72.767 0.912500000000 91257 11.43
75.647 0.925000000000 92504 13.33
79.103 0.937500000000 93772 16.00
81.023 0.943750000000 94387 17.78
83.327 0.950000000000 95015 20.00
85.823 0.956250000000 95632 22.86
89.023 0.962500000000 96264 26.67
92.927 0.968750000000 96877 32.00
95.551 0.971875000000 97190 35.56
99.199 0.975000000000 97502 40.00
104.063 0.978125000000 97814 45.71
110.015 0.981250000000 98128 53.33
118.911 0.984375000000 98439 64.00
124.799 0.985937500000 98596 71.11
131.967 0.987500000000 98752 80.00
142.591 0.989062500000 98907 91.43
159.103 0.990625000000 99064 106.67
179.839 0.992187500000 99220 128.00
193.791 0.992968750000 99297 142.22
214.143 0.993750000000 99375 160.00
241.279 0.994531250000 99454 182.86
279.295 0.995312500000 99533 213.33
341.247 0.996093750000 99610 256.00
371.199 0.996484375000 99649 284.44
410.623 0.996875000000 99688 320.00
456.447 0.997265625000 99727 365.71
510.207 0.997656250000 99766 426.67
600.575 0.998046875000 99805 512.00
674.303 0.998242187500 99826 568.89
712.703 0.998437500000 99844 640.00
816.639 0.998632812500 99864 731.43
900.095 0.998828125000 99883 853.33
1003.519 0.999023437500 99903 1024.00
1142.783 0.999121093750 99913 1137.78
1224.703 0.999218750000 99922 1280.00
1339.391 0.999316406250 99932 1462.86
1538.047 0.999414062500 99942 1706.67
1674.239 0.999511718750 99952 2048.00
1762.303 0.999560546875 99957 2275.56
1899.519 0.999609375000 99961 2560.00
2285.567 0.999658203125 99966 2925.71
2502.655 0.999707031250 99971 3413.33
3004.415 0.999755859375 99976 4096.00
3557.375 0.999780273438 99979 4551.11
3764.223 0.999804687500 99981 5120.00
4689.919 0.999829101563 99983 5851.43
5533.695 0.999853515625 99986 6826.67
6008.831 0.999877929688 99988 8192.00
6348.799 0.999890136719 99990 9102.22
6664.191 0.999902343750 99991 10240.00
7491.583 0.999914550781 99992 11702.86
8228.863 0.999926757813 99993 13653.33
9076.735 0.999938964844 99994 16384.00
12320.767 0.999945068359 99995 18204.44
14376.959 0.999951171875 99996 20480.00
14376.959 0.999957275391 99996 23405.71
16203.775 0.999963378906 99997 27306.67
16203.775 0.999969482422 99997 32768.00
18153.471 0.999972534180 99998 36408.89
18153.471 0.999975585938 99998 40960.00
18153.471 0.999978637695 99998 46811.43
22298.623 0.999981689453 99999 54613.33
22298.623 0.999984741211 99999 65536.00
22298.623 0.999986267090 99999 72817.78
22298.623 0.999987792969 99999 81920.00
22298.623 0.999989318848 99999 93622.86
24838.143 0.999990844727 100000 109226.67
24838.143 1.000000000000 100000
#[Mean = 59.587, StdDeviation = 171.029]
#[Max = 24838.143, Total count = 100000]
#[Buckets = 24, SubBuckets = 2048]
This test measures latency of chunk uploading operation. On each iteration test creates new batch run, new chunk of size 1000, upload the chunk and the complete the batch. Test measures performance of each operation separately. Results are in the following tables.
Value Percentile TotalCount 1/(1-Percentile)
16.039 0.000000000000 1 1.00
19.503 0.100000000000 1025 1.11
20.383 0.200000000000 2014 1.25
21.391 0.300000000000 3005 1.43
22.399 0.400000000000 4004 1.67
23.151 0.500000000000 5024 2.00
23.423 0.550000000000 5512 2.22
23.759 0.600000000000 6006 2.50
24.271 0.650000000000 6514 2.86
24.703 0.700000000000 7005 3.33
25.263 0.750000000000 7506 4.00
25.679 0.775000000000 7751 4.44
26.623 0.800000000000 8000 5.00
27.839 0.825000000000 8252 5.71
29.247 0.850000000000 8501 6.67
30.447 0.875000000000 8750 8.00
31.103 0.887500000000 8875 8.89
31.935 0.900000000000 9000 10.00
32.959 0.912500000000 9126 11.43
34.015 0.925000000000 9250 13.33
35.231 0.937500000000 9378 16.00
35.999 0.943750000000 9438 17.78
36.575 0.950000000000 9501 20.00
37.599 0.956250000000 9565 22.86
38.815 0.962500000000 9625 26.67
40.991 0.968750000000 9689 32.00
41.919 0.971875000000 9720 35.56
42.463 0.975000000000 9750 40.00
43.359 0.978125000000 9782 45.71
44.383 0.981250000000 9814 53.33
45.375 0.984375000000 9844 64.00
46.239 0.985937500000 9862 71.11
46.847 0.987500000000 9875 80.00
47.679 0.989062500000 9891 91.43
48.447 0.990625000000 9907 106.67
49.151 0.992187500000 9922 128.00
49.407 0.992968750000 9930 142.22
50.015 0.993750000000 9938 160.00
50.687 0.994531250000 9947 182.86
51.871 0.995312500000 9954 213.33
53.695 0.996093750000 9961 256.00
54.463 0.996484375000 9965 284.44
56.223 0.996875000000 9970 320.00
59.903 0.997265625000 9973 365.71
65.247 0.997656250000 9977 426.67
73.343 0.998046875000 9981 512.00
79.551 0.998242187500 9983 568.89
83.327 0.998437500000 9985 640.00
86.143 0.998632812500 9987 731.43
117.247 0.998828125000 9989 853.33
126.271 0.999023437500 9991 1024.00
128.639 0.999121093750 9992 1137.78
130.623 0.999218750000 9993 1280.00
181.503 0.999316406250 9994 1462.86
290.047 0.999414062500 9995 1706.67
291.839 0.999511718750 9996 2048.00
291.839 0.999560546875 9996 2275.56
319.231 0.999609375000 9997 2560.00
319.231 0.999658203125 9997 2925.71
383.231 0.999707031250 9998 3413.33
383.231 0.999755859375 9998 4096.00
383.231 0.999780273438 9998 4551.11
419.583 0.999804687500 9999 5120.00
419.583 0.999829101563 9999 5851.43
419.583 0.999853515625 9999 6826.67
419.583 0.999877929688 9999 8192.00
419.583 0.999890136719 9999 9102.22
599.551 0.999902343750 10000 10240.00
599.551 1.000000000000 10000
#[Mean = 24.825, StdDeviation = 11.291]
#[Max = 599.551, Total count = 10000]
#[Buckets = 24, SubBuckets = 2048]
Value Percentile TotalCount 1/(1-Percentile)
232.831 0.000000000000 1 1.00
276.991 0.100000000000 1016 1.11
286.207 0.200000000000 2013 1.25
293.631 0.300000000000 3027 1.43
299.519 0.400000000000 4018 1.67
306.943 0.500000000000 5000 2.00
311.039 0.550000000000 5533 2.22
315.391 0.600000000000 6008 2.50
321.535 0.650000000000 6506 2.86
327.935 0.700000000000 7000 3.33
334.079 0.750000000000 7507 4.00
338.687 0.775000000000 7762 4.44
344.063 0.800000000000 8002 5.00
352.511 0.825000000000 8253 5.71
361.215 0.850000000000 8503 6.67
372.991 0.875000000000 8751 8.00
380.927 0.887500000000 8879 8.89
389.887 0.900000000000 9001 10.00
401.919 0.912500000000 9127 11.43
423.423 0.925000000000 9250 13.33
455.679 0.937500000000 9375 16.00
499.199 0.943750000000 9438 17.78
547.839 0.950000000000 9501 20.00
595.967 0.956250000000 9563 22.86
614.399 0.962500000000 9627 26.67
630.783 0.968750000000 9689 32.00
639.487 0.971875000000 9720 35.56
646.655 0.975000000000 9751 40.00
654.847 0.978125000000 9782 45.71
665.087 0.981250000000 9813 53.33
675.327 0.984375000000 9844 64.00
684.031 0.985937500000 9860 71.11
690.175 0.987500000000 9876 80.00
697.343 0.989062500000 9891 91.43
703.487 0.990625000000 9908 106.67
706.559 0.992187500000 9922 128.00
709.631 0.992968750000 9930 142.22
712.703 0.993750000000 9938 160.00
716.799 0.994531250000 9947 182.86
720.895 0.995312500000 9954 213.33
724.479 0.996093750000 9961 256.00
729.599 0.996484375000 9965 284.44
731.647 0.996875000000 9969 320.00
735.743 0.997265625000 9973 365.71
777.215 0.997656250000 9977 426.67
859.135 0.998046875000 9981 512.00
875.519 0.998242187500 9983 568.89
1139.711 0.998437500000 9985 640.00
1223.679 0.998632812500 9987 731.43
1401.855 0.998828125000 9989 853.33
1503.231 0.999023437500 9991 1024.00
1606.655 0.999121093750 9992 1137.78
2024.447 0.999218750000 9993 1280.00
2136.063 0.999316406250 9994 1462.86
2420.735 0.999414062500 9995 1706.67
2666.495 0.999511718750 9996 2048.00
2666.495 0.999560546875 9996 2275.56
2818.047 0.999609375000 9997 2560.00
2818.047 0.999658203125 9997 2925.71
3096.575 0.999707031250 9998 3413.33
3096.575 0.999755859375 9998 4096.00
3096.575 0.999780273438 9998 4551.11
4069.375 0.999804687500 9999 5120.00
4069.375 0.999829101563 9999 5851.43
4069.375 0.999853515625 9999 6826.67
4069.375 0.999877929688 9999 8192.00
4069.375 0.999890136719 9999 9102.22
4304.895 0.999902343750 10000 10240.00
4304.895 1.000000000000 10000
#[Mean = 333.507, StdDeviation = 117.132]
#[Max = 4304.895, Total count = 10000]
#[Buckets = 24, SubBuckets = 2048]
Value Percentile TotalCount 1/(1-Percentile)
17.727 0.000000000000 1 1.00
20.895 0.100000000000 1007 1.11
22.143 0.200000000000 2015 1.25
22.895 0.300000000000 3021 1.43
23.439 0.400000000000 4027 1.67
24.047 0.500000000000 5019 2.00
24.399 0.550000000000 5517 2.22
24.719 0.600000000000 6009 2.50
25.151 0.650000000000 6511 2.86
25.791 0.700000000000 7000 3.33
26.895 0.750000000000 7501 4.00
27.599 0.775000000000 7764 4.44
28.351 0.800000000000 8008 5.00
29.407 0.825000000000 8251 5.71
30.575 0.850000000000 8506 6.67
31.935 0.875000000000 8753 8.00
32.543 0.887500000000 8875 8.89
33.375 0.900000000000 9005 10.00
33.983 0.912500000000 9131 11.43
34.751 0.925000000000 9252 13.33
36.191 0.937500000000 9376 16.00
37.119 0.943750000000 9439 17.78
38.559 0.950000000000 9500 20.00
39.903 0.956250000000 9564 22.86
41.919 0.962500000000 9625 26.67
43.871 0.968750000000 9688 32.00
44.575 0.971875000000 9719 35.56
45.055 0.975000000000 9751 40.00
46.271 0.978125000000 9783 45.71
47.839 0.981250000000 9813 53.33
49.087 0.984375000000 9844 64.00
49.695 0.985937500000 9861 71.11
50.399 0.987500000000 9876 80.00
50.815 0.989062500000 9891 91.43
51.551 0.990625000000 9907 106.67
52.671 0.992187500000 9922 128.00
53.183 0.992968750000 9930 142.22
53.951 0.993750000000 9939 160.00
55.583 0.994531250000 9946 182.86
59.039 0.995312500000 9954 213.33
60.575 0.996093750000 9961 256.00
63.199 0.996484375000 9965 284.44
66.047 0.996875000000 9969 320.00
71.487 0.997265625000 9973 365.71
79.615 0.997656250000 9977 426.67
89.599 0.998046875000 9981 512.00
101.887 0.998242187500 9983 568.89
114.367 0.998437500000 9985 640.00
117.183 0.998632812500 9987 731.43
138.495 0.998828125000 9989 853.33
211.199 0.999023437500 9991 1024.00
351.999 0.999121093750 9992 1137.78
386.303 0.999218750000 9993 1280.00
416.511 0.999316406250 9994 1462.86
562.175 0.999414062500 9995 1706.67
676.863 0.999511718750 9996 2048.00
676.863 0.999560546875 9996 2275.56
808.447 0.999609375000 9997 2560.00
808.447 0.999658203125 9997 2925.71
1612.799 0.999707031250 9998 3413.33
1612.799 0.999755859375 9998 4096.00
1612.799 0.999780273438 9998 4551.11
2383.871 0.999804687500 9999 5120.00
2383.871 0.999829101563 9999 5851.43
2383.871 0.999853515625 9999 6826.67
2383.871 0.999877929688 9999 8192.00
2383.871 0.999890136719 9999 9102.22
2392.063 0.999902343750 10000 10240.00
2392.063 1.000000000000 10000
#[Mean = 26.899, StdDeviation = 39.841]
#[Max = 2392.063, Total count = 10000]
#[Buckets = 24, SubBuckets = 2048]
Value Percentile TotalCount 1/(1-Percentile)
14.271 0.000000000000 1 1.00
19.327 0.100000000000 10138 1.11
19.839 0.200000000000 20206 1.25
20.719 0.300000000000 30049 1.43
21.967 0.400000000000 40064 1.67
22.975 0.500000000000 50216 2.00
23.263 0.550000000000 55113 2.22
23.599 0.600000000000 60142 2.50
24.255 0.650000000000 65057 2.86
24.847 0.700000000000 70079 3.33
26.063 0.750000000000 75022 4.00
27.279 0.775000000000 77511 4.44
28.543 0.800000000000 80009 5.00
29.439 0.825000000000 82560 5.71
30.111 0.850000000000 85016 6.67
31.135 0.875000000000 87529 8.00
31.759 0.887500000000 88782 8.89
32.735 0.900000000000 90003 10.00
33.599 0.912500000000 91261 11.43
34.335 0.925000000000 92528 13.33
35.327 0.937500000000 93753 16.00
35.807 0.943750000000 94416 17.78
36.319 0.950000000000 95000 20.00
37.567 0.956250000000 95631 22.86
39.935 0.962500000000 96257 26.67
41.343 0.968750000000 96900 32.00
41.599 0.971875000000 97238 35.56
41.823 0.975000000000 97502 40.00
42.303 0.978125000000 97826 45.71
43.103 0.981250000000 98131 53.33
43.583 0.984375000000 98456 64.00
43.839 0.985937500000 98607 71.11
44.287 0.987500000000 98752 80.00
45.151 0.989062500000 98912 91.43
45.727 0.990625000000 99070 106.67
46.335 0.992187500000 99230 128.00
46.815 0.992968750000 99298 142.22
47.167 0.993750000000 99382 160.00
47.487 0.994531250000 99456 182.86
47.839 0.995312500000 99535 213.33
48.255 0.996093750000 99614 256.00
48.479 0.996484375000 99652 284.44
48.767 0.996875000000 99688 320.00
49.055 0.997265625000 99730 365.71
49.439 0.997656250000 99767 426.67
49.983 0.998046875000 99805 512.00
50.495 0.998242187500 99825 568.89
51.391 0.998437500000 99844 640.00
52.863 0.998632812500 99865 731.43
55.263 0.998828125000 99883 853.33
58.015 0.999023437500 99903 1024.00
58.975 0.999121093750 99913 1137.78
60.255 0.999218750000 99923 1280.00
63.903 0.999316406250 99932 1462.86
71.935 0.999414062500 99942 1706.67
89.791 0.999511718750 99952 2048.00
98.047 0.999560546875 99957 2275.56
112.639 0.999609375000 99961 2560.00
132.607 0.999658203125 99966 2925.71
168.703 0.999707031250 99971 3413.33
205.823 0.999755859375 99976 4096.00
221.951 0.999780273438 99979 4551.11
222.207 0.999804687500 99981 5120.00
268.543 0.999829101563 99983 5851.43
316.159 0.999853515625 99986 6826.67
374.015 0.999877929688 99988 8192.00
456.191 0.999890136719 99990 9102.22
546.815 0.999902343750 99991 10240.00
693.247 0.999914550781 99992 11702.86
739.839 0.999926757813 99993 13653.33
748.543 0.999938964844 99994 16384.00
821.759 0.999945068359 99995 18204.44
1011.711 0.999951171875 99996 20480.00
1011.711 0.999957275391 99996 23405.71
1116.159 0.999963378906 99997 27306.67
1116.159 0.999969482422 99997 32768.00
1373.183 0.999972534180 99998 36408.89
1373.183 0.999975585938 99998 40960.00
1373.183 0.999978637695 99998 46811.43
3274.751 0.999981689453 99999 54613.33
3274.751 0.999984741211 99999 65536.00
3274.751 0.999986267090 99999 72817.78
3274.751 0.999987792969 99999 81920.00
3274.751 0.999989318848 99999 93622.86
3760.127 0.999990844727 100000 109226.67
3760.127 1.000000000000 100000
#[Mean = 24.694, StdDeviation = 18.901]
#[Max = 3760.127, Total count = 100000]
#[Buckets = 24, SubBuckets = 2048]
The service and clients are assumed to work on controlled network. Possible issues on public networks (I did't test it yet) aeron-io/aeron#697