ThingWorx Model and Data Best Practices > Data Ordering > Performance Results
Performance Results
The following test setup and observations were made during the performance testing.
Test Environment-Single Server (with Azure PostgreSQL Flex Server Database)
◦ CPU: 32
◦ Processor: Intel(R) Xeon(R) Platinum 8168 CPU @ 2.70GHz
◦ RAM: 64 GB
◦ Database: PostgreSQL
Test Environment-HA Setup (4 Node configuration with Azure PostgreSQL Flex Server Database)
For each node, the following setup was used:
◦ CPU: 8
◦ Processor: Intel(R) Xeon(R) Platinum 8168 CPU @ 2.70GHz
◦ RAM: 16 GB
◦ Database: Azure Flex Server
Test Environment-Axeda Edge
◦ eMC Server: 1 for Single server and 4 for High Availability
◦ Data Ordering Enabled: true
Simulator Details
◦ Devices (Things): ~30000
◦ Properties per Thing: Integer (10 nos.) and String (10 nos.)
◦ Property update frequency: 60 seconds
◦ Data Ingestion frequency: 60 seconds
◦ Properties with Alerts: 5
All property updates were pushed to Value Stream.
JAVA Simulator
• For ThingWorx Single Server + Azure PostgreSQL Flex + Connection Server (Large Size) (~30000 devices)
Data Ordering Status
Platform Consumption
Database Consumption
WS Requests
VS Queue Requests
(WPS)
CPU (%)
(32 cores)
Memory
(out of 128 GB)
CPU (%)
Memory
(out of 128 GB)
Writes
(OPS)
OFF
30
40.46 GB
(31.6%)
23.53
41 GB
(32%)
9366
56720
ON
35.98
32.06 GB
(25%)
15.16
40GB
(31%)
7949
59220
• For ThingWorx High Availability (4 nodes) + Azure PostgreSQL Flex + Connection Server (Small Size) (~30000 devices)
Data Ordering Status
Platform 1
Platform 2
Platform 3
Platform 4
Database
CPU (%)
Memory (GB)
VS Queue Rate (WPS)
Stream Queue Rate (WPS)
CPU (%)
Memory (GB)
VS Queue Rate (WPS)
Stream Queue Rate (WPS)
CPU (%)
Memory (GB)
VS Queue Rate (WPS)
Stream Queue Rate (WPS)
CPU (%)
Memory (GB)
VS Queue Rate (WPS)
Stream Queue Rate (WPS)
CPU (%)
Memory (GB)
WS Requests (Writes)
OFF
51.5
7.13
17364
1.5
51.5
7.13
17364
1.5
55.59
3.60
17325
1.17
49.82
7.23
17438
1.4
50.73
37.59
7453
ON
57.98
6.08
17291
1.27
49.48
7.36
17953
1.27
51.89
3.66
17736
1.35
51.56
7.19
17997
1.14
48.41
35.79
7593
Axeda EDGE
• ThingWorx Single Server + Axeda Edge + EMC + Azure PostgreSQL Flex (Large Size) (~30000 devices)
Data Ordering Status
Platform
Database
CPU (%)
(16 Cores)
Memory (GB)
(out of 64GB)
VS Queue Rate (WPS)
CPU(%)
(16 Cores)
Memory (GB)
(out of 64GB)
OFF
34.55
5.84
23057
9.96
31.07
ON
24.68
5.42
19738
7.68
30.89
• ThingWorx High Availability (4 Nodes) + Axeda Edge + EMC + Azure PostgreSQL Flex (Large Size) (~30000 devices)
Data Ordering Status
Platform 1
Platform 2
Platform 3
Platform 4
Database
CPU (%)
(16 Cores)
Memory (GB)
(out of 64GB)
VS Queue Rate (WPS)
CPU (%)
Memory (GB)
(out of 64GB)
VS Queue Rate (WPS)
CPU (%)
Memory (GB)
(out of 64GB)
VS Queue Rate (WPS)
CPU (%)
Memory (GB)
(out of 64GB)
VS Queue Rate (WPS)
CPU (%)
Memory (GB)
(out of 64GB)
OFF
44.92
5.45
7212
42.84
5.73
7142
39.98
4.51
7169
43.09
6.95
7212
11.26
32.46
ON
33.48
4.8
7539
33.50
7.55
7556
33.23
4.69
7455
33.23
5.17
7444
15.84
31.90
Summary
Simulator
Setup
Observation for Data Ordering ON compared with Data Ordering OFF
JAVA Simulator
ThingWorx Single Server + Azure PostgreSQL Flex + Connection Server (Large Size)
No Performance Impacts
JAVA Simulator
ThingWorx High Availability (4 nodes) + Azure PostgreSQL Flex + Connection Server (Small Size)
Minor performance drop is observed on few nodes. Overall system performance is consistent.
Axeda EDGE
ThingWorx Single Server + Axeda Edge + EMC + Azure PostgreSQL Flex (Small Size)
Minor performance drop is observed.
Axeda EDGE
ThingWorx High Availability (4 Nodes) + Axeda Edge + EMC + Azure PostgreSQL Flex (Small Size)
Minor performance drop is observed on few nodes. Overall system performance is consistent.
• Scope of Observations: The findings are based on a single performance run. Variability in outcomes may occur depending on specific configurations, including edge device setups, application load, and system configuration adjustments.
• Performance in High Availability (HA) Setup: During testing, a minor performance degradation was observed on a few nodes within the HA setup. However, the overall system performance remained consistent, with no substantial impact noted across the setup.
• Scalability Considerations: Despite minor deviations on certain nodes, the system demonstrates resilience and scalability under HA configurations, maintaining reliability and operational consistency.
• Contextual Influences: Performance behavior may differ in production environments due to factors such as real-world network conditions, resource allocation policies, and operational workloads.
Was this helpful?