blob: 655559d414a3eb3d4678ef20a75cac117d49b57e (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
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
101
102
103
|
.. This work is licensed under a Creative Commons Attribution 4.0 International License.
.. http://creativecommons.org/licenses/by/4.0
.. (c) Cisco Systems, Inc
RELEASE NOTES
+++++++++++++
Release 2.0
===========
Major release highlights:
- Dedicated chain networks
- VxLAN support with VTEP in the traffic generator
- Enhanced chain analysis
- Code refactoring and enhanced unit testing
- Miscellaneous enhancement
Dedicated chain networks
------------------------
NFVbench 1.x only supported shared networks across chains.
For example, 20xPVP would create only 2 networks (left and right) shared by all chains.
With NFVbench 2.0, chain networks will become dedicated (unshared) by default with an option in
the nfvbench configuration to shared them. A 20xPVP run will create 2x20 networks instead.
Enhanced chain analysis
-----------------------
The new chain analysis improves at multiple levels:
- there is now one table for each direction (forward and reverse) that both read from left to right
- per-chain packet counters and latency
- all-chain aggregate packet counters and latency
- supports both shared and dedicated chain networks
Code refactoring and enhanced unit testing
------------------------------------------
The overall code structure is now better partitioned in the following functions:
- staging and resource discovery
- traffic generator
- stats collection
The staging algorithm was rewritten to be:
- a lot more robust to errors and to handle better resource reuse use cases.
For example when a network with a matching name is discovered the new code will verify that the
network is associated to the right VM instance
- a lot more strict when it comes to the inventory of MAC addresses. For example the association
from each VM MAC to a chain index for each Trex port is handled in a much more strict manner.
Although not all code is unit tested, the most critical parts are unit tested with the use of
the mock library. The resulting unit test code can run in isolation without needing a real system under test.
OPNFV Fraser Release
====================
Over 30 Jira tickets have been addressed in this release (Jira NFVBENCH-55 to NFVBENCH-78)
The Fraser release adds the following new features:
- support for benchmarking non-OpenStack environments (with external setup and no OpenStack openrc file)
- PVVP packet path with SRIOV at the edge and vswitch between VMs
- support logging events and results through fluentd
Enhancements and main bug fixes:
- end to end connectivity for larger chain count is now much more accurate for large chain count - avoiding excessive drops
- use newer version of TRex (2.32)
- use newer version of testpmd DPDK
- NDR/PDR uses actual TX rate to calculate drops - resulting in more accurate results
- add pylint to unit testing
- add self sufficient and standalone unit testing (without actual testbed)
OPNFV Euphrates Release
=======================
This is the introductory release for NFVbench. In this release, NFVbench provides the following features/capabilities:
- standalone installation with a single Docker container integrating the open source TRex traffic generator
- can measure data plane performance for any NFVi full stack
- can setup automatically service chains with the following packet paths:
- PVP (physical-VM-physical)
- PVVP (physical-VM-VM-physical) intra-node and inter-node
- can setup multiple service chains
- N * PVP
- N * PVVP
- supports any external service chain (pre-set externally) that can do basic IPv4 routing
- can measure
- drop rate and latency for any given fixed rate
- NDR (No Drop Rate) and PDR (Partial Drop Rate) with configurable drop rates
- traffic specification
- any fixed frame size or IMIX
- uni or bidirectional traffic
- any number of flows
- vlan tagging can be enabled or disabled
- user interface:
- CLI
- REST+socketIO
- fully configurable runs with yaml-JSON configuration
- detailed results in JSON format
- summary tabular results
- can send logs and results to one or more fluentd aggregators (per configuration)
|
