[Notes] 802.11ac Aps performance compared to Clisco APs.

IxChariot utilized the script "High-Performace_TCP.scr" to generate maximum TCP traffic, while handling downstream traffic.

Each access point was attached horizontally against a dropdown ceiling.

Four different distance and obstruction scenarios were used:

• 5 feet LOS
• 30 feet Non-LOS
• 75 feet Non-LOS
• 120 feet Non-LOS

Each client device was tested in each set one at a time.

• a Samsung Galaxy S4, 1SS
• a MacBook Air notebook, 2SS(Now is 3SS)
• a MacBook Pro laptop, 3SS

http://www.arubanetworks.com/pdf/technology/TR_Miercom_02Oct2013.pdf

Theoretical Calculation of wire speed.

Wire speed = ( ByterPerSecond - Noumber Of Flags ) = ( 74,880,000 - 1 ) = 74,879,999

[perforance] QoS One to Many

sequence:
1. GE0 to FE0, 每一個Priority 打100 packets ( per burst) 共800個 packets.
2. GE0 to FE1, 每一個Priority 打100 packets ( per burst) 共800個 packets.
3. .......以至類推
4. 共 loop 9300次 (4 個port × 8 個priority 共 32個 Streams )

[performance] Partially Meshed

Partially Meshed Test
是用來測定DUT的傳送frames最大容許能力，多個傳送Ports 到多個接收ports在一個網路格式，在那傳送Ports 沒有接收 和接收Ports沒有傳送時。

per burst 4 packets.

[performance] Head of line blocking.

Description:
這個測驗 尋找 在一個"不擁塞(uncongested)" output interface上 所增加的delay,每當從一個input interface 那裡得到框架,再試圖把frames轉傳送到一個擁擠的output interface. 如果那Port 是設定為half duplex, collisions 應該被發現的在那擁擠的interface. 但如果是 full duplex 和 flow control 是enabled, flow control frames 應該被發現.
Topology:
Port A, B 傳送封包到 Port C. 而 Port A 也傳送封包到 Port D.
(Port A 傳到 Port C, D 而Port B傳到 C, 此時 Port C 會收到 Port A 和Port B傳送來的封包 而造成 Port C的擁擠 100% offered load. 但 Port D只有收到 Port A 所傳送來的封包 所以Port D 是不擁擠的Port 只有 50% 的offered load.)

[performance] Back to Back.

Description:
這個測驗是測定，在沒有 frames loss 情況下，DUT 能夠 接收 和 轉送 的最大時間。
frames 是傳送 user-specified rate, 通常 最大理論 rate是基於port speed.
Topology:
This test uses pairs of ports with one-to-one traffic mapping; one port transmits to one receive port.

[performance] latency test.

在 latency 測試， frames 是傳送一個固定的時間. 在 每秒 test 框架加標簽和透過持續時間時間 half way 傳導它。 當這test 傳導什麼時候收到了frames時, 測驗把連接的框架的 timestamp 與 timestamp 比較。 二個 timestamps 之間的差別是等待時間。

Address learning test.

Description:
在沒有 folooding 或者 droppig 框架的情況下，Switch 能夠學習新 MAC address 的Max rate.
(學的多快)
Topology:
This test uses the one-to-many mapping but only one port is used for transmission at a time; received address packets should appear on the specified receive ports.

Address cache testing

Description:
這個測試是用Binary search 的方式找出 每個 port 或一台 swcitch 的 address table size.

IXIA port 1 是DA 遞增，port 2,3,4 是 SA 遞增
當port 2,3,4 傳送packet出去後，然後L2 會將mac學起來。再由Ixia的port 1 傳到 Port 2,3,4以確定那些mac是有被學起來的!

Topology:
Then the test only needs to be run on one transmit port to all receive ports on the switch; This test is configured with a one-to-many traffic mapping.

[Performance] Frame Error Filtering of IXIA.

The LM1000T-5 Load Module(GE module) not support 'Alignment' and 'Dribble'.
The LM100TX Load Module(FE module) suport all item. (Undersize,Oversize,Fragment,Alignment,Dribble,Bad CRC)