Design and Performance Evaluation of Distributed Virtual Router Based on OpenStack

2014 ◽  
Vol 687-691 ◽  
pp. 2888-2892
Author(s):  
Zhen Shan Bao ◽  
Bao Ping Wang ◽  
Wen Bo Zhang
Keyword(s):  
Performance Evaluation ◽  
Data Center ◽  
Data Flow ◽  
Network Node ◽  
Bottle Neck ◽  
Distributed Router ◽  
And Performance ◽  
East West

With the booming of Cloud Compute, OpenStack has been more widely used in Data Center, but in tradition deployment of OpenStack, the traffic must traverse network node to get routing decision. So the network node would be a bottle-neck due to the increasing amount of data flow. In order to solve this problem, this paper propose an extensible design, we create a internal virtual router in each compute node, the internal virtual router just can process the East-West traffic in lay2. With the help of distributed router, the East-West traffic can be delivered directly from the source hypervisor to the destination hypervisor, no longer need to traverse the router in network node to get routing decision, Therefore, North-South and East-West traffic don't impact each other anymore. North-South traffic enjoys all the bandwidth of router in network node exclusively, hence reduce the load of the network node. Experiment results show that OpenStack with distributed router has a much better performance evaluation than original OpenStack. This will dramatically increase the total bandwidth, and also bring other advantages.

Babcock Bottle Certification Appparatus Performance Evaluation

1995 ◽  
Vol 78 (2) ◽  
pp. 463-470 ◽  
Author(s):  
Joanna M Lynch ◽  
David M Barbano ◽  
George Houghton ◽  
J Richard Fleming
Keyword(s):  
Performance Evaluation ◽  
Internal Pressure ◽  
Pressure Effect ◽  
Short Neck ◽  
Additional Experiment ◽  
Volume Determination ◽  
Bottle Neck ◽  
Computer Controlled ◽  
And Performance ◽  
Volume Estimates

Abstract The use and performance of a computer-controlled apparatus for certification of Babcock bottles used for payment testing were evaluated. The apparatus delivered mercury into the bottle neck using a syringe mounted on a motorized pump. Syringe movement and the collection and calculation of data were controlled by computer. The apparatus was evaluated using four 8% milk bottles (total volume 1.600 mL; bottles rejected if deviation was ≥0.008 mL) and four 50% short-neck cream bottles (total volume 5.000 mL; bottles rejected if deviation was ≥0.050 mL). Six milk and 5 cream bottle trials were conducted; each bottle was read 5 consecutive times for each trial. As a percentage of the rejection criteria, average repeatability was 5–6% and reproducibility was 9–10%. These values were similar for both types of bottles. The evaluation of the apparatus demonstrated acceptable within- and between- days performance in relation to the rejection criteria and volumes measured. Because the apparatus creates a closed system during certification, an additional experiment was conducted with 50% short-neck cream bottles to determine the effect of internal pressure within a bottle on volume estimates. Four trials were conducted, as previously described, using 4 control and 4 experimental bottles. Volume between the 0 and 50% marks (5 mL) was determined. Holes were drilled in the experimental bottles to eliminate internal pressure for the final 2 trials. The estimated volume of bottles under pressure was 0.0054 ± 0.0012 mL greater than without internal pressure (i.e., about 11% of the rejection criteria). We concluded that any pressure effect was small relative to other factors that affect volume determination.

Sign in / Sign up

Export Citation Format

Share Document