Tools
Sorted by:
Try your query at:
 |
 |
 |
 |
 |
 |
Results 1 - 10
of
7,742
Table 3: Network bandwidth consumption for common file-system operations. Shows the total number of bytes transmitted between all the nodes for each operation. Overhead shows the percentage of the bandwidth used by harbingers and duplicate updates.
2002
"... In PAGE 11: ... Instead, its network usage is near- optimal, with less than 2% of the bandwidth wasted. Table3 shows network bandwidth consumption for com- mon file-system update operations. Operations such as cre- ating a file or writing one byte show a high percentage of overhead, since they are sent directly without harbingers, but they have only a minor impact on the overall wasted bandwidth since their size is small.... ..."
Cited by 90
Table 3: Network bandwidth consumption for common file-system operations. Shows the total number of bytes transmitted between all the nodes for each operation. Overhead shows the percentage of the bandwidth used by harbingers and duplicate updates.
"... In PAGE 11: ... Instead, its network usage is near- optimal, with less than 2% of the bandwidth wasted. Table3 shows network bandwidth consumption for com- mon file-system update operations. Operations such as cre- ating a file or writing one byte show a high percentage of overhead, since they are sent directly without harbingers, but they have only a minor impact on the overall wasted bandwidth since their size is small.... ..."
Table 3: Network bandwidth consumption for common file-system operations. Shows the total number of bytes transmitted between all the nodes for each operation. Overhead shows the percentage of the bandwidth used by harbingers and duplicate updates.
"... In PAGE 11: ... Instead, its network usage is near- optimal, with less than 2% of the bandwidth wasted. Table3 shows network bandwidth consumption for com- mon file-system update operations. Operations such as cre- ating a file or writing one byte show a high percentage of overhead, since they are sent directly without harbingers, but they have only a minor impact on the overall wasted bandwidth since their size is small.... ..."
Table 4: Comparing the number of messages transmitted and number of replicated copies available for di erent coherence protocols on a network with 5 sites. Control (cntl) messages are used to send update or invalidation messages. Data messages are used to send the entire page data to another site.
1997
"... In PAGE 59: ...Table4 along with the number of replicated copies of a page (per operation). The total number of copies that exist is one plus the number of replicated copies.... In PAGE 59: ... As a result, the page remains in write mode, and invariant (1) is not enforced. Table4 compares the behavior of WB, WI, BR(2,5), and BR(3,5) for the same sequence of read and write operations as shown in Figure 8. The number of messages transmitted and replicated copies available are shown.... In PAGE 60: ... 1. Competitive operation costs - By inspecting the total number of messages transmitted by the di erent protocols in Table4 , we observe that BR costs as much as WI and WB. After having scaled the network to a larger size, Table 5 shows that BR transmits less messages than WB and only a few more control messages than WI.... ..."
Cited by 1
Table 1: An excerpt of server packet trace At the start of this episode, cwnd = 2; ssthresh = 1; cong count = 2, the server sends a fresh cwnd of packets out with the retransmission timer set for packet 1. The transmit queue includes packets 1 and 2. After some time, the ACK for packet 1 arrives and cwnd, ssthresh, and cong count are updated accordingly. RTT estimation and RTO computation are performed with RTO value updated to 390ms. New data packets 3 and 4 are sent out. The transmit queue is shifted and now it consists of packets 2, 3, and 4. The retransmission timer for packet 2 is set to 0.390 second. When this timer goes o at time 12.729123 + 0.390 second, cwnd, ssthresh, and cong count are recomputed, and packet 2 is resent with timer set to 0.780 second.
1999
"... In PAGE 19: ... We need to know when an ACK (as a separate packet or piggybacked) arrives in order to trigger the update of cwnd related information and the shift of the transmit queue. An excerpt of server side tcpdump trace ( Table1 ) illustrates the e ect of transmit queue. This episode is part of the trace when transferring 50KB le with delay = 0 and loss rate = 21%.... ..."
Cited by 2
Table 1. Performance statistics network connection between the client and the server is slower. However, the size of the data transmitted and recorded in the remote tests appears to be smaller. This is because the RFB client is adaptive to its computing environment, i.e. when it detects a slower connection, it sends less update requests to the RFB server. In other words, there exists a trade-off between the network bandwidth and the granularity of the frame buffer updates.
1998
"... In PAGE 4: ... In other words, there exists a trade-off between the network bandwidth and the granularity of the frame buffer updates. For comparison, we have recorded the tasks using the equivalent MPEG video, and the results are as listed in the last two rows of Table1 . The X Time is the time taken by the same user to complete the same tasks on a local X display and the MPEG Size is the size of the resulting MPEG stream generated by an MPEG encoder [16].... In PAGE 4: ... When the screen images are not complex, as in Task1 and Task2, our medium can save up to 80% of the storage space. We observe that it takes a longer time to complete the tasks with our applications than with a local X display, sometimes three times as much ( Table1 ). This is partially because of the heavy traffic involved between the RFB client/server, and partially because of the slow drawing mechanism of the Java programming language [6].... ..."
Cited by 6
Table 1. Performance statistics network connection between the client and the server is slower. However, the size of the data transmitted and recorded in the remote tests appears to be smaller. This is because the RFB client is adaptive to its computing environment, i.e. when it detects a slower connection, it sends less update requests to the RFB server. In other words, there exists a trade-off between the network bandwidth and the granularity of the frame buffer updates.
1998
"... In PAGE 4: ... In other words, there exists a trade-off between the network bandwidth and the granularity of the frame buffer updates. For comparison, we have recorded the tasks using the equivalent MPEG video, and the results are as listed in the last two rows of Table1 . The X Time is the time taken by the same user to complete the same tasks on a local X display and the MPEG Size is the size of the resulting MPEG stream generated by an MPEG encoder [16].... In PAGE 4: ... When the screen images are not complex, as in Task1 and Task2, our medium can save up to 80% of the storage space. We observe that it takes a longer time to complete the tasks with our applications than with a local X display, sometimes three times as much ( Table1 ). This is partially because of the heavy traffic involved between the RFB client/server, and partially because of the slow drawing mechanism of the Java programming language [6].... ..."
Cited by 6
Table 3 Percentage of transmitted packets for DPS run with different time series forecasting methods
in sensor
"... In PAGE 7: ...1. Gains in update rate Table3 reports the percentage of packets sent when running the DPS with the CM and AR models with orders from 1 to 5 (AR1; .... In PAGE 8: ...Table3 contains the model that yielded the lowest update rate, and that was consequently selected by the AMS procedure. 5.... ..."
Table 4: Number of messages transmitted per hour for ICP and Cache Digests. Of course, Cache Digests require signi cantly fewer messages. Cache Digests use about one message per digest update. For this experiment, digests were expired with one hour intervals. Thus the number of digests received is approximately the number of cache peers that support Cache Digests. We send more digest messages than we receive because some messages go to peers which are either dead or do not have Cache Digest support.
1998
Cited by 83
Table 4: Number of messages transmitted per hour for ICP and Cache Digests. Of course, Cache Digests require signi cantly fewer messages. Cache Digests use about one message per digest update. For this experiment, digests were expired with one hour intervals. Thus the number of digests received is approximately the number of cache peers that support Cache Digests. We send more digest messages than we receive because some messages go to peers which are either dead or do not have Cache Digest support.
1998
Cited by 83
Results 1 - 10
of
7,742
