Scribd
Upload
409 views

Avg Data Rate To Browsers: 1.50 Mbps

The document summarizes the assumptions and consequences of a network configuration with a 1.54 Mbps access link between an institutional router and origin servers. Given an average request rate of 15 requests per second and average object size of 100K bits, the access link utilization would be around 99% leading to congestion and delay in the minutes. Two solutions are proposed: 1) Upgrade the access link bandwidth to 154 Mbps to reduce utilization to 9.7% but this is costly. 2) Install a web cache with a 40% hit rate to serve 60% of requests locally, reducing access link utilization to 58% and total delay to around 1.2 seconds, providing a cheaper and faster alternative.

Uploaded by

Rayhan Ahmed
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
409 views

Avg Data Rate To Browsers: 1.50 Mbps

The document summarizes the assumptions and consequences of a network configuration with a 1.54 Mbps access link between an institutional router and origin servers. Given an average request rate of 15 requests per second and average object size of 100K bits, the access link utilization would be around 99% leading to congestion and delay in the minutes. Two solutions are proposed: 1) Upgrade the access link bandwidth to 154 Mbps to reduce utilization to 9.7% but this is costly. 2) Install a web cache with a 40% hit rate to serve 60% of requests locally, reducing access link utilization to 58% and total delay to around 1.2 seconds, providing a cheaper and faster alternative.

Uploaded by

Rayhan Ahmed
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Assumptions: Consequences:

avg object size: 100K bits LAN utilization: 15%


avg request rate from browsers to origin servers:15/sec Access link utilization = 99%
avg data rate to browsers: 1.50 Mbps Total delay = Internet delay + access delay +
RTT from institutional router to any origin server: 2 sec LAN delay
(Internet delay) = 2 sec + minutes + usecs
Access link rate: 1.54 Mbps

Consequences elaboration:

Traffic Intensity on LAN Utilization =15%?

15 rec/sec * (100,000 bits/req) / 10,000000 bits/sec = 0.15 = 15% [Note: 10 Mbps LAN = 10,000000 bps]

Traffic Intensity on Access Link utilization = 99% (approximately)?

15 rec/sec * (100,000 bits/req) / 1540000 bits/sec = 0.9740 = 97% [i.e. close to 99%] (approximately) [Note: 1.54
Mbps Access link = 1540000 bps]

Since the access link is being utilized around 97-99%, there will be a congestion, hence delay will be more.

Total delay = Internet delay (2 sec) + Access delay (since intensity is 97-99%, which is high, it will be in order of
minutes, which is also a larger delay in networking (NOT OK)) + LAN delay (typically is milliseconds (OK))

= ~ 2 seconds ++

SOLUTIONS:

1. We can upgrade the access link. If updated with a larger bandwidth, it will reduce the traffic intensity on
the access link and consequently delay will be reduced in the access link.

Let us, increase the bandwidth of access link from 1.54 Mbps to 154 Mbps.

The access link intensity becomes = 15 rec/sec * (100,000 bits/req) / 154000000 bits/sec = 0.09740 = 9.7%.

Access Link utilization is reduced.


Total delay = Internet delay (2 sec) + Access delay (since intensity is 9.7%, which is low, it will be in order of
milliseconds (OK)) + LAN delay (typically is milliseconds (OK))

= ~ 2 seconds ++

However, increasing access link bandwidth is COSTLY. Since, we need to pay for this large bandwidth every
month.

2. We can install Web cache (proxy server) in the institutional network.

Suppose after installing the proxy server in the LAN, the cache-hit rate is 0.4 (40%), 60% requests are satisfied at
origin servers.

Now 60% of requests use the access link, instead of 97-99%.

Data rate to browsers over the access link = 0.6*1.50 Mbps = 0.9 Mbps

Hence, access link utilization becomes = 0.9/1.54 Mbps = 0.584 = 58.4% (reduced than original 97-99%)

Now, Total delay = 0.6 * (delay from origin servers) + 0.4 *(delay when satisfied at web cache) = 0.6* 2.01 +
0.4* (~ milliseconds) = ~ 1.2 seconds (which is lower than the delay with increased bandwidth of 154 Mbps also
(VERY GOOD-CHEAPER- FASTER)).

You might also like