I.J.

Information Technology and Computer Science, 2013, 11, 42-53

Published Online October 2013 in MECS (http://www.mecs-press.org/)
DOI: 10.5815/ijitcs.2013.11.05

A Proposed Model for Web Proxy Caching

Techniques to Improve Computer Networks
Performance
Nashaat el-Khameesy
Prof. and Head of Computers & Information systems Dept- SAMS, Maady Cairo, Egypt
E-mail: [email protected]

Hossam Abdel Rahman Mohamed

Computer & Information System Dept- SAMS, Maady Cairo, Egypt
E-mail: [email protected], [email protected]

Abstract— one of the most important techniques for are located between users and web sites for lessening of
improving the performance of web based applications is the response time of user requests and saving of
web proxy caching Technique. It is used for several network bandwidth. We also should build an efficient
purposes such as reduce network traffic, server load, caching approach in order to achieve better response
and user perceived retrieval delays by replicating time.
popular content on proxy caches that are strategically
Generally, we use web Proxy servers to provide
placed within the network. The use of web proxy
internet access to users within a firewall. For security
caching is important and useful in the government
reasons, companies run a special type of HTTP servers
organizations that provides services to citizens and has
called "proxy" on their firewall machines. When a
many branches all over the country where it is
Proxy server receives any requests from the clients, it
beneficial to improve the efficiency of the computer
forwards them to the remote servers intercepts the
networks performance, especially remote areas which
responses, and sends the replies back to the clients. Due
suffer from the problem of poor network
to we use the same proxy servers for all clients inside of
communication. Using of proxy caches provides all
the firewall in the same organization, these clients share
users in the government computer networks by reducing
common interests and they would probably access the
the amount of redundant traffic that circulates through
same set of documents and each client tends to browse
the network. It also provides them by getting quicker
back and forth within a short period of time, So this
access to documents that are cached. In addition, there
provides the effectiveness of using these proxies to
are a lot of benefits we can obtain from the using of
cache documents. Therefore, this will increase the hit
proxy caches but we will address them later. In this
ratio for a previously requested and cached document
research, we will use web proxy caching to provide all
on the proxy server in the future. In addition to web
of the above benefits that we are mentioned above and
caching at proxy server saves network bandwidth, it
to overcome on the problem of poor network
also provides lower access latency for the clients.
communication in ENR (Egyptian National Railways).
We are going to use a scheme to achieve the integration Most Web proxy servers are still based on traditional
of forward web proxy caching and reverse proxy caching policies. These traditional caching policies only
caching. consider one factor in caching decisions and ignore the
other factors that have impact on the efficiency of the
Web proxy caching. Due to this reason these
Index Terms— Web Proxy Caching Technique, conventional policies are suitable in traditional caching
Forward proxy, Reverse Proxy like CPU caches and virtual memory systems, but they
are not efficient in Web caching area. [3], [4]
We use the proxy cache of the proxy server and it is
I. Introduction located between the client machines and origin server.
The work of the proxy cache is similar to the work of
One of the most well-known strategies for improving the browser cache in storing previously used web
the performance of Web-based system is Web proxy objects. The difference between them is the browser
caching by keeping Web objects that are likely to be cache which deals with only a single user, the proxy
used again in the future in location closer to user. The server services hundreds or thousands of users. The
mechanisms The Web proxy caching are implemented work of the proxy cache is as follow, when the proxy
at the following levels: client level, proxy level and server receives a request it checks its cache at first if the
original server level. [1], [2] it is known that proxy servers

Copyright © 2013 MECS I.J. Information Technology and Computer Science, 2013, 11, 42-53
 A Proposed Model for Web Proxy Caching Techniques to Improve Computer Networks Performance 43

request is found the proxy server sends the request B. The reduction in network traffic will make
directly to the client but if the request is not found the retrieving not cached contents faster because of
proxy server forwards the request to the origin server less congestion along the path and less workload
and after the origin server replies to the proxy server it at the server.
forwards the request to the client and also save a copy
from the request in local cache for future use. The proxy  Web proxy caching reduces the workload of the
caching is used to reduce user delays and to reduce origin Web server by caching data locally on the
Internet congestions it is widely utilized by computer proxy servers over the wide area network.
network administrators, technology providers, and  The robustness and reliability of the Web service is
businesses. [5], [6], [7] enhanced because in case the origin server in
The proxy server uses its filtering rules to evaluate unavailable due to any failure in the server itself or
the request, so it may use IP address or protocol to filter any failure in the network connection between the
the traffic. If the request is valid by the filter, the proxy proxy server and the origin server, the proxy server
provides the content by connecting to the origin server will retrieve the required data from its local cache.
and requesting the service on behalf of the client in case  Web caching has a side effect that allows us a chance
the required content is not cached on the proxy server. to analyze an organization's usage patterns.
The proxy server will return the content directly to the
client if it was cached before by the proxy server In addition to proxy caches provide a significant
We must consider the following problems before reduction in latency, network traffic and server load,
applying web proxy caching: they also produce set of issues that must be considered.

Size of Cache: In traditional architectures each proxy  A major disadvantage is the resend of old documents
server keeps records for data of all other proxy servers. to the client due to the lack of proper proxy updating.
This will lead in increasing in cache size and if cache This issue is the focus of this research.
size becomes large this will be a problem because as  A single proxy is a single point of failure.
cache size is larger, Meta data become difficult to be
managed. [8]  A single proxy may become a bottleneck. A limit has
to be set for the number of clients a proxy can serve.
Cache Consistency: We should ensure that Cache
Consistency is verified to avoid Cache Coherence
Therefore, in all government institution those provide
problem. Cache Consistency means when a client send
services to citizen, we must be searched about methods
requests for data to proxy server that data should be up-
and solutions to enhance the efficient of services
to-date. [9]
delivery , and as we know that most places away from
Load balancing: There is must be a limit for number Cairo state is facing failure in the network because the
of connections to certain proxy server to avoid the lack of infrastructure and possibilities of the services
problem of overloaded only one server than the other in provider (ISP).
case we use load balancing. [10]
There has been a lot of research and enhancement in
Extra Overhead: When all the proxy servers keep computer technology and the Internet has emerged for
the records of all the other proxy servers, this will lead the sharing and exchange of information. There has
to extra overload in the system which already produces been a growing demand for internet-based applications
congestion on all proxy servers. This extra head due to that generates high network traffic and puts much
each proxy server in the system must check the validity demand on the limited network infrastructure. We can
of its data with respect to all other proxy servers. [11] use addition of new resources to the network
infrastructure and distribution of the traffic across more
In addition to the proxy cache provide some
resources as a possible solution to the problems of
advantages such as a reduction in latency, network
growing network traffic.
traffic and server load, it also provides some more
advantages Using of proxy caches in the government computer
networks is useful to the server administrator, network
 Web proxy caching decreases network traffic and administrator, and end user because it reduces the
network congestion and this automatically reduces amount of redundant traffic that circulates through the
the consumption of bandwidth network. And also end users get quicker access to
 Web proxy caching reduces the latency because of documents that are locally cached in the caches.
the followings: However, there are additional issues that need to be
considered by using of proxies. In this study, we will
A. When a client sends to the proxy server a focus on Web proxy caching since it is the most
request already cached in the proxy server so in common caching method.
this case the proxy server will reply directly to
the client instead of send the request to the
origin server.

Copyright © 2013 MECS I.J. Information Technology and Computer Science, 2013, 11, 42-53
44 A Proposed Model for Web Proxy Caching Techniques to Improve Computer Networks Performance

1.1 Problem Statement When the forward proxy receives a request from the
client, the request can be rejected by the forward server
The governmental organizations which provide
or allowed to pass to the internet or [13] retrieved from
services to citizen must target efficient and more
the cache to the client. The last one reduces the network
reliable services while keeping cost-effective criteria.
traffic and improves the performance.
Throughout this paper we consider the case of the
Egyptian National Railway (ENR) datacenter which On the other hand, the forward proxy treats the
serve many applications supported to many branches requests by two different ways according to the requests
spreaded allover Egypt which are quite faraway from are blocked or allowed. In case the request is blocked,
Cairo state. Current infrastructure faces so many the forward proxy returns an error to the client. In case
challenging problems leading to poor reliability as well the request is allowed, the forward proxy checks either
as ineffective services and even more discontinuity of the request is cached or not; if it is cached, the forward
such services even at the headquarter datacenter. The proxy returns cached content to the client. If it is not
main attributes of the problems facing the ENR network cached, the forward proxy forwards the request to the
are summarized as following: internet then returns the retrieved content from the
intent to the client.
 A major problem of the remote site is unstructured
and their heavy network traffic. The above figure explains the work of the forward
proxy in case the request is allowed but not cached on
 The network overloading might result in the loss of the forward proxy A. the forward proxy will send the
data packets request to the server on the internet then the server on
 The origin servers loaded most of the time the internet return the required content to the forward
proxy and finally the forward proxy return the received
 Transmission delay – normal traffic data but low content to the client and cached it on its cache for future
speed of the line. and same request. The cached content on the forward
 Queuing delay due to huge network traffic proxy will reduce the network traffic in the future and
actually improves the performance of the whole system.
 Slow the services that provided to citizens.
 Processing delay – due to any defection of the
2.2 Reverse Proxy Caching
network device
The other common form of a proxy server is a
 Network faults can cause loss of data reverse proxy; it performs the reverse function of the
 Broadcast delay – due to presence of broadcasting on forward proxy, it is used to forward requests from an
network internet network to the intranet network. [14] This
provides more security by preventing any hacking or an
illegal access from the clients on the internet to
important data stored on the content servers on the
II. Proxy Caching Overview intranet network. By the same way, if the required
Caches are often deployed as forward proxy caches, content is cached on the reverse proxy, this will reduce
reverse proxy caches or as transparent proxies. the network traffic and improves the performance.[15]

2.1 Forward Proxy Caching

The most common form of a proxy server is a

forward proxy; it is used to forward requests from an
intranet network to the internet. [12]

Fig. 2.2: Reverse proxy caching

The advantages of reverse proxy are

 Solving single point of failure problem by using load
balancing for content servers.
 Reducing the traffic on the content servers in case the
request is blocked by the reverse proxy. In this case
the request is rejected directly by the reverse proxy
Fig. 2.1: Forward proxy caching
without interrupt the content servers.

Copyright © 2013 MECS I.J. Information Technology and Computer Science, 2013, 11, 42-53
 A Proposed Model for Web Proxy Caching Techniques to Improve Computer Networks Performance 45

 Reducing the bandwidth consumes by blocked The main object of using caching hierarchy is to
requests as it is blocked directly by reverse proxy reduce the network traffic and minimize the times that a
before reaching to the content servers. proxy server needs to contact to the content server in
the internet or in the internal network to provide the
The function of the reverse proxy is the same as the client with needed content information .These multiple
function of the forward proxy except the request is caches works in that manner in case of forward proxy,
initiated from the client on the internet to the content at first the client initiate a request to the bottom level
servers in the internal network. At first, the client on the cache. If the needed content information is found and
internet sends a request to the reverse proxy. If the cached on it, it returns this information to the client
request is blocked, the reverse proxy returns an error to directly. If this information is not cached on it, it will
the client. If the request is allowed, the reverse proxy forward the client request to the next level cache that is
checks if the request is cached or not. In case the institutional. If this cache found the needed information
request is cached, the reverse proxy returns the content cached on it, it will return it to bottom level cache then
information directly to the client on the internet. In case the bottom level cache returns them to the client. If the
the request is not cached, the reverse proxy sends the needed information is not cached on it, it will forward
request to the content server in the internal network then the request to regional level. If the needed information
resends the retrieved content from the content server to is cached on it, it will return the needed information to
the client and also cached the content information from the institutional level cache then the institutional level
the content server for future requests to same content cache returns them to the bottom level cache and finally
information [16] bottom level cache returns them to the client. If the
needed information is not found not found on it, it will
forward the request to the last level of cache that is
2.3 Transparent Caching national, if the needed information is found on that
cache, it works the same way as above till the
Transparent proxy caching eliminates one of the big information reach to the client. If the needed
drawbacks of the proxy server approach: the information is not cached on that cache, it will forward
requirement to configure Web browsers. Transparent the request to the content server on the internet and also
caches work by intercepting HTTP requests and repeat the same steps as above till the information
redirecting them to Web cache servers or cache clusters. reached to the client.
[17]
This style of caching establishes a point at which
different kinds of administrative control are possible; In case of the reverse proxy, the same steps above are
for example, deciding how to load balance requests repeated except the request will forward by reverse way
across multiple caches. There are two ways to deploy as in the forward proxy. Here, the request will forward
transparent proxy caching: at the switch level and at the from national level cache then to then to regional then
router level. [18] to institutional bottom and finally to the content server
in the internal network. The important note in caching
Router-based transparent proxy caching uses policy- hierarchy either in case of the forward proxy or the
based routing to direct requests to the appropriate reverse proxy is each cache receives information from
cache(s). For example, requests from certain clients can another level cache will cache a copy from that
be associated with a particular cache. [19] information for future need to the same request.
In switch-based transparent proxy caching, the
switch acts as a dedicated load balancer. This approach
is attractive because it reduces the overhead normally
incurred by policy-based routing. Although it adds extra
cost to the deployment, switches are generally less
expensive than routers. [20]

III. Proxy Caching Architecture

The following architectures are popular: hierarchical,
distributed and hybrid.

3.1 Hierarchical Caching Architecture

Caching hierarchy consists of multiple levels of
caches. In our system we can assume that caching
hierarchy consists of four levels of caches. These levels Fig. 3.1: Hierarichal caching architecture
are bottom, institutional, regional, and national levels
[21]

Copyright © 2013 MECS I.J. Information Technology and Computer Science, 2013, 11, 42-53
46 A Proposed Model for Web Proxy Caching Techniques to Improve Computer Networks Performance

3.2 Distributed Caching Architecture other systems. This technique has been implemented in
the OceanStore prototype (Eaton et al., 2004). [25]
In distributed Web caching systems, there are no
other intermediate cache levels than the institutional
caches, which serve each others' misses. In order to
decide from which institutional cache to retrieve a miss IV. Design Goals & Proposed Architecture
document, all institutional caches keep meta-data
information about the content of every other To improve the computer network performance,
institutional cache. With distributed caching, most of decrease the workload for data center and ensure
the traffic flows through low network levels, which are continual service improvement, we aim to design
less congested. In addition, distributed caching allows efficient mechanisms for reducing the workload of a
better load sharing and are more fault tolerant. data center and business Continuity verification and
Nevertheless, a large-scale deployment of distributed achieve the following goals:
caching may encounter several problems such as high  Reduces network bandwidth usage consumption
connection times, higher bandwidth usage, which leads to reduce network traffic and network
administrative issues, etc. [22] congestion
There are several approaches to the distributed  Decrease the number of messages that enter the
caching. Internet Cache Protocol (ICP), which supports network by satisfying requests before they reach the
discovery and retrieval of documents from neighboring server.
caches as well as parent caches. Another approach to
distributed caching is the Cache Array Routing protocol  Reduces loads on the origin servers.
(CARP), which divides the URL-space among an array  Decreases user perceived latency
of loosely coupled caches and lets each cache store only
the documents whose URL are hashed to it. [23]  Reduced page construction times during both normal
loading and peak loading
 If the remote server is not available due to a server
3.3 Hybrid Caching
\crash" or network partitioning, the client can obtain
A hybrid cache scheme is any scheme that combines a cached copy at the proxy.
the benefits of both hierarchical and distributed caching
architectures. Caches at the same level can cooperate
together as well as with higher-level caches using the 4.1 Proposed Architecture
concept of distributed caching. [24]
We define before two types of proxies, the forward
A hybrid caching architecture may include proxy and the reverse proxy. The forward proxy is used
cooperation between the architecture's components at to forward clients from the clients on the internal
some level. Some researchers explored the area of network to the content server in the internet. The
cooperative web caches (proxies). Others studied the reverse proxy is used to forward requests from the
possibility of exploiting client caches and allowing clients in the internet to the content server in the
them to share their cached data. internal network. Fig 5-1 shows that the forward proxy
serves as a servant for internal clients and as a cache
One study addressed the neglect of a certain class of
because it cached the content received from the content
clients in researches done to improve Peer-to-Peer
server on the internet. So for any the same repeated
storage infrastructure for clients with high-bandwidth
request, the forward server can return the cached
and low latency connectivity. It also examines a client-
content on it to the client directly without backing again
side technique to reduce the required bandwidth needed
to the content server. On the same time the forward
to retrieve files by users with low-bandwidth.
proxy does an important rule as it hides the internal
Simulations done by this research group has proved that
clients from outside world as the request is initiated
this approach can reduce the read and write latency of
from the forward proxy.
files up to 80% compared to other techniques used by

Fig. 4.1: Proposed Architecture

Copyright © 2013 MECS I.J. Information Technology and Computer Science, 2013, 11, 42-53
 A Proposed Model for Web Proxy Caching Techniques to Improve Computer Networks Performance 47

Fig 4-1 also shows the reverse proxy that used to 2 If the content requested by Remote Site user is not
forward the requests from external clients to content contained in the Forward proxy cache, the request is
servers in internal network. In this case the reverse forwarded to an upstream Reverse proxy caching.
proxy makes encrypting content, compressing content,
3 If the upstream Reverse Proxy Cache has a valid
reducing the load on content servers. It also hides the
copy of the requested content in cache, the content is
responses from internal networks and as them come
returned to Forward proxy cache (Remote Site).
from the reverse proxy which increases the security. It
Forward proxy cache places the content in its own
also caches the content and forwards it directly to
cache and then returns the content to the Remote Site
clients if they repeated again without backing again to
user who requested the content.
the content server. Finally, we can use load balancing to
balance between content servers and in this case the 4 If the upstream Reverse proxy caching does not
reverse proxy and forward the request from the client to contain the requested content in its cache, it will
any of this content serves which increase the forward the request to the Web Application server.
availability of the system.
5 The Web Application server returns the requested
content to reverse proxy caching. Reverse proxy
caching places the content in cache.
4.2 Proposed Architecture Workflow
6 Web Application server returns the content to reverse
1 The Remote Site client sends a request for Web
proxy caching. Reverse proxy caching server places
Application content to the Forward proxy cache. If
the content in its cache. Reverse proxy caching server
Forward proxy caching contains a valid version of
returns the content from its cache to Forward proxy.
the Web Application content in its cache, it will
Forward proxy cache places the content in its own
return the content to the requesting user.
cache and then returns the content to the Remote Site
user who requested the content.

Fig. 4.2: Proposed workflow

 Requests returned from the application server.

V. Performance Analysis
In this chapter, we evaluate cache performance of  Requests returned from cache without verification.
web proxy caching for web applications and compare it  Requests returned from the application server,
to the case of not using web proxy caching at all. We updating a file in cache.
will monitor and evaluate the performance of web
proxy caching in three cases:  Requests returned from cache after verifying that
they have not changed.
 Without using web proxy caching.
 At the beginning of using web proxy caching.
5.1 Performance Metrics
 After certain period (one month) from using web
proxy caching. The seven main categories of performance metrics
are:
We will take in our consideration the following
parameters in evaluation process

Copyright © 2013 MECS I.J. Information Technology and Computer Science, 2013, 11, 42-53
48 A Proposed Model for Web Proxy Caching Techniques to Improve Computer Networks Performance

1. Cache Performance: how requested Web objects VI. Results

were returned from the Web Server cache or from the
In this section we will investigate the performance
network. It will be measured the according to
analysis of cache, Network traffic, Failure
2. Traffic: the amount of network traffic, by date, sent communication, Dropped packets and queue length
through Web Proxy including both Web and non-
Web traffic.
6.1 Cache Performance
3. Daily traffic: average network traffic through Web
Proxy at various times during the day. This report The cache performance results for each of the log
includes both Web and non-Web traffic. files are shown below. The percentage of requests
returned from cache without verification is high. It
4. Web Application Responses: how ISA Server
shows that between 38% of all requests result in a
responded to HTTP requests during the report period.
request returned from cache without verification, which
5. Failures communicating: Web proxy Cache is consistent with previously published results. Wills
encountered the following failures communicating and Mikhailov[26] reported that only 15% to 32% of their
with other computers during the report period. proxy logs result in requests returned from cache
without verification. Yin, et al.[27] revealed that 20% of
6. Dropped Packets: shows the users who had the
requests to the server are due to revalidation of cached
highest number of dropped network packets during
documents that have not changed. These results are
the report period Users that had the most dropped
consistent with the results found in our logs as
packets are listed first
discussed before. However, with the current logs, the
7. Queue Length: The System\Processor Queue Length number of requests returned from cache without
counter shows how many threads are ready in the verification has increased a little. This may be due to
processor queue, but not currently able to use the the duration of the analysis being longer for this study
processor. or to the use of different logs. It is assumed that a large
fraction of these frivolous requests are due to embedded
objects that do not change often.

Table ‎0-1: Cache Performance Results

Reques % of Total Total

Status
ts Requests Bytes
Objects returned from the application
20873 59.30 % 617.73 MB
server
Objects returned from cache without
13450 38.20 % 26.93 MB
verification
Objects returned from cache after
489 1.40 % 0.99 MB
verifying that they have not changed
Information not available 354 1.00 % 49.74 KB
Objects returned from the application
59 0.20 % 14.62 KB
server, updating a file in cache
Total 35225 100.00 % 645.72 MB

6.2 Traffic of the objects themselves. Hence, we can reflect the size
factor of web object. An average object size hit ratio
The results for average network traffic through web
reflects the factor of object size to object-hit ratio.
proxy caching Server at various times during the day at
Average-object Hit Ratio: The cache-hit ratio can
the beginning of using web proxy caching and after
indicate an object-hit ratio in web caching. The average
certain time from using web proxy caching are in Table
object-hit ratio can calculate an average value of an
below. The results indicate that the average processing
object-hit ratio on a requested page the performance is
time for handling the request is reduced by 43% after
evaluated by comparing an average object-hit ratio and
certain time of using web proxy caching because proxy
response time [28] Response Time gain factor (RTGF):
caches the previous visited pages and return them
This factor give you amount of advantage in web cache
directly to client without waste time to ask application
response time. [29]
server each time.
To reflect the physical environment of the network,
we have to consider factors influencing traffic. Of
various factors influencing traffic, object size is a factor

Copyright © 2013 MECS I.J. Information Technology and Computer Science, 2013, 11, 42-53
 A Proposed Model for Web Proxy Caching Techniques to Improve Computer Networks Performance 49

Table 6-2: Average Network Traffic

At the beginning of using Web Proxy Cache After certain time of using Web Proxy Cache

Average Average RTGF

Requests Total Bytes Cache Hit Ratio Requests Total Bytes Cache Hit Ratio
Processing Time Processing Time
1924 133.00 sec 3.61 GB 0.00 % 6297 58.90 sec 2.94 GB 1.00 % 55.7142
1708 126.40 sec 9.75 MB 0.00 % 6157 61.30 sec 37.91 MB 0.00 % 51.5031
1654 127.10 sec 8.33 MB 0.00 % 5954 62.30 sec 36.27 MB 0.00 % 50.9834
1927 115.20 sec 8.42 MB 0.00 % 6004 58.00 sec 36.19 MB 0.00 % 49.6527
1731 121.10 sec 8.41 MB 0.00 % 5934 57.90 sec 36.12 MB 0.00 % 70.1156
1713 121.60 sec 8.51 MB 0.20 % 5901 58.30 sec 35.80 MB 0.00 % 50.1187
1642 126.60 sec 8.36 MB 0.00 % 5899 57.30 sec 35.80 MB 0.00 % 54.6318
1835 128.70 sec 8.48 MB 0.00 % 6249 54.50 sec 38.34 MB 2.80 % 57.6543
1859 126.30 sec 9.13 MB 0.00 % 6014 57.60 sec 36.69 MB 0.30 % 54.3942
2455 135.10 sec 11.50 MB 0.00 % 7212 52.00 sec 40.92 MB 12.00 % 69.7113

6.3 Traffic by Time of Day

The following Table (6.3) summarizes average
network traffic through web proxy caching Server at
various times during the day.

Table 6-3: Traffic by Time of day

Average Requests Average Bytes Average Response T Average Response

Time Interval
Per Second Per Second ime for Cached Requests Time for Non Cached Requests

00:00 15.4 67.49 KB - 55.30 sec

00:15 17.1 13.15 MB - 56.90 sec
00:30 14.8 10.57 MB 0.00 sec
00:45 14.1 83.45 KB - 65.20 sec
01:00 16.1 75.27 KB 0.00 sec
01:15 16.2 69.21 KB 0.00 sec
01:30 14.5 92.42 KB 0.00 sec
01:45 14.0 77.47 KB - 59.20 sec
02:00 15.8 67.94 KB - 55.10 sec
02:15 16.9 68.87 KB - 56.80 sec
02:30 15.2 93.02 KB 0.00 sec
02:45 14.3 77.61 KB - 52.60 sec
03:00 16.1 70.17 KB 0.00 sec
03:15 16.7 70.64 KB 0.00 sec
03:30 15.2 93.35 KB 0.00 sec
03:45 14.5 77.47 KB - 51.90 sec
04:00 15.9 70.57 KB 0.00 sec
00:00 15.4 67.49 KB - 55.30 sec
00:15 17.1 13.15 MB - 56.90 sec
00:30 14.8 10.57 MB 0.00 sec

Copyright © 2013 MECS I.J. Information Technology and Computer Science, 2013, 11, 42-53
50 A Proposed Model for Web Proxy Caching Techniques to Improve Computer Networks Performance

As we shown in 6.3 Table the average response time results indicate that the percentage of success is
for non cached requests (returned from origin servers) increased.
55.30 sec for 67.49 KB in the first item, but the average
response time for cached requests was it 0.00 sec for
10.57 MB Which shows that the use web proxy cache 6.5 Failures Communication
has Significant impact in the speed of retrieval of
contents The results for how web proxy server encountered
the following failures communicating with other
computers during the report period. The results indicate
Table 6-4: Web Application Responses that the percentage of errors after using web proxy
At the beginning of using Web Proxy Cache
caching is enhanced by noticeable value
Response Requests HTTP Responses
Table ‎0-5: Failures Communication
Success 30910 87.80 %
Authorization failure 2565 7.30 % At the beginning of using Web Proxy Cache

Other 1094 3.10 % Error Type Errors % Total Errors

Object moved 656 1.90 % Firewall Service errors 165263 95.50 %
Object not found 0 0.00 % Web Proxy errors 7821 4.50 %
Total 35225 100.00 % Total 173084 100.00 %
After certain time of using Web Proxy Cache
After certain time of using Web Proxy Cache
Response Requests HTTP Responses
Error Type Errors % Total Errors
Success 43518 91.60 %
Firewall Service errors 13062 58.70 %
Authorization failure 2403 5.10 %
Web Proxy errors 9184 41.30 %
Other 976 2.10 %
Object moved 622 1.30 % Total 22246 100.0

Object not found 0 0.00 %

Total 47519 100.0
6.6 Dropped Packets
The result below shows the users who had the highest
6.4 Web Application Responses number of dropped network packets during the report
period. Users that had the most dropped packets are
The results for how web proxy caching server listed first. We can observe that the percentage of
responded to HTTP requests during the report period at dropped packets is reduced by the time.
the beginning of using web proxy caching server and
after certain time from using web proxy caching. The

Table 6-6: Dropped Packets

At the beginning of using After certain time of using

Web Proxy Cache Web Proxy Cache
% of Total % of Total
Dropped Dropped
User Dropped User Dropped
Packets Packets
Packets Packets
10.1.12.13 36857 23.40 % 10.1.12.13 692 11.90 %
10.1.12.14 35573 22.60 % 10.1.12.14 642 11.00 %
172.31.1.2 33909 21.50 % 172.31.1.2 183 3.10 %
172.31.1.1 31718 20.10 % 172.31.1.1 203 3.50 %
10.1.12.12 4752 3.00 % 10.1.12.12 794 13.60 %
10.1.12.20 2212 1.40 % 10.1.12.20 233 4.00 %
10.1.12.23 2211 1.40 % 10.1.12.23 211 3.60 %
10.1.12.22 2188 1.40 % 10.1.12.22 224 3.80 %
10.1.12.15 1536 1.00 % 10.1.12.15 565 9.70 %
10.1.19.12 1028 0.70 % 10.1.19.12 399 6.90 %
10.1.10.17 315 0.20 % 10.1.10.17 60 1.00 %
Total 152299 100.00 % Total 7612 100.00 %

Copyright © 2013 MECS I.J. Information Technology and Computer Science, 2013, 11, 42-53
 A Proposed Model for Web Proxy Caching Techniques to Improve Computer Networks Performance 51

6.7 Queue Length

After analyzing the network traffic, we found that
network traffic is very burst over a large range of
timescales and shows asymptotic self-similarity and
multi-fractal behavior at intermediate time scales. Our
study of request process of web-server shows that the
busy-period traffic is very similar to a Harm parameter
of around 0.7. [30] Also, several studies show that WAN
traffic is multifractal in nature. We captured a sample
from the state of network traffic of Giza and a Luxor
station during the problem is occurring. Figure (6.7, 6.8) Fig. 6-8: Traffic Analysis for Luxour -Station
shows that Giza line Speed: 2048.0 kbits/s and
Incoming packets 1990.9 kb/s (97.2%) ,Out packets
2001.3 kb/s (97.7%) and Louxur Line Speed: 2048.0 In Figure (6.7), (6.8) the multifractal behavior shows
kbits/s with Incoming 1931.3 kb/s (94.3%) and Out that there is a lot of requests arrive at the server [31] And
packets 1931.5 kb/s (94.3%) ,This means that the any new arrival process will cause heavy-tailed queue-
network overloading might result in the loss of data length distribution and as a result will lead to more
packets and queuing delay loads on servers and the server cannot able to process
any new request and return error to client this error
“server-too-busy”.‎Due‎to‎the‎above‎problem,‎a‎multiple‎
complaint from the system administrators for the origin
servers loaded most of the time, and always those
problems at certain times when the citizens buying
tickets. also The widely use of web-based user
interfaces for conducting business on the ENR network
has brought to focus the problem of dealing with
overloads to which the web servers (especially those
that form the front end servers).

Fig. ‎0-7: Traffic Analysis for Giza-Station

Table 6-7: CPU Utilization and Avg Queue Length

At the beginning of using Web Proxy Cache After certain time of using Web Proxy Cache
Overload Processor Avg Queue Length Overload Processor Avg Queue Length
(requests/s) Utilization % (Reqs) (requests/s) Utilization % (Reqs)
36.857 80.95 37.98 % 6.92 692 0.97
35.225 73.67 47.54 % 6.42 642 1.02
30.910 74.67 19.02 % 1.83 183 0.21
31.718 79.67 66.41 % 2.03 203 0.37

case the request is blocked by the reverse proxy. In this

VII. Conclusions
case the request is rejected directly by the reverse proxy
We proposed a hybrid of web proxy caching without interrupt the content servers and reducing the
architecture to improve the performance of computer bandwidth consumes by blocked requests as it is
network, by integrating forward proxy caching blocked directly by reverse proxy before reaching to the
technique and reverse proxy caching technique. The content servers. The forward proxy serves proxy is used
forward proxy is used to forward clients from the clients as a servant for internal clients and as a cache because it
on the internal network to the content origin server. We cached the content received from the content on origin
are use the forward proxy caching to reduce the network server. So for any the same repeated request, the
traffic, reduced page construction times during both forward server can return the cached content on it to the
normal loading and peak loading , decreasing dropped client directly without backing again to the content
packets. The reverse proxy is used to forward requests server. On the same time the forward proxy does an
from the clients in the internet to the content server in important rule as it hides the internal clients from
the internal network. We are use the reverse proxy outside world as the request is initiated from the
caching Reducing the traffic on the content servers in forward proxy.

Copyright © 2013 MECS I.J. Information Technology and Computer Science, 2013, 11, 42-53
52 A Proposed Model for Web Proxy Caching Techniques to Improve Computer Networks Performance

Web proxy caching is effective solution to lessen Faculty, Cairo University The authors thankfully
Web service bottleneck, reduce traffic over the network acknowledge the support provided by Trans IT
and improve scalability of the Web Application system. Datacenter Operation Team
This research presents an overview of the Web Proxy
caching Technique design and management of a cache
consistency algorithm and examines the proposed
solutions to improve computer networks performance in References
recent literature. [1] J. B. Pati , B. V. Pawar / Improving Performance
on WWW using Intelligent Predictive Caching for
 We proposed an effective and flexible for ENR
Web Proxy Servers, IJCSI International Journal of
network by using web proxy caching technique to
Computer Science Issues, Vol. 8, Issue 1, January
Reduces network bandwidth usage, Decrease the
2011
number of messages that enter the network, Reduces
loads on the origin servers. [2] Chen, T.: Obtaining the optimal cache document
replacement policy for the caching system of an
 We have designed and implemented our web proxy
EC website. European Journal of Operational
caching in a system involving two major components.
Research 181(2), 828–841, 2007
A forward proxy cache for users, Reverse proxy
cache‎for‎server’s‎application. [3] Romano, S., ElAarag, H.: A neural network proxy
cache replacement strategy and its implementation
 We have performed an analytical investigation of in the Squid proxy server. Neural Computing &
four proposed consistency algorithms to identify how Applications 20(1), 59–78, 2011
read and writes could affect the operation of these
algorithms and to provide taxonomy of when the [4] Cobb, J., ElAarag, H.: Web proxy cache
algorithms are most appropriate. replacement scheme based on back-propagation
neural network. Journal of Systems and Software
 We evaluated our algorithm and compared it with 81(9), 1539–1558 ,2008
existing and proposed algorithms using a testbed of
cluster nodes to run an emulation of the Web. This [5] Kaya, C.C., Zhang, G., Tan, Y., Mookerjee, V.S.:
included the used of actual Web proxies and servers An admission-control technique for delay
that are driven by a workload of server and proxy reduction in proxy caching. Decision Support
logs. Systems 46(2), 594–603 , 2009

 We investigated the amount of unnecessary traffic [6] Kumar, C., Norris, J.B.: A new approach for a
that is generated on the Web application today. This proxy-level web caching mechanism. Decision
provides estimates of the possible reduction of Support Systems 46(1), 52–60 ,2008
unnecessary messages that could occur if our [7] Kumar, C.: Performance evaluation for
approach were available for use in the Web implementations of a network of proxy caches.
application today. Decision Support Systems 46(2), 492–500 ,2009
 Our background study determines whether significant [8] Rajeev Tiwari et al. /Robust Distributed Web
network savings would occur if server invalidation caching: A Dynamic clustering Approach ,
played some part in the maintenance of consistency. International Journal of Engineering Science and
These evaluations show the potential for substantial Technology (IJEST), Vol. 3 No. 2 Feb 2011
improvement offered by our proposal.
[9] Ali, W., Shamsuddin, S.M., Ismail, A.S.: Web
 A survey of existing techniques shows that many proxy cache content classification based on
individual techniques are good in their own context, support vector machine. Journal of Artificial
but we have shown in our results that an integration Intelligence 4(1), 100–109 ,2011
of these techniques performs significantly better
when compared to the techniques in isolation. [10] Tiwari‎Rajeev‎and‎Khan‎Gulista,‎“Load‎Balancing‎
in Distributed Web Caching: A Novel Clustering
 We implemented a web proxy-based content Approach”,‎ Proc.‎ of‎ ICM2ST-10, International
accelerator, which combines the best features of Conference on Methods and models in science and
fragment - caching and proxy-caching, with the technology pp. 341-345, November 6, vol.1324,
intention of reducing bandwidth requirement. 2010
[11] Rajeev‎ Tiwari,‎ Gulista‎ khan,‎ “Load‎ Balancing‎
through‎ distributed‎ Web‎ Caching‎ with‎ clusters”,‎
Acknowledgements Proceeding of the CSNA 2010 Springer, pp 46-54,
Chennai, India, 2010
I would like to express our gratitude to Prof. DR/
Nashaat El- Khameesy Computers & Information [12] Takaaki Komura, Hiroaki Sano, Noritoshi Demizu,
systems Dept- Sadat Academy, Asmaa Ahmed - Ken Makimura, "Design and Implementation of
Communications and Electronic Dept- Engineering Web Forward Proxy with Shibboleth

Copyright © 2013 MECS I.J. Information Technology and Computer Science, 2013, 11, 42-53
 A Proposed Model for Web Proxy Caching Techniques to Improve Computer Networks Performance 53

Authentication," saint, pp.321-326, 2011 [24] Jinsuk Baek,Gurpreet Kaur and Junghoon Yang /A
IEEE/IPSJ International Symposium on New Hybrid Architecture for Cooperative Web
Applications and the Internet, 2011 Caching, journal of ubiquitous convergence and
technology, VOL. 2, NO. 1, MAY 2008
[13] Wang, Xiao-Ming Fu, Fang-Wei /Forward secure
proxy signature scheme, Tongxin Xuebao (Journal [25] S. Rhea, P. Eaton, D. Geels, H. Weatherspoon, B.
on Communications). Vol. 26, no. 11, pp. 38-42. Zhao, and J. Kubiatowicz. Pond: the OceanStore
Nov. 2005 prototype. In Proc. Of USENIX FAST, pages 1–14,
Mar. 2003.
[14] P. Wurzinger,C. Platzer,C. Ludl,E. Kirda, C.
Kruegel /SWAP: Mitigating XSS attacks using a [26] C. E. Wills and M. Mikhailov. Toward a better
reverse proxy, Pages 33-39 , IWSESS '09 understanding of web resources and server
Proceedings of ICSE Workshop on Software responses for improved caching. pages 1231-1243,
Engineering for Secure Systems,2009 1999
[15] Lin, C.-H., Liu, J.-C., and Lien, C.-C.,‎“Detection‎ [27] J. Yin, M. Dahlin, and A. Iyengar. Engineering
Method Based on Reverse Proxy against Web server-driven consistency for large- scale dynamic
Flooding‎ Attacks,”‎ 8th‎ International‎ Conference‎ web services. In Proceedings of the 10th
on Intelligent Systems Design and Applications International World-wide Web Conference, pages
(ISDA-2008), pp.281- 284. Kaohsiung City, 46-57, Hong Kong, May 2001
Taiwan. (2008)
[28] Yun Ji Na ,Seok KO2 / Design and performance
[16] Peter Sommerlad /Reverse Proxy Patterns, (c) Analysis of the Web Caching Algorithm for
2003 by Peter Sommerlad,Erlenstrasse 79, CH- Performance Improvement of the Response Spee,.
8832 Wollerau, Switzerland,2003 T.-J. Cham et al. (Eds.): MMM 2007, LNCS 4352,
Part II, pp. 686 – 693, 2007. Springer-Verlag
[17] C. Amza and G. Soundararajan and E. Cecchet.
Berlin Heidelberg 2007
Transparent Caching with Strong Consistency in
Dynamic ContentWeb Sites. In ICS, 2005. [29] Dhawaleswar rao. ch / study of the web caching
algorithms for performance improvement of the
[18] Laura Abba, Marina Buzzi, and Massimo Ianigro.
response speed, Journal of Computer Science and
Introducing transparent Web caching in a local
Engineering (IJCSE) ,Vol. 3 No. 2 Apr-May 2012
area network. In Proc. of Intl. CMG Conference,
pages 99–110, 2000. [30] Lin, C.-H., Liu, J.-C., and Lien, C.-C.,‎“Detection‎
Method Based on Reverse Proxy against Web
[19] Wenting‎Tang‚‎Fan‎Du‚‎Matt‎W.‎Mutka‚‎Lionel‎M.‎
Flooding‎ Attacks,”‎ 8th‎ International‎ Conference‎
Ni‚‎ and‎ Abdol-Hossein Esfahanian. Supporting
on Intelligent Systems Design and Applications
global replicated services by a routing-metric-
(ISDA-2008), pp.281- 284. Kaohsiung City,
aware DNS. In Proceedings of the 2nd
Taiwan. (2008)
International Workshop on Advanced Issues of E-
Commerce and Web-Based Information Systems [31] Peter Sommerlad /Reverse Proxy Patterns, (c)
(WECWIS‎2000)‚‎pages‎67–74‚‎June‎2000 2003 by Peter Sommerlad,Erlenstrasse 79, CH-
8832 Wollerau, Switzerland,2003
[20] L Abba, M Buzzi, D Pobric, M Ianigro/
Introducing transparent web caching in a local area
network, Centre National de la Recherche
Scientifique Paris, France, France ©2000 Authors’ Profiles
[21] Rodriguez, P., Spanner, C., Biersack, and E., Dr Nashaat el-Khameesy: Prof. and
Analysis of Web Caching Architectures: Head of Computers & Information
Hierarchical and Distributed Caching, IEEE/ACM systems Chair, Sadat Academy, Maady,
Transactions on Networking, Vol. 9(4):404-418, Cairo-Egypt.
August 2001.
[22] S. V. Nagaraj, web caching and its applications,
page 24, Caching Architectures, print ©2004
kluwer Academic Publishers- new york , boston- Hossam Abdel Rahman Mohamed:
2004 Master of computer science &
[23] Suresha , Caching Techniques for Dynamic Web information system - Sadat Academy
Servers, Doctor of Philosophy Department of - Dept Computer and Information
Computer Science and Automation ,Indian System, Maady, Cairo, Egypt - He is
Institute of Science Bangalore 560 012 INDIA - currently position Data Center
June 2007 Operation Team leader at Trans IT
Company - Ministry of Transportation

Copyright © 2013 MECS I.J. Information Technology and Computer Science, 2013, 11, 42-53