So you're looking to optimize your WAN for your business and need to know what your options are. Upgrading your bandwidth to MPLS or Ethernet isn't necessarily the solution. Bandwidth utilization is the key for optimizing your WAN.
Packeteer, Riverbed and Juniper are WAN optimization systems that do an excellent job with latency, especially if CIFS WAN-unaware situations such as opening, copying and modifying files using Windows are the main problem.
To identify the delay, you need to find the path your IP packets take. A satellite connection versus a landline connection (such as earthbound wireless, land or underwater fibre) can be the source of the latency problem. Check with your service provider to get a service level agreement to guarantee the quality parameters you want concerning packet loss, delay and jitter.
It's very important to not buy or install any new network system before identifying existing delays.
Once the latency issue is identified, a valid solution for optimizing your WAN is to purchase a WAN-accelerator from a vendor that offers such a product. The amount of traffic your WAN uses will also determine which vendor you will choose for this product. For example, if your business uses such programs as Citrix and SAP, then a traffic-prioritizing WAN accelerator such as Packeteer or Ipanema would be the best solution. But if your business uses more file-oriented applications, your best option would be to purchase a data reduction accelerator like Riverbed, Juniper Citrix and Cisco. Riverbed is by far the best solution.
While optimization systems effect only TCP flows, latency issues with UDP flows (VoIP) can't be resolved since real-time traffic is naturally subject to delays. Citrix WANScaler, Cisco WAAS, Juniper, Riverbed and Expand as well as other vendors sell systems for optimizing generic TCP flows which would meet the needs of your business.
As a word of caution, all the vendors hide the effects of latency. WAN optimization solutions act like an invisible TCP proxy, simultaneously spoofing both the server's IP address to the client and the client's IP address to the server. The TCP ACK messages are sent locally, speeding up the delivery of the ACK message and avoiding latency effects. TCP connections and throughput are subject to latency when the ACK messages are returned.
Vendors also hide the effects of latency by utilizing the entire WAN bandwidth link for implementing selective ACKs and large initial flow control windows. Compression mechanisms, layer 7 optimizations (mostly CIFS file-sharing as well as HTTP-based compression), embedded QoS are used for artificially increasing bandwidth not in use. This is known as "WAN optimization".
w
Showing posts with label wan optimization. Show all posts
Showing posts with label wan optimization. Show all posts
Monday, March 15, 2010
Monday, February 1, 2010
When Should You Add More Bandwidth To Your Business Network (eg WAN)?
Many businesses ask this question. It's common ..... so don't feel alone.
Is there a general rule of thumb, a percentage of bandwidth usage, that is an indicator that it is time to lease more bandwidth?
The true answer here is, "It depends".
Of course you need to know how many users you have ..... and what they're using the network for (e.g. CAD, Multi-media applications, sharing large data files, medical imagery, etc.). That's a factor in your ultimate decision obviously. As is what is the operation, are users on one side and servers on the other, are servers and users on both sides, how many points? Plus .... are either of those categories (number of users, applications) going to grow in the near future.
If growth of users and/or applications is in the distant future ..... you have more room to plan and maybe go with a gradual solution in the interim. In this case you may be able to go with a bonded T1 or burstable DS3 bandwidth rather than a full DS3 or OC3 bandwidth.
Before you can make a real decision on increasing your bandwidth though (thus increasing your expenses) the first step is to log/record your bandwidth usage patterns for a specific period of time - I'd say at least a month (30 day period). This will create a decent benchmark on where you stand in regards to your bandwidth needs.
Ignore occasional spikes -- spikes will happen, this is a reality of networking. However if you have sustained spikes lasting for several minutes at a time and this occurs at least once every day that is a hint of a larger problem.
(If that happened I'd note the time of day such spiking occurs, then correlate the times if they match up to any specific task the company is doing during that time period each day).
However, as a general rule ..... look at the bandwidth usage throughout the day. If the average usage is at 50% then start to talk about increasing the bandwidth. If it's at 70%, you will start seeing major trouble, don't just talk about increasing bandwidth then, DO IT.
If you want to be a bit more lenient rather than conservative on your benchmarks ..... get excited about increasing bandwidth when you consistently see usage at 75% bandwidth, when it hits 85% there is no debate, no talking ..... just order more bandwidth.
But, remember to base your decision on the average usage ..... not the spikes that may occur during the day. Again, spikes will happen no matter what your bandwidth is.
Here's another tip (or two) .......
Much of the time bandwidth is not the problem, but the amount of clutter and the lack of optimization at the machine level.
So ..... on every computer on the network do the following:
First, delete any temporary files (even temporary system files) in the c:\windows\temp directory and in their profile under local settings.
Second, get rid of any unused applications.
Third, ensure that the swap file is optimized.
Fourth, Stick more memory in the system if possible.
Last, Defrag the hard drive!
Following the above you should be able to reclaim some speed. If speed is your main issue (it isn't always).
Also .... if you get your files of a network server don't forget to optimize that as well. But the first rule we tell users is do not work off server copies if you can help it.
Now, if after going through all of the above you still decide you do need more bandwidth ..... you can get free assistance to find the right solution (at the best price too) at DS3-Bandwidth.com.
Michael is the owner of FreedomFire Communications....including DS3-Bandwidth.com and Business-VoIP-Solution.com. Michael also authors Broadband Nation where you're always welcome to drop in and catch up on the latest BroadBand news, tips, insights, and ramblings for the masses.
Is there a general rule of thumb, a percentage of bandwidth usage, that is an indicator that it is time to lease more bandwidth?
The true answer here is, "It depends".
Of course you need to know how many users you have ..... and what they're using the network for (e.g. CAD, Multi-media applications, sharing large data files, medical imagery, etc.). That's a factor in your ultimate decision obviously. As is what is the operation, are users on one side and servers on the other, are servers and users on both sides, how many points? Plus .... are either of those categories (number of users, applications) going to grow in the near future.
If growth of users and/or applications is in the distant future ..... you have more room to plan and maybe go with a gradual solution in the interim. In this case you may be able to go with a bonded T1 or burstable DS3 bandwidth rather than a full DS3 or OC3 bandwidth.
Before you can make a real decision on increasing your bandwidth though (thus increasing your expenses) the first step is to log/record your bandwidth usage patterns for a specific period of time - I'd say at least a month (30 day period). This will create a decent benchmark on where you stand in regards to your bandwidth needs.
Ignore occasional spikes -- spikes will happen, this is a reality of networking. However if you have sustained spikes lasting for several minutes at a time and this occurs at least once every day that is a hint of a larger problem.
(If that happened I'd note the time of day such spiking occurs, then correlate the times if they match up to any specific task the company is doing during that time period each day).
However, as a general rule ..... look at the bandwidth usage throughout the day. If the average usage is at 50% then start to talk about increasing the bandwidth. If it's at 70%, you will start seeing major trouble, don't just talk about increasing bandwidth then, DO IT.
If you want to be a bit more lenient rather than conservative on your benchmarks ..... get excited about increasing bandwidth when you consistently see usage at 75% bandwidth, when it hits 85% there is no debate, no talking ..... just order more bandwidth.
But, remember to base your decision on the average usage ..... not the spikes that may occur during the day. Again, spikes will happen no matter what your bandwidth is.
Here's another tip (or two) .......
Much of the time bandwidth is not the problem, but the amount of clutter and the lack of optimization at the machine level.
So ..... on every computer on the network do the following:
First, delete any temporary files (even temporary system files) in the c:\windows\temp directory and in their profile under local settings.
Second, get rid of any unused applications.
Third, ensure that the swap file is optimized.
Fourth, Stick more memory in the system if possible.
Last, Defrag the hard drive!
Following the above you should be able to reclaim some speed. If speed is your main issue (it isn't always).
Also .... if you get your files of a network server don't forget to optimize that as well. But the first rule we tell users is do not work off server copies if you can help it.
Now, if after going through all of the above you still decide you do need more bandwidth ..... you can get free assistance to find the right solution (at the best price too) at DS3-Bandwidth.com.
Michael is the owner of FreedomFire Communications....including DS3-Bandwidth.com and Business-VoIP-Solution.com. Michael also authors Broadband Nation where you're always welcome to drop in and catch up on the latest BroadBand news, tips, insights, and ramblings for the masses.
Friday, January 15, 2010
LAN & WAN - Does a Letter Really Change the Meaning of the Word?
Of course! Particularly when that one letter is a part of an acronym. As with the topic at hand, the difference between LAN and WAN lies deeper than the obvious spelling and phonetic. In many respects, there exist no single distinction between local area network (LAN) and wide area network (WAN). But there are, in reality, several differences that may well distinguish LAN from WAN.
The points to be considered are the following: geographic, connections, topology, usage, reliability and of course, the cost. We will first touch the comparison with regard to the geographic scope.
Scope
Given what LAN and WAN stand for, one can easily identify an immediate difference: the scope of WAN is clearly vast as compared to that of LAN. The former possesses networks that could spread across city to city, sometimes even on a broader scope. With that said, LAN may be deemed as an undersized version of a WAN and is usually limited to a single edifice or a residential house. So if you connect all the LANs then you may have just developed a WAN---a small scale WAN, that is.
Connections
On a local area network, Ethernet is the established standard for connecting computers or workstations. In here, wires and cables are but permanent. This arrangement allows interconnectivity among separate local area networks. As for the wide area networks, they almost always depend on common carriers or a system where they utilize or hire service providers to be able to interconnect.
With the issue on speed, LAN is approximately ten times faster than WAN. One cannot impartially compare the LAN's speed reaching 10 gigabits per second to a WAN's fastest speed of 155 megabits per second. (Talk about no competition.)
Network Topology
As shown in the previous paragraphs, WANs and LANs are configured differently. More often than not, the latter is inclined to be a peer-to-peer networking where one computer may share (and sometimes access) data and applications in other computer units within the network. While WAN operates in a client-to-server level and most, if not all, data originates from a central computer or server.
Usage
Continuing the discussion, the peer-to-peer sharing is commonly used to access files, as well as printers, scanners and other peripherals. The usage is somewhat controlled and restricted among a few groups of people. These users have their own passwords that will allow them to access a particular data (not the entire folder or drive). An administrator may also be present. In a WAN, it is not really the sharing of peripherals or whatnot, it stresses more on communications.
Reliability
LAN boasts a higher reliability rate because when one computer fails, all other will continue to operate. In WAN, this is not the scenario. When a unit falters in a WAN, most probably the rest will fail to function as well.
Cost
With the extent of WAN's reach, obviously this costs a lot. Not to mention the need for field experts like engineers to install and maintain all their machines. Sometimes, WAN also necessitates satellite links; yet another reason for its high set-up cost. Local area network, though housed in a single building, will prove to be more cost efficient.
So, in simple equation, WAN=vast reach=costly while LAN=limited extent=cost efficient.
Finally, to answer question posted above (yet again), there is a yawning gap between the WAN and the LAN.
Benedict Yossarian is specialises in internet marketing. Benedict recommends Comm store for networking hardware and Cat5 Cable. For IT training Benedict recommends New Horizons Learning Centres
The points to be considered are the following: geographic, connections, topology, usage, reliability and of course, the cost. We will first touch the comparison with regard to the geographic scope.
Scope
Given what LAN and WAN stand for, one can easily identify an immediate difference: the scope of WAN is clearly vast as compared to that of LAN. The former possesses networks that could spread across city to city, sometimes even on a broader scope. With that said, LAN may be deemed as an undersized version of a WAN and is usually limited to a single edifice or a residential house. So if you connect all the LANs then you may have just developed a WAN---a small scale WAN, that is.
Connections
On a local area network, Ethernet is the established standard for connecting computers or workstations. In here, wires and cables are but permanent. This arrangement allows interconnectivity among separate local area networks. As for the wide area networks, they almost always depend on common carriers or a system where they utilize or hire service providers to be able to interconnect.
With the issue on speed, LAN is approximately ten times faster than WAN. One cannot impartially compare the LAN's speed reaching 10 gigabits per second to a WAN's fastest speed of 155 megabits per second. (Talk about no competition.)
Network Topology
As shown in the previous paragraphs, WANs and LANs are configured differently. More often than not, the latter is inclined to be a peer-to-peer networking where one computer may share (and sometimes access) data and applications in other computer units within the network. While WAN operates in a client-to-server level and most, if not all, data originates from a central computer or server.
Usage
Continuing the discussion, the peer-to-peer sharing is commonly used to access files, as well as printers, scanners and other peripherals. The usage is somewhat controlled and restricted among a few groups of people. These users have their own passwords that will allow them to access a particular data (not the entire folder or drive). An administrator may also be present. In a WAN, it is not really the sharing of peripherals or whatnot, it stresses more on communications.
Reliability
LAN boasts a higher reliability rate because when one computer fails, all other will continue to operate. In WAN, this is not the scenario. When a unit falters in a WAN, most probably the rest will fail to function as well.
Cost
With the extent of WAN's reach, obviously this costs a lot. Not to mention the need for field experts like engineers to install and maintain all their machines. Sometimes, WAN also necessitates satellite links; yet another reason for its high set-up cost. Local area network, though housed in a single building, will prove to be more cost efficient.
So, in simple equation, WAN=vast reach=costly while LAN=limited extent=cost efficient.
Finally, to answer question posted above (yet again), there is a yawning gap between the WAN and the LAN.
Benedict Yossarian is specialises in internet marketing. Benedict recommends Comm store for networking hardware and Cat5 Cable. For IT training Benedict recommends New Horizons Learning Centres
Friday, January 1, 2010
How Do You Determine the Right Wide Area Network (WAN) Architecture For a Computer Network?
One of the most complicated decisions to make for an IT department concerns the architecture for a multi-site voice/data network. Should the network be based on simple T1 point-to-point bandwidth circuits? How about burstable DS3 bandwidth or maybe a single OC3 bandwidth cloud? No matter the bandwidth chosen .... the most efficient design should base the communication backbone on a MPLS configuration (Multi-Protocol Label Switching).
To get there, first and foremost in deciding about the WAN architecture of any organization communication network you must determine ....
- How many locations you would like to connect?
- Architecture - hub and spoke or mesh architecture?
- What applications would be run on this network .... Voice, Video, Data, all of the above?
- Precisely which applications will be run in case of Data?
- What will be the % each will take up of your QoS, the total of all three should be 100%?
- Voice is the premium level QoS and hence the most expensive as it is a real time communication, followed by video and then data.
- How many users precisely will be using the network at a given point of time at each location?
- What will be the concurrency factor? Are you are looking for 100% concurrency or you can manage with lesser concurrency?
- What is the scope of scalability at each location and hub location?
- Will the access to internet also be given to users?
- Internet at a central location can help you in implementing and enforcing various security policies of your organization.
- Do you want to give access to the network resources to a mobile user?
The answer to all these questions will help in arriving at the MPLS bandwidth required at each hub and spoke location.
Honestly speaking no organization should ideally try to do this calculation themselves. Instead they can hire a consultant or a telecom service provider to do this activity as they are experts in designing this solution. With their help you can easily decide upon the bandwidth for each location, select suitable router, make redundancy plans, routing the traffic on Atlantic or pacific routes, blah blah.
For help walking through the analysis .... and determining the best solution .... take advantage of the free help at DS3-Bandwidth.com.
Ideally I also recommend to give the freedom of providing and managing the routers at each location to the Telecom service provider. Then it becomes a managed solution and the service provider can easily monitor your network in the event of an outage. They then can remotely login into the routers and manage your complete network giving you higher uptimes and SLAs.
In the end .... take a deep breath and work through the steps above, also helps to increase SharePoint Performance. Following this simple game plan will get you to the right decision for your organizations network architecture.
To get there, first and foremost in deciding about the WAN architecture of any organization communication network you must determine ....
- How many locations you would like to connect?
- Architecture - hub and spoke or mesh architecture?
- What applications would be run on this network .... Voice, Video, Data, all of the above?
- Precisely which applications will be run in case of Data?
- What will be the % each will take up of your QoS, the total of all three should be 100%?
- Voice is the premium level QoS and hence the most expensive as it is a real time communication, followed by video and then data.
- How many users precisely will be using the network at a given point of time at each location?
- What will be the concurrency factor? Are you are looking for 100% concurrency or you can manage with lesser concurrency?
- What is the scope of scalability at each location and hub location?
- Will the access to internet also be given to users?
- Internet at a central location can help you in implementing and enforcing various security policies of your organization.
- Do you want to give access to the network resources to a mobile user?
The answer to all these questions will help in arriving at the MPLS bandwidth required at each hub and spoke location.
Honestly speaking no organization should ideally try to do this calculation themselves. Instead they can hire a consultant or a telecom service provider to do this activity as they are experts in designing this solution. With their help you can easily decide upon the bandwidth for each location, select suitable router, make redundancy plans, routing the traffic on Atlantic or pacific routes, blah blah.
For help walking through the analysis .... and determining the best solution .... take advantage of the free help at DS3-Bandwidth.com.
Ideally I also recommend to give the freedom of providing and managing the routers at each location to the Telecom service provider. Then it becomes a managed solution and the service provider can easily monitor your network in the event of an outage. They then can remotely login into the routers and manage your complete network giving you higher uptimes and SLAs.
In the end .... take a deep breath and work through the steps above, also helps to increase SharePoint Performance. Following this simple game plan will get you to the right decision for your organizations network architecture.
Tuesday, December 29, 2009
What You Should Consider For a WAN Optimization Solution
You're evaluating your WAN (Wide Area Network) and wondering how you can optimize performance across the network. What are you're realistic options for a solution?
Keep in mind that this isn't necessarily a bandwidth solution, such as say MPLS or ethernet is. Rather it's a bandwidth utilization question.
I would suggest looking at WAN optimization appliances: Riverbed, Juniper, Packeteer. In many cases they do a very good job with latency, especially if the main source of the problem is CIFS WAN-unaware issues (copying/opening/modifying files using Windows).
The first step in my view is to find out the path your IP packets are taking. A delay can be caused if your connection is going through satellite and not a land line (Earthbound wireless, fibre- underwater or land, etc.).
First I would suggest tracing your packets' route ..... then making an SLA with your service provider and see if they can guarantee the quality parameters you want concerning BW, delay, jitter etc.
I wouldn't jump into buying or installing anything before that!
Keep in mind that all WAN-accelerators solve Latency issues. So choose any Vendor that does WAN-acceleration. But choosing the right one is not that easy. It depends on your WAN-traffic.
If you use applications (like Citrix, SAP) than a traffic priorizing Accelerator (Like Packeteer or Ipanema) will offer you the best results. However, if your traffic is more file oriented (MS-Office) ..... than you will get the best results with the Data reduction Accelerators, like Riverbed, Juniper Citrix and Cisco. In general you'll likely get the best results with Riverbed.
Wan Optimizations appliances have an effect only on TCP flows. If you have latency issues with UDP flows (voip) I don't think you can do anything against that (real time traffic is very sensitive to latency)
About generic TCP flows optimization, you'll find many vendors in the market selling products which fit well according to your needs : Citrix WanScaler, Cisco WAAS, Juniper, Riverbed, Expand and probably a lot of other vendors...
Here is the trick they all use to hide the effects of latency (they don't hide latency, because it already exists, they hide the effects), Wan optimization solutions act like a transparent TCP proxy, spoofing server's IP address to the client, and spoofing client's IP address to the server, and they send the TCP Ack messages locally. So you don't have to wait for the round trip time to get the ACK message. It helps to "fill-the-pipe" quickly. This is the great solution to avoid latency effects, as TCP throughput is highly sensitive to latency, due to the Ack return messages, also congestion control has an horrible effect when the TCP connections start.
Also they implement selective Acks and large initial flow control windows to use the total bandwidth of the WAN link. Add to this some compression mechanisms to artificially increase the bandwidth available and some layer 7 optimizations (mainly CIFS for file sharing, but also HTTP-based compression), embedded QoS stuff, and you have an overview of what we call "Wan optimization".
Whatever the vendor you choose, they all implement what I described above.
Michael is the owner of FreedomFire Communications... including DS3-Bandwidth.com. Michael also authors Broadband Nation where you're always welcome to drop in and catch up on the latest BroadBand news, tips, insights, and ramblings for the masses.
Keep in mind that this isn't necessarily a bandwidth solution, such as say MPLS or ethernet is. Rather it's a bandwidth utilization question.
I would suggest looking at WAN optimization appliances: Riverbed, Juniper, Packeteer. In many cases they do a very good job with latency, especially if the main source of the problem is CIFS WAN-unaware issues (copying/opening/modifying files using Windows).
The first step in my view is to find out the path your IP packets are taking. A delay can be caused if your connection is going through satellite and not a land line (Earthbound wireless, fibre- underwater or land, etc.).
First I would suggest tracing your packets' route ..... then making an SLA with your service provider and see if they can guarantee the quality parameters you want concerning BW, delay, jitter etc.
I wouldn't jump into buying or installing anything before that!
Keep in mind that all WAN-accelerators solve Latency issues. So choose any Vendor that does WAN-acceleration. But choosing the right one is not that easy. It depends on your WAN-traffic.
If you use applications (like Citrix, SAP) than a traffic priorizing Accelerator (Like Packeteer or Ipanema) will offer you the best results. However, if your traffic is more file oriented (MS-Office) ..... than you will get the best results with the Data reduction Accelerators, like Riverbed, Juniper Citrix and Cisco. In general you'll likely get the best results with Riverbed.
Wan Optimizations appliances have an effect only on TCP flows. If you have latency issues with UDP flows (voip) I don't think you can do anything against that (real time traffic is very sensitive to latency)
About generic TCP flows optimization, you'll find many vendors in the market selling products which fit well according to your needs : Citrix WanScaler, Cisco WAAS, Juniper, Riverbed, Expand and probably a lot of other vendors...
Here is the trick they all use to hide the effects of latency (they don't hide latency, because it already exists, they hide the effects), Wan optimization solutions act like a transparent TCP proxy, spoofing server's IP address to the client, and spoofing client's IP address to the server, and they send the TCP Ack messages locally. So you don't have to wait for the round trip time to get the ACK message. It helps to "fill-the-pipe" quickly. This is the great solution to avoid latency effects, as TCP throughput is highly sensitive to latency, due to the Ack return messages, also congestion control has an horrible effect when the TCP connections start.
Also they implement selective Acks and large initial flow control windows to use the total bandwidth of the WAN link. Add to this some compression mechanisms to artificially increase the bandwidth available and some layer 7 optimizations (mainly CIFS for file sharing, but also HTTP-based compression), embedded QoS stuff, and you have an overview of what we call "Wan optimization".
Whatever the vendor you choose, they all implement what I described above.
Michael is the owner of FreedomFire Communications... including DS3-Bandwidth.com. Michael also authors Broadband Nation where you're always welcome to drop in and catch up on the latest BroadBand news, tips, insights, and ramblings for the masses.
Saturday, December 26, 2009
Are You Confident Your Application Will Work in the WAN?
LANs vs. WANs
LANs are big, fast and reliable networks normally found within an organisation's main site or building (or the test lab). They can be rightly viewed as being super fast highways. The amount of data they carry is light relative to their capacity and the distances this data travels is relatively short. In contrast, WANs are geographically disbursed networks of often lower capacity which are required to carry a high volume of data compared with their capacity, over much greater distances.
There are three network key characteristics that influence an application's performance:
Available Bandwidth
WANs tend to have a much lower bandwidth than LANs which means that individual applications have to compete for space. This lower bandwidth can have a detrimental affect on software performance. Additionally, network administrators can set up their networks to favour certain applications like Voice over IP (VoIP) over 'conventional' applications, so once in the production environment the application is again competing for supremacy over others.
Latency
Latency is the delay encountered when running an application between two networks. It occurs because standard TCP/IP networks do not do not send data in a continuous stream, instead breaking it down into packets (like envelopes in the post) and sending it in batches. They also wait for confirmation that the packets have arrived safely before sending more, causing further delay. Also, the journey itself is not direct and various network devices will be encountered along the way, which all add their own additional delay. As a result, it can take 90ms to complete a round trip journey and an application transaction will consist of many such trips.
Packet Loss, Error and Reordering
As the packets of data travel over the network they can be lost, errored or reordered so that they arrive out of sequence, or don't arrive at all. It's like sending a bus down the road that either doesn't arrive or doesn't arrive entirely intact and is therefore probably useless when it reaches its destination. Wireless WAN, Satellite and 3G/Mobile Phone networks are generally subject to higher loss and error rates than wired networks. The increased use of wireless networks within buildings and as a way of allowing mobile 'on-the-road' members of the workforce to receive data means that applications need to be developed to cope with this potential for loss, erroring or reordering.
WAN emulation / network simulation
Reproducing these three network conditions is impossible if testing is confined to the internal LAN. However, WAN emulation / network simulation technology can be deployed in the same room as a normal test rig or even on a desktop. It allows the user to recreate a wide variety of different WAN or Wireless conditions and enables testing during prototype, development, quality assurance and pre-deployment stages.
A WAN emulator also gives complete control over the conditions in a single test in and also has the ability to reproduce these conditions time and again. This cannot be guaranteed if using a live network and additionally testing on a live network can interfere with existing mission-critical business applications already running.
Written by Phil Bull from iTrinegy, developers of sophisticated, yet easy-to-use, application response time monitoring, network traffic analysis and WAN/Network emulation technology to help organizations address networked application performance issues, de-risk application roll-outs and optimize network utilization.
LANs are big, fast and reliable networks normally found within an organisation's main site or building (or the test lab). They can be rightly viewed as being super fast highways. The amount of data they carry is light relative to their capacity and the distances this data travels is relatively short. In contrast, WANs are geographically disbursed networks of often lower capacity which are required to carry a high volume of data compared with their capacity, over much greater distances.
There are three network key characteristics that influence an application's performance:
Available Bandwidth
WANs tend to have a much lower bandwidth than LANs which means that individual applications have to compete for space. This lower bandwidth can have a detrimental affect on software performance. Additionally, network administrators can set up their networks to favour certain applications like Voice over IP (VoIP) over 'conventional' applications, so once in the production environment the application is again competing for supremacy over others.
Latency
Latency is the delay encountered when running an application between two networks. It occurs because standard TCP/IP networks do not do not send data in a continuous stream, instead breaking it down into packets (like envelopes in the post) and sending it in batches. They also wait for confirmation that the packets have arrived safely before sending more, causing further delay. Also, the journey itself is not direct and various network devices will be encountered along the way, which all add their own additional delay. As a result, it can take 90ms to complete a round trip journey and an application transaction will consist of many such trips.
Packet Loss, Error and Reordering
As the packets of data travel over the network they can be lost, errored or reordered so that they arrive out of sequence, or don't arrive at all. It's like sending a bus down the road that either doesn't arrive or doesn't arrive entirely intact and is therefore probably useless when it reaches its destination. Wireless WAN, Satellite and 3G/Mobile Phone networks are generally subject to higher loss and error rates than wired networks. The increased use of wireless networks within buildings and as a way of allowing mobile 'on-the-road' members of the workforce to receive data means that applications need to be developed to cope with this potential for loss, erroring or reordering.
WAN emulation / network simulation
Reproducing these three network conditions is impossible if testing is confined to the internal LAN. However, WAN emulation / network simulation technology can be deployed in the same room as a normal test rig or even on a desktop. It allows the user to recreate a wide variety of different WAN or Wireless conditions and enables testing during prototype, development, quality assurance and pre-deployment stages.
A WAN emulator also gives complete control over the conditions in a single test in and also has the ability to reproduce these conditions time and again. This cannot be guaranteed if using a live network and additionally testing on a live network can interfere with existing mission-critical business applications already running.
Written by Phil Bull from iTrinegy, developers of sophisticated, yet easy-to-use, application response time monitoring, network traffic analysis and WAN/Network emulation technology to help organizations address networked application performance issues, de-risk application roll-outs and optimize network utilization.
Subscribe to:
Posts (Atom)