WAN Bandwidth Calculation

Good morning all,

I have a nice easy one for someone out there i'm sure......

I have been asked to bridge and existing LAN to another site office at the other end of the country:


Currently the existing LAN has no external connectivity and everything is on the same subnet, servers and users.

I have no NETFLOW or network management or monitoring software as this LAN started off very small and grew over time and was thrown together by those using it and not anyone working in the network field.

All i have is the stats from the two main ESX servers hosting the VMs at the existing site showing throughput in KBps (attached).

I need to be able to calculate what the WAN bandwidth requirements are based on this information, I have had a go (shown below) but I am not sure if I am working this out correctly and I have no one I can ask in my company to verify this for me.

My working out

Server ending .100 shows avg throughput of 10,000KBps (10MBps)

10MB (10,000KB) * 8 = 80,000Kbps (8Mbps)

So on average this server uses 8Mbps throughout the day peaking at 40MB or 32Mbps in the early hours (which I suspect to be a backup)

Server 2 ending .101 is the same working out which = avg 16Mbps

8Mbps + 16Mbps + overhead of roughly 30% (8Mbps) = 32Mbps bandwidth required on the WAN link for this service

This shows us the total bandwidth required for these servers, but most of this is going to be local to the main site i.e. from Main Site Existing LAN to the Server resources hosted locally at that site.
We are looking at a small number of users in the Remote site transiting the WAN to access these server resources (exact number to be confirmed but I expect this to be around 10 as a maximum at least initially)

Therefore we could take the not overly scientific approach as divide 32Mbps of bandwidth currently being utilized by the servers and devide this by the current number of users using these central servers (understood to be around 300) and then multiply by 10 to get the total bandwidth required over the WAN.

For example,  32Mbps /300 = 0.1 Mbps per user * 10 (users) = 1Mbps bandwidth required

Given an average medium usage office worker’s IT bandwidth consumption is roughly calculated to be 80Kbps by industry guidelines I guess this is a plausible figure.

Does this sound about right, I really need some help and your guidance is greatly appreciated?

Thanks in advance,

Who is Participating?
giltjrConnect With a Mentor Commented:
Not, sure.  I will say that our remote site with developers has had as many as 200 developers RDP'ing to VM's over a single VPN tunnel over the Internet.  

Both sides of the tunnel had 100 Mbps Internet links and we have never had any performance problems.
Hi David,

Before we can advise we would probably need to know what the reason is for linking the two offices so we can assess what traffic would flow between them.

As you correctly note, a lot of the traffic included in your figures is internal and we need to establish how much of that would be transferred across the link.

Things to consider:

Are there any applications that will be communicating across the link?
Are the networks on windows domains? if yes, will they be linked to remain separate?
Would files be accessed or transferred across the link very often? If yes, what kind of files (small word docs, large video files etc)
canttalkeatingAuthor Commented:
Hi Totallytonto,

The reason is to employ some new staff at the Remote office to do the same exact job as staff at the Main office.  The LAN in the existing office consists on Windows hosts and Linux hosts with all of the windows based applications being hosted on a pair of ESX Servers running VMware.  The Linux hosts talk to physical Linux servers at the moment, but will be virtualized further down the line.

Initially this was believed to be around 10 staff members but it has since been revealed that we are looking at up to 120 users at the Remote office so the bandwidth required is going to increase substantially.

The files that would need to be opened are large in size (around 1GB) as these are code repositories that need to be accessed by our developers.  We are looking at getting the software vendor in for consultation as the software is not designed to be used over a WAN but the vendor does have a software add-on for muti-site requirements such as this which we will consider.

Given the number of staff at the Remote office, I think the ideal would be to perhaps have the same resources/services at the New Remote Site and then just have replication traffic passing over the WAN along with an out-of-hours backup back to the Main site, but this may not be an option financially and I only look after the networks!

The alternative to the above would be to keep everything at the main site and have all users at the New Remote site connect to all resources over the WAN, which is were the bandwidth calculation will be the deciding factor as I may be able to leverage an existing network link or install a new one

I need to consider both options and determine which is best :)

Thanks for getting involved

Free Tool: Subnet Calculator

The subnet calculator helps you design networks by taking an IP address and network mask and returning information such as network, broadcast address, and host range.

One of a set of tools we're offering as a way of saying thank you for being a part of the community.

--> For example,  32Mbps /300 = 0.1 Mbps per user * 10 (users) = 1Mbps bandwidth required

The problem with this example is when the network is hitting 32Mbps you don't know if that is 1 user or 300 users driving the 32Mbps.  

I would suggest you look into using VDI/RDP from the new site to the current site.   That way the majority of the traffic flowing between the sites is screens and not huge documents/programs.

If you try to replicate files after hours that means that one site could be looking at old files for a full day.

You may also want to look at WAN accelerators/optimizers.
canttalkeatingAuthor Commented:
Thanks for the response Giltjr,

Would it matter if the peak usage was one user driving it up to 32Gbps or 300 users doing this, the bandwidth used would still be 32Gbps?

I have considered RDP or perhaps a Thin Client solution using Citrix to control the type of traffic and the volume.

With regard to replication, I meant to replicate over the WAN throughout the day and run back ups out of hours.

I am familiar with Riverbed Steelhead and BlueCoat PacketShapers, both of which I agree would be good assets but I need to understand the volume of traffic before considering the components of the design

Thanks for the suggestions and for getting involved

Happy to discuss further

--> Would it matter if the peak usage was one user driving it up to 32Gbps or 300 users doing this, the bandwidth used would still be 32Gbps?

Think about it, it would.  Based on your formula a single user would only need a 100Kbps link.   Do you think a single user would get the performance they need over a 100Kbps link?

Image opening that 1GB file over a 100Kbps link.
canttalkeatingAuthor Commented:

I see where you're coming from, do you have a suggestion on how this could be measured more accurately?

Ultimately I need to know what size pipe I need between the two sites in order to allow the users at the Remote office (120 users) to do their work unhindered.


What you can start off with is answering the question "How long is a single user willing to wait to do function "X"?  Then figure out what is the minimum bandwidth required for that function.  

Example, if a single user only wants to wait 10 seconds to open a 1 GB file, then you would need a 1 Gbps link.  If they are willing to wait roughly 100 seconds (just over 1.5 minutes),  then a 100 Mbps link would do.  However a 100Kbps link it would take close to 20 minutes.

Depending on what is going on, there will be some "back ground" noise always flowing over the link.  However you can make the assumption that 80% of the time only 20% of the users will actually be doing function that cause data to flow over the link.  

The tough part is convincing users that expecting to open a 1 GB file over a WAN link should take 2+ minutes.

Using WAN accelerators will reduce the bandwidth required on the physical link.  

Something else to think about to reduce the cost of the physical links would be to use the Internet with a combination of VPN and WAN accelerators.
canttalkeatingAuthor Commented:
Thanks Giltjr,

I get what your saying.  The users currently work on a closed 1Gbps LAN so they are used to sub 10 second speeds for opening 1GB files and I would imaging that they would want as close to this as possible for the new staff working out of the new Remote office.

hypothetically, if we had a 1Gbps dedicated WAN link, a user would be able to open a 1GB file in <10 seconds, but if a second user were to require the same or another file of 1GB in size to be opened at the same time, this would open in 20 seconds right?

Depending on how far apart the two locations are it would take about 10 seconds.  The Further apart the locations (that is the greater the latency) the longer it takes to open the file.

Even in your current office, it is possible that users opening 1GB file would double their time.

If two users attempted to open two 1 GB files from the same server unless the server had more than 1 Gbps of bandwidth the time it takes to open both files would increase.
Realistically, with your assessment of user numbers and filesizes, a WAN link isnt really suitable at all.
Your users would get considerably less than the 1gbps internal network they are used to.
No wanlink can get anywhere near that for multiple users, especially the number youre looking at.

In reality, you can either look to reduce the need for huge data transfers by
A) moving to web based apps or getting users to open these 1GB files via a terminal server
B) replicating the data to each site in realtime with DFS so access is LAN instead of WAN wont work with database files though)
canttalkeatingAuthor Commented:
Thanks totallytonto,

A) Moving to web based apps - By this do you mean accessing the same file repositories via a web front end from the remote location, rather than opening the file itself? Could you advise further on this solution?

With regard to Terminal Server do you mean something like Citrix?  If so, the remote workers would connect to the Terminal Citrix Server via ThinClient and from there access the files they require locally i.e. the >1GB files would remain within the LAN at the Main Site and therefore not traverse the WAN?

Cheers for the help on this
Remote web workplace, sharepoint and various other systems can provide a web front end to users over a wan link. Its the same as many of the 'cloud, services available but on your own servers instead of theirs.
Terminal servers can work over web too if required.

And yea, just like citrix. It keeps the traffic on the LAN and only has screen updates over the wan.
canttalkeatingAuthor Commented:
Thanks totallytonto,

So if you store the files on a SharePoint server and then access the SharePoint server over the WAN and open the large files is this not the same as accessing the file when it is stored in a normal windows fileshare?  Is the penny about to drop for me with regard to the benefits of SharePoint? Is SharePoint going to be using HTTP/HTTPS where accessing via a windows fileshare would use CIFS/SMB which has a greater overhead than the Web Traffic?

SharePoint aside, I could in theory stand up a Citrix Server and have Users connect in remotely via Hardware Desktops to a Citrix Web Front End to access their desktops and all traffic over the WAN would be screenshots reducing the overhead. (or we could do the same thin using regular Citrix).

I am determined to learn this so I can build myself a little Knowledgebase so I can answer these sorts of design requirement questions in future as I think this is the route I am going to end up going down from here.  This being my first :)

Thanks again for all the help
SteveConnect With a Mentor Commented:
Accessing files by standard sharepoint still downloads the file so thats about the same as any other traffic, but it can provide a web front end (like office 365) where a 'web' version of word/excel handles the file which never actually leaves your LAN.

From a Citrix point of view, yes youre spot on. You can use various types if citrix setup as long as the aim is to prevent the large files ever having to leave your lan.
This method is suitable for most circumstances, with the exception of graphically intensive applications (as they tend to cause so much screen update traffic the advantages are cancelled out.)
canttalkeatingAuthor Commented:
Thanks again totallytonto,

I still don't fully understand the SharePoint 'web' version of Excel and how it keeps the file local to the server.  Are you saying that normal SharePoint works like a standard file share but you can implement 'web based' excel so that when the remote user accesses the file, it keeps the file local to the SharePoint server rather than opening it within the locally installed copy of excel?

I am leaning towards a Citrix solution as the file types I think these guys will be accessing are IBM specific which I assume don't have an equivalent web version for SharePoint (but I will look into this).
Citrix is a vendor specific version of the VDI/RDP option I mentioned earlier.

That I am aware of the "Sharepoint" solution with the "web" version of Office products just prevents you from installing Office on the computers at the remote site.  The files still have to get sent over the network.  

Using WAN accelerators/optimization I think be better than the Sharepoint solution.   There are many out there, but I have not used any of them.  Bluecoat and BigIP's WOM are two that come to mind.  However most WAN accelerators/optimizers are dsigned for low speed high latency links, that is T1/E1 links maybe 100 Mbps links.  Anything faster than 100 Mbps will not gain too much if anything unless you have very high latency.

How far apart are the two sites?
canttalkeatingAuthor Commented:
Thanks giltjr,

Apologies I didn't realize they were the same thing.  When I heard RDP I was thinking traditional Remote Desktop via Windows :)

I have done some further investigation into the software applications the staff are going to be using and it involves actually downloading 2GB file repositories to their local workstation which they use to compile their code. The code as I understand it is then saved back to the central server.  They also use some other software along side this which I am looking into.  

With this in-mind WAN optimization is looking like a possible candidate as it's going to be the same files being downloaded over and over by the same users from the same target servers, but I won't rule out Citrix either to save downloading altogether!  Once I know more about how the other tools and apps are used I can use the information I have learned here from you and totallytonto to decide on the best solution overall.

I would still have liked to have the maths behind calculating the current bandwidth to derive a WAN pipe size for the Remote site, but I guess I might have to decide on the best solution/most optimal way of working first and then workout the bandwidth requirements needed to support that solution.

More than happy to discuss further

giltjrConnect With a Mentor Commented:
O.K., sounds like you are doing some of what we have done.

Our current setup is a
1) Remote DC where all servers are located.
2) HQ with most of our developers
3) Remote site with developers.

Our HQ and DC are less that 10 miles apart.
The remote developer site is half-way around the world.

Our HQ and DC are connected currently connected with four links totaling 600 Mbps.  That will be replaced soon with a single 1 Gbps link.  Developers at our HQ user their desktops for all development with the code repository being at the DC.  So they actually download/upload the code between the DC and HQ when they do their work.

The remote site connects to our DC via VPN over the Internet.  We have a VMWare ESXi cluster.  Each developer has their own VM and they RDP to their VM.  So when they do download/upload it stays within the DC.

Our future plan is to implement VDI for all developers.   The VDI solution for the developers at our HQ is more for disaster recovery not performance issues.  This way if something happens to our HQ building, the developers still can do their work.
SteveConnect With a Mentor Commented:
If youre using specific applications youre right there wont be web versions available so it may not be an option, but have a look at googledocs and the microsoft verion for an example of web based apps. Its like the difference beween outlook and webmail. One runs on your pc and the other is web-based.

Citrix and rdp are the same idea by different companies. I think they can both provide a 'remote app' facility for most applications. Worth trying as i think youll find transferring such large files a pain

Alternatively, consider something like a fie server at each site with replication overnight.
What product from IBM are you using, it sounds as if you are doing application development.
canttalkeatingAuthor Commented:
Good morning Gents and thanks for the comments and suggestions.

giltjr, the IBM products are IBM ClearCase and ClearQuest predominantly with a couple of others to provide additional functionality.  This is for app dev.

When your Developers access the Code repositories are they downloading the snapshots to their local machines or is this opened on the server and worked on over the WAN link?
They download snapshots to their desktops and work from there.  I know we are using ClearCase, not sure about ClearQuest.

Like I said, right now we have 4 WAN links with a combined total of  600 Mbps of bandwidth between our HQ and DC.

At our HQ we have about 400 employees, 200 application developers and the rest are either operations support or end users.
canttalkeatingAuthor Commented:
Thanks giltjr,

My second site will have 120 developers.  If I used RDP to connect back into their individual VMs on the ESXi Centralized servers would this not consume too much bandwidth and server resources when everyone is connected at the same time?

Is their a common model explaining how much RDP traffic takes up per second that I could use to work out a worse case scenario i.e. cumulatively derive the total amount of bandwidth consumed if all developers were connected back to the main site via RDP at the same time + some additional % added for other traffic traversing the WAN during the day such as TCP overhead at 30% and data synchronization depending on what I can install locally at the new office?
canttalkeatingAuthor Commented:
Good to know giltjr, was this just regular RDP from a standard Windows Desktop fat client to VM Desktop images hosted within ESX?
Yes, standard Windows RDP to VM Windows Desktop within ESX.
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.