Network Guy's - WAN Capacity Planning question

[6172crew]

How many T1s are you using now for this one? Could you use a VDSL line and same some $$ or do you need a DS3?

I am assuming you are talking about the pipe between 2 points.

I know the telco companies like to sell in blocks of T1s which are ADSL, bt if you could get a VDSL line at 16mb for $100 a month...I dont think they sell a block of IPs with those but attaching a DLCI or multiple IPs to VDSL might take care of your issue on the cheap.

On a side note, is there one group that is using a service more than another? Could you ad another T1 for the bandwith hogs and create a new VLAN so that they use that T1 for X,Y but c and z is for trucks?

If they offer u-verse in that area then VDSL is available, only issue I can see is that they may not want to sell the addresses needed to get it done.

I do know ATT stopped bonding dsl lines by attaching email addresses to them so they wont bond correctly.

Then again I might be talking about something completely different here.

[nick]

This is a question for you hard-core network guys that have done a lot of enterprise WAN work.

One of the locations on my WAN is starting to really hit their bandwidth limits pretty hard. Our 95th Percentile for transmit usage has grown to 89% of existing capacity and I'm starting to receive complaints about performance from our employees.

1st Question: How much headroom should I keep between the 95th Percentile and the max capacity of any given WAN circuit?

2nd Question: Where did you get that metric, and is it documented anywhere?

With the current economic environment I'll need to fully justify any requests to upgrade this circuit.

Thanks,
John

1. I start looking at an upgrade when I see 50% average usage. Assuming a clean line, that's when you start experiencing delays, and start really paying attention to QoS.

95 percentile isn't as good at indicating sustained bandwidth usage, but you hitting 89% of it does show that at the very least your traffic peaks a lot.

Do you have any other ways to measure your traffic than what you get from your ISP (I presume that's where you get your 95 percentile from)? If you control the end devices, it's not hard to do.

You have to know your actual traffic pattern before deciding on upgrading the link, as you need to identify the actual problems first. For example, if some users experience delays and it's a problem, you might just need to play with QoS (I presume you already some QoS, given that you're considering upgrading your link).

2. I believe, I saw Cisco and Juniper docs with updated best practices on bandwidth utilization, which includes their best practices on when to upgrade the link. They'll push WAN optimization, of course, as there's more money in it for them, but they have the info on when to upgrade the link, too. Too lazy to search right now, I've had my fill of IT for the day. That's why I'm on a gun forum right now, you know

However, best practices would work if you need to justify this to the IT side. If you need to justify the need for upgrade to the business side, you need to do it in terms of ROI and compliance ROI/risk assessment.

To get there, do you have the traffic history data for that link? If yes, take the data, show the pattern of growth, and use it to create the projected increase in utilization. Then create the cost comparison between using the link as is an upgrading it "gradually", going with a considerably larger link, potentially with a different technology (the per-Mbps cost there would be lower, of course), and the cost of doing traffic shaping. Include the potential benefits and pitfalls.

Then supplement it with the ROI analysis on the cost of delays (effective user downtime (when the users aren't technically down, but they're waiting on something), projects that can't be done because of the slow link, support costs, including the support calls to respond to user complaints about slowness and support calls resulting from something not synchronizing on time, etc.) vs. the cost of the new link/whatever solution you come up with (as upgrading the link isn't the only solution). Of course, you present this paragraph first, before presenting the ROI analysis for which technology to pick

Then, if there're compliance/security/disaster recovery issues, do the risk assessment. For example, what if the data doesn't get synchronized. What if your IDS doesn't get adequate bandwidth to monitor the network behind that WAN link? What if your directory doesn't get synched? And so on.

This is just a generic description, of course, as I don't have the slightest idea of your environment. If you have more questions, PM me.

[nick]

Ah, I see there's more info in the thread already. One day I'll read the entire thread before posting, but that day isn't today. Not yet.

[artherd]

so much is dependent on your business and usage patterns.

Echo the bonding and QoS for voip, times a thousand!!!

[jmlivingston]

Just to follow-up on this...

I received permission to upgrade their internet to a 10 Mb MetroE circuit from Cogent. Once this is installed, I can move their Snapmirror replication traffic off their WAN and push it through a VPN back to HQ. This should relieve their WAN congestion plus enhance their internet service (currently their internet is 3Mb).

John