Is there an easy way to evaluate IXP benefits? There is. NPV is a simple, but powerful financial calculation that is useful as a snapshot of a business case.
3. • Net Present Value
• Compares an investment made today
• Against the present value of future cash flow(s)
• Discounted by a rate of return
• Comparing doing nothing (transit), to public peering
4. What is “CAPEX” and “OPEX”?
• Capital Expenditure
• A business expense for a fixed asset that outlives the tax
year
• Router and switch ports, optics, tools, testing hardware
• Operational Expense
• Normal business expenses
• Wages, data center rent, conditioned power, IXP port fees
5. • Get Costs Together
• Capex
• Opex
• Derive the Benefit Targets “TED”
• Transit Mb/s
• Effective Mb/s
• Discount Mb/s
DISCOUNT
+/-
Effective
Cost
Transit
Cost
6. • Base Modeling
• Generally expected results
• Worst Case Modeling
• If you’re wrong
• Best Case
• If you’re wrong again, but in a good way
• Standard of Care
• Risk is always appropriate, but call it out
7. Business Case Worst Base Best
Utilization Mb/s 5,000 5,000 5,000
Investment $(16,350) $(16,350) $(16,350)
Discount Rate 14% 14% 14%
Cash Flow YR1 $3,825 $17,325 $30,825
YR 2 $4,208 $19,058 $33,908
YR 3 $4,628 $20,963 $37,298
YR 4 $5,091 $23,060 $41,028
YR 5 $5,600 $25,366 $45,131
NPV $6,157 $85,592 $165,028
8. Business Case Worst Base
Utilization Mb/s 5,000 5,000
Investment $(16,350) $(16,350)
Discount Rate 14% 14%
Cash Flow YR1 $225 $22,725
YR 2 $248 $24,998
YR 3 $272 $27,497
YR 4 $299 $30,247
YR 5 $329 $33,272
NPV $(15,026) $117,367
9. • Do port fees matter?
• What about intangibles?
• IXP benefits vs. transit?
• Golden Packets?
• Market Pressure?
• What if I am upside down?
10. • Credits
• Ben Hedges, for reviewing
• Bill Norton, for ’golden packet’ conversations
• Job Snijders, for his inspirational hard work.
• Contact info (and for a copy of the model)
• Martin.Hannigan@microsoft.com
• LinkedIn http://bit.ly/LMJH-GPF12
• Twitter @theicelandguy
• Cell +1-617-821-6079
Editor's Notes
A few quick disclosures
I work at Microsoft
I’m on the Interconnection Team, full time.
I’m a part time student and them some, studying at Boston University and MIT for credentials in data center Real Estate Finance and Development
I’m anticipating completing my MIT studies in June and BU in December. Very exciting and busy times.
Today, I’m presenting a method to take a snapshot of business viability related to a port on a public internet exchange. You can do this as little as as much in depth as you like, my approach is effectively a reliable way to snapshot the benefits or lack thereof.
What this presentation is not is a judgement if peering is good or not. It’s just a method to use to analyze your own unique situation.
Net present value is a calculation that compares an amount invested today to the present value of future cash receipts from that investment.
In laymans terms, the amount invested is compared to future cash amounts after they are discounted by a specified rate of return, or YIELD. The discount itself is referred to as the DISCOUNT RATE.
Today, we will snapshop transit vs. peering
Capital expenditure is the cost of developing or providing non-consumable parts for the product or system. Like routers, switches, WDM gear or a building to operate a data center in.
An operating expense or expenditure, is an ongoing cost for running a product, business, or system. These include wages, rack rents, cost of conditioned power, transit and port fees.
In order to do the analysis, we need to generate the data for the business case
First, we roll up all capital expenditures like allocated ports, related chassis costs, optics and other hardware. The capex allows me to create the basis for the investment, or the capital spend to do the work or project
Then I tally up the expenses to subtract them from the savings (or add to the losses) we project. The snapshot works both ways, highlighting a win or a loser.
So simple, think net operating income.
Take your transit mb/s cost
Subtract your effective mb/s
Derive the positive or negative “discount mb/s”.
Generate cash flows with expected utilization
Subtract opex
Calculate for each year
Results
Gotcha: Make sure you are comparing apples for apples, insure all costs are yearly, including depreciation
After all my data is compiled and checked for accuracy, I copy it to three categories and tweak to map to the categories I’m looking at.
A base model, where I can show what I believe are the results we should see
A worst case model, where I might take an extremely conservative approach and exagerate costs and minimize benefits.
The best case, where I hope to be, but I take more risk, perhaps I increase knobs attributing more to benefits or I expect that I’m a super mighty negotiation and I’ll have less investment costs
When dealing in risk, I find that it is always best to call out how much risk you are anticipating in the cases. Transparency for the win. No surprises.
Here are what some of the results look like. In this snapshot, I look at an IXP port and associated costs.
Modeling a single IXP port purchase, in a single location.
Assumptions of transit at $2.00, Worst case IXP port at 1500, Base 750 and Best 0.
Everything else is equal
I made the yield a little higher than I expect that most of you would use.
You can find out what yield your company uses by asking your Treasury department.
Comparing two IXPs and assumptions are
Different locations
IXP port fees at $1500 vs. $750
Rack costs at $1400 vs. $1200
What’s different? The effective cost per mb/s, which increased then decreased the discount mb/s benefit.
Benefits before yield
Total benefits: $138,738
After Yield payoff $117,367
Yield Cost $21,372
Do Port Fees Matter? Of course they do. But they’re at least equal in value to any other. If a rack rent is $1200 and an IXP port is $1200, then which one do you target for reduction? Right Answer? BOTH. By modeling the number you provide yourself and others a visualization of where your expenses are and which ones you can tweak, how much and what the overall impact is.
Intangibles: Intangible benefits of an IXP port are measurable. Without an IXP in a market, the benefit of PNI’s may not exist or be minimized. Technically, the benefits of a PNI _should_ be added to the cash flows from savings. I would argue that in most cases the typical $700 delta should be added directly to them thereby mitigating the IXP port cost entirely. There would also be less overall market pressure on transit costs. For simplicity sake, and since we are talking about snapshots, I’d simply blend the costs into a single number while raising the available capacity for the benefit inflows. You still need demand capacity to drive the discounts though, but the benefts are much higher from that view.
Transit: Is transit better? It can be. But the value is pressured significantly by the increasing reliability and capabilities that IXP’s are stepping up to provide and when including the benefits of PNI in the equation.. I can see the day in the near future where IXPs will be SSAE-16 and HIPAA compliant by default while providing cloud and other services on a localized basis. I’m near certain that the internet won’t scale without them in todays terms.
Golden Packets: This is a concept that Bill Norton and I have been debating. There is a point where a financial loss on an IXP becomes only a technical matter. If those losses are blended into product P/L which are dependent on them, it may just be an opportunity cost which ends up resulting in overall increased benefits. It’s not always crystal clear.
Market Pressure: Then there are the ports which are obviously not going to be cost effective on day one. But they too hold hidden value. We should all support new IXP’s where we can and can justify investing in the eco-system. Adding IXPs in markets that don’t have them tend to drive costs down significantly when done right. A small investment today can result in a bigger in vestment in the future. You could use NPV to project traffic flows and include at least initial losses. This is much more speculative. I can tell you from experience for at least “the big guys” it is well worth it. From many perspectives.
Last but not least, what about realizing that the benefits aren't worth it. Depending on how much traffic you have, you can either make a decision that the intangibles contribute to the big picture, say performance, or you can trade and go to transit without peering.