Professional Documents
Culture Documents
TTG does not participate in TTG starts its own carrier- TTG operates a dedicated TTG partners with global
an open data hub in an specific data hub targeting carrier neutral datacenter1 players for datacenter and for
attempt to limit its ICPs, carriers and regional serving international players IXP activities to create a
Strategic Intent & Tactics
attractiveness, keep global operators in a controlled as well as local MNOs / ISPs carrier-neutral data hub
carriers at bay and hence environment that protects TT open to everyone
Enterprise clients of TTG not
protect TT’s domestic from corporate customer
allowed in TTG’s data center TTG aims to attract a
revenues cannibalization
maximum of international
TTG generates upside from
TTG peers with content TTG generates upside from players to create IP Gravity in
connectivity and colocation
providers abroad and ensures int’l backhaul to and TT’s data hub, in order to
while limiting enterprise
high quality connectivity with collocation in own hub, while substitute cannibalized
customer cannibalization
them limiting the attractiveness of domestic Corporate and
an open data hub Carrier-neutrality enhances Wholesale revenues with new
TTG serves int’l demand only
value proposition to and wholesale revenue streams,
reactively – no proactive TTG fully serves int’l and
uptake of tenants, but international and domestic
efforts to bring traffic to Turkey domestic connectivity demand
introduces wholesale
undertaken from a potential “parallel” open
competition
data hub
Examples
today
1) Data center can be owned by a separate TTG-owned entity not subject to regulation
Note: Etisalat’s and Omantel’s business model are more “closed” and less conducive to carriers than the business model for “Proactive” considered here
Sources: Strategy& analysis
Strategy& Confidential June 2015 Prepared for TTG 2
In “reactive”, TTG does not participate in the “data hub”
ecosystem in an attempt to limit its attractiveness
TTG play options illustration – “Reactive”
TTG Competing Data Hub
TTG Carrier–Specific Data Center Competing Carrier Neutral DC
Key 1 TTG does not participate in the data hub 3 TTG serves traffic demand to other DCs but reactively
Highlights 2 TTG peers with content providers abroad
Key 1 TTG’s DC is carrier specific 3 TTG hub potentially competes with “open” alternative
Highlights 2 Designated area not including corporates
Key 1 TTG’s DC becomes carrier neutral 3 TTG DC becomes integral part of a unified Turkey DH
Highlights 2 Designated area not including corporates 4 TTG DC potentially within a regulatory “free zone”
Google, Akamai,
ICPs. Some T1s setup large base of smaller
presence. No IP gravity
Facebook, and most T1
players setup presence.
hubs in Turkey, in the
setup presence. Turkey
becomes a moderately- Turkey becomes a major case of the “proactive”
sized regional hub hub with global reach play
• Subsequent analysis
Hub Competition
Competing hub attracts Traffic is split between Own hub attracts Google, play is better fit under
Google, Akamai and the two hubs with Akamai and Facebook different scenarios
Facebook and majority of and majority of regional
regional demand demand
30 1 24 12
Insights
No
Change -24 Regulation has a critical
-131 -146 impact on the upside that can
-279 be realized from the data hub
opportunity
In the “Proactive” play, TTG
265
Moderate 181 166 can capture a significant
36 97 upside if it successfully
Change 23 8
attracts key tenants such as
-125 Google and Akamai (High
scenario)
548
In a “Hybrid” play, TTG
362 limits cannibalization, and
277
Data Hub 209 219 204 still captures a significant
114 71
Friendly upside
Incremental Net Impact - NPV Incremental Net Impact – NPV Breakup by Category
(TL Millions) (2015-2025) (TL Millions) (2025)
No change
Xyz Data hub-friendly
105 17 -2
114 -7 114
1 0 0 17 2 0
95 7 1 0 0
0: Status Quo 129
0: Status Quo
6
-2 -14 -24
Key Highlights
In the data hub-friendly regulatory scenario, a data hub is established in Turkey even without TTG’s participation
While TTG does not incur cannibalization, it looses significant market share and upside from international and DC
metro connectivity
Incremental Net Impact - NPV Incremental Net Impact – NPV Breakup by Category
(TL Millions) (2016 - 2025) (TL Millions) (2025)
No change
Xyz Data hub-friendly 544 -8 10 0 -4 5 1 548
1 6 10 0 5 1 0 5 1 2
548
30
302 Quo
0: Status 0: Status Quo
14 510
10
-1
Key Highlights
Int’l Connectivity revenues and DC metro connectivity are the main drivers of the total upside
IP Transit cannibalization is limited as transit share of TTG is already very low and mainly concerns national
access to its own eyeballs
No Corporate Connectivity cannibalization occurs in this scenario as competitors are not in TTG’s data centers
Incremental Net Impact - NPV Incremental Net Impact – NPV Breakup by Category
(TL Millions) (2016 - 2025) (TL Millions) (2025)
No change
Xyz Data hub-friendly 50 1 362
-2 0 1
362 313 -9 10 0 48
12 4 4 10 0 7 4 2
262
8
0: Status Quo 0: Status Quo
9 294
-3
Key Highlights
Hybrid behaves similarly to the “Proactive scenario”
International connectivity upside capture is somewhat lower than in the “Proactive” option because TTG does not
benefit from a ‘boost’ in its market share to the carrier specificity of its data center
Incremental Net Impact - NPV Incremental Net Impact – NPV Breakup by Category
(TL Millions) (2015-2025) (TL Millions) (2025)
No change
Xyz Data hub-friendly 313 -9 10 -291
4 0
71
147
Key Highlights
DC-related corporate connectivity incurs full cannibalization within 2 years
Benefits Risks
1. International Connectivity
Forecast traffic
Split traffic demand by Forecast TTG’s market Calculate TTG’s
demand attracted from
type share Revenues
the region
1.1 1.2 1.3
% Turkey % Traffic TTG
TTG Prices
share Split Market Share
Total
Total Int’l
TotalInt’l
Int’l Total
Traffic
Traffic
Traffic Int’l TTG IP
X Traffic X X X Transit
Volume
(per
Volume Revenues
(per to
country)
(per Turkey
country)
country)
Forecast traffic demand Split traffic into: Int’l Forecast TTG’s market Calculate TTG revenues,
based on connectivity, backhaul, IP Transit, and share based on regulatory costs, and cash flows
plays, and regulatory levers Peering levers as well as the play
adopted by TTG1
Input Calculation Assumption Intermediate Output Final Output
1) For example, play 1 enables TTG to lock-in demand and increase its market share
Sources: Strategy& analysis
Strategy& Confidential May 2015 Prepared for TTG 16
1. International Connectivity
• Google, Akamai, and Facebook • All major ICPs and T1s as well as
• Limited presence of major ICPs as well as most T1 setup a large base of smaller players
Impact • Some T1s setup presence presence setup presence
• No IP gravity • Turkey becomes a moderately- • Turkey becomes a major regional
sized regional hub hub with global reach
% of Potential
Traffic Captured
10% 40% 100%
Content • Current Law no. 5651 curbs • Regulated Free Zone where DCs • Removal of the restrictive
Regulation demand from affected tenants are exempted from 5651 legislation
Assumptions
1. International Connectivity PRELIMINARY
Reg. Moderate
40% 4.0% 5.0% 5.7% 5.7%
Impact Change
• Table above shows the Turkey’s share of the total target country demand
• The share driving the model is the product of regulatory impact and the share by play option under a hub-friendly environment
Assumptions Calculation
1. International Connectivity
Comments
• In the most general case, regional traffic to
Turkey will be split between TTG’s own hub
A (%) and the competing data hub
Regional
• TTG is a backhaul connectivity provider to
1 Traffic to the competing “open” data hub. TTG’s
Own Carrier-Specific
Data Hub Turkey market share in the long term is driven by
the number of players in the market,
• This in turn is heavily dependent on the
2 attractiveness / size market and hence the
regulatory environment
1- A (%)
Open Carrier-
• TTG’s gets 100% Market Share (MS) of the
traffic intended for its own carrier-specific
Neutral Data Hub
data hub
• Considering the above, the effective market
share of TTG will be:
Traffic to own data hub is A % of the total traffic to Turkey. TTG’s share of – TTG market share =
1
backhaul is 100%. A is non-zero only in the “Proactive” play = [A x 100% +
+ (1 - A) x MS for Open Data Hub] x adj
Traffic to competing data hub is (1 – A) % of the total traffic to Turkey. TTG’s
2 • An adjustment is applied for “reactive” to
share of backhaul for that portion of the traffic depends on regulation reflect attempts limit the data hub
Assumptions Calculation
1. International Connectivity PRELIMINARY
A B C
% TTG Own Carrier- TTG Market Share in
X 100 % + (1 - TTG Own-Hub Traffic) X X adj
Specific Hub Traffic “Open” Environment
A• Non-zero only for “proactive”. TTG’s Hub traffic depends on with which players TTG secures deals. A sensitivity
Comments
is conducted as part of assessment of the “proactive”. Multiplied with 100% market share as TTG is the de-
facto backhaul supplier in this case
C• Adjustment factor of 0.7 applied for “reactive”. This is to reflect TTG’s attempts to limit the attractiveness of the
data hub
Assumptions Calculation
1. International Connectivity PRELIMINARY
Assumptions Calculation
1. International Connectivity
-12%
99 90
-10%
60 1G
36 -12%
10G
100
-12% 30
238 27
-8% 21
144 16
96 11 9
6
Forecast metro Calculate market Forecast TTG’s market Calculate TTG’s Cash
capacity revenues share Flows
1.1
Total Int’l
Traffic to
Turkey
Forecast purchased metro Determine an appropriate Forecast TTG’s market Calculate TTG OpEx,
required BW for the different pricing level and calculate share and calculate TTG’s CapEx and cash flows
type of links emerging in the DC metro connectivity metro connectivity revenues
data hub revenue
Model Overview
• Model calculates the capacity required by type of link
2d T1 – RO
• Assumptions include % share of traffic by CP,
Data exchange through direct peering peering as % content traffic tenant is not collocated
Data exchange through the IXP in the same IXP node with other type of tenants
3.4
Moderate change
9.0 -28%
-10%
3.6 2.5
1.8 1.7 2.3 Benchmark average
0.7 0.3 1.0 0.4 0.2
1.9 for 1G in 2015
1.7 1.7
2015 2020 2025 1.5
1.2 1.1 1.1 1.5
Data hub friendly
0.8
-34% 0.6
9.0
-11%
3.6
1.8 1.1 0.5 0.2 0.6 0.3 0.1 TR TR ES BE FR TR AT SE DE IE TR NL UK
(15’) (20’) (20’) (20’)
2015 2020 2025
1) Prices have been decreasing over the previous years at a sustained 15%-20% rate
Sources: Strategy& analysis
Strategy& Confidential May 2015 Prepared for TTG 27
2. DC Metro Connectivity
Comments
-10%
• Competition in the leased line market is
65% -23% expected to strengthen over time
59%
56%
-40% • Accordingly, for the “hybrid” and
“liberal” plays, TTG’s market share is
45% expected to decline and reach more
43%
competitive levels
• For the “reactive” and “proactive” plays,
27%
TTG is not actively facilitating DC
metro connectivity needs of the
competing hub
• As a consequence, competing
infrastructure players supporting the
DCs are expected over time to capture
2016 2020 2025
a higher market share
Reactive / Proactive Hybrid / Liberal
Comments
OpEx as % Revenues (%)
• Assumed incremental OpEx / Revenues
27% ratio for P2P metro Ethernet service half
27%
26% than reported OpEx / Rev, which is
25% 25%
23%
calculated on a fully allocated basis
• Trend for “no change” scenario is in line
20% 20% 20%
with historical data. For other scenarios
we are assuming a further increase to
account for the effect of decreased prices
• A CapEx of USD 16,000 per Gbps for
upgrading metro capacity is assumed
based on a high-level model from
Strategy& accounting for metro switch
and edge CapEx
• The model considers that extra dig, duct
and fiber costs are not required
2016 2020 2025
Total Int’l
Traffic to
Turkey
Forecast collocation Determine pricing level and Forecast TTG’s market Calculate TTG OpEx,
demand from content calculate DC collocation share and calculate TTG’s CapEx and cash flows
providers, carriers and other revenue collocation revenues
types of tenants
Tier-2
10 1.0 4.0 1.5
Carriers
Regional
44 1.0 4.0 1.2
Operators
Sources: Strategy& analysis
Strategy& Confidential May 2015 Prepared for TTG 32
3. Data Center Collocation
Comments
Proactive (Own hub dominates subscenario)
• For “hybrid” and “liberal”, TTG’s DC is assumed
Hybrid / Liberal
to become the 3rd DC in size with market share
increase to 25% of new space annually in the, in
58% 58% light of that:
– TTG does not have spare capacity,
– DE-CIX announced their first DC in Turkey
and is soon to announce a second one
• For “proactive”, TTG’s share of new “data hub”
29% space in Turkey is significantly higher in the case
25%
that TTG’s own hub dominates, offering access
to a unique tenant ecosystem that includes the
15%
likes of Google and Akamai
10%
1.3
TTG’s Cash Flows – Cost assumptions
Capacity
Unit Cost Net Effect on Backhaul Costs:
UC0 UC1
B Δ(C) = V0 * UC0 – V1 * UC1
Forecast Backhaul Cost Savings Due to Traffic Localization Calculate Net Savings
4.1 4.2
Current All calculation done by
% Localized
Capacity upstream route to Europe
TTG Capacity
Localize Capacity Cost
Own X Needs X Savings
d Traffic Decrease Savings
Traffic
A 4.2
Capacity Unit Cost / Volume Net
Relationship - Savings
Forecast Cost Increase Due to Increase in Capacity Unit Cost Calculate Net Savings
Sources: Strategy& analysis
Strategy& Confidential May 2015 Prepared for TTG 37
75% 75%
74% 74% 10G 100G
71% 2.6 -11%
2.4
3.3
• Showing % for
data hub friendly -4%
scenario. Time
to reach 75% 2.3 -30%
takes longer in
1.2
other scenarios 1.0 1.0 1.0 1.9
• IP transit is not 1.6
localized (TT 1.3
47% continue to buy 1.1
IP transit abroad
to avoid
cannibalization
Vienna
Milan
London
Frankfurt
Sofia
Amsterdam
of existing IP
transit revenues)
34% 2015 2020 2025
3.1 3.2
% DC Inputs from TT Additional Inputs from TT
TTG
Share Corporate Churn Corporate Prices
Total
Total
Total Int’l
TT
Int’l TotalTTInt’l TotalTTInt’l
Total Int’l
Corporate Total Int’l
Corporate Incremental
Traffic
Corporate
Traffic Traffic
Traffic Traffic
Traffic
Volume X DC
Volume x DC
Volume X Revenues /
Connectivity
Volume Volume Volume
Related Losses
Related
(per (per
(per
Lines
(per (per (per
Lines
country)
Lines
country) country)
country)
country) country)
Calculate DC related revenues that Forecast additional churn on total DC Forecast cannibalization impact based
would be affected by the TRDH related lines on additional churn and revenues
No incremental price erosion vs. the
base case (“reactive”, “no change”) is
assumed1
Input Assumption Intermediate Output Final Output
1) Decreasing prices to fight churn of DC-related connectivity services would have an averse impact on non-DC related services
Sources: Strategy& analysis
Strategy& Confidential May 2015 Prepared for TTG 39
5. Corporate Connectivity
3c: “Liberal” 100% of total base 100% of DC related revenues lost within 2 years of operations
3.2
Price – TTVPN Product (85% of DC-related connectivity revenues)
TTVPN Blended Prices Revenues, lines, and prices are as per the
business plan
(2016 – 2025) (TL Thousands / Line / Year)
This assumes an upsell in TTVPN, which results
+10% in price increase per line
Price evolution will remain the same across
538 738
303 scenarios due to the small portion of revenues
affected, which does not warrant a reduction in
overall prices from TTG
2016 2020 2025
Strategy& Prepared for TTG 40
Assumptions
6. Enterprise Cloud
5.1
% Enterprise TTG TTG Costs
Market Share
Total
Total
Int’l
Int’l
Traffic
Traffic
ICT Addressable
X Cloud X TTG X TTG Cash
Volume
Volume
Market Cloud Flows
Market
(per
(per
countr
countr
y)
y)
Forecast Addressable Cloud market Forecast TTG’s market share of Calculate TTG revenues, costs, and cash
demand for Turkey and Region enterprise cloud market in Turkey and flows
Region
Hub Friendly 3% 14% Open data hub maximizes competition and further reduces TTG’s potential market share
Assumptions Calculation
6. Enterprise Cloud PRELIMINARY
Moderate Change in regulation will enable regional operators to open up more and thus access to these markets would be
0% 2%
Change easier
Hub Friendly 0% 4% Open data hub optimizes competition and further reduces TTG’s potential market share
Assumptions Calculation
7. IP Transit
Internet Comments
at large
• Turkcell / Superonline have an
aggregate internet traffic estimated at
~1.5 Tbps
• Turkcell / Superonline are currently
purchasing over 600 Gbps of IP Transit,
of which only 30 Gbps from TT
0.5 Gbps
~ 600 Gbps • The totality of IP Transit purchased by
Superonline is used to reach TT
eyeballs. Virtually none of it (0.5 Gbps)
transits TT to the internet
• Accordingly, the IP transit module
30 Gbps
assumed that 90% of current TTG’s IP
transit traffic is for own eyeballs and
hence shielded from cannibalization
• TTG does not participate in the “data hub” ecosystem in an attempt to limit its
Strategic Intent attractiveness, keep global carriers corporate and wholesale arms at bay and hence
protect TT’s domestic revenues
• TTG lobbies with the regulator to elevate the disadvantages of a data hub and counter
Regulatory Strategy
efforts aimed at creating a more hub-friendly regulatory environment
• Attract ICPs, carriers and regional operators within a controlled environment that protects
TT from revenue cannibalization and with the objective to not only generate a financial
Strategic Intent
upside but also to inhibit the growth of the ecosystem of competing data hub. Offer
incentives to players that TTG cannot capture to stay outside the country
• Use TT’s own Carrier-Specific data center, by reserving space in TT’s data centers –
hence, mixed-use space for corporate / wholesale domestic / global customers
Data Center Strategy
• Carve out space for “data hub” use in both Istanbul and Ankara data centers while
providing competitive backhaul service between the data centers
• Create an interconnect mechanism within own DC with clear business rules to start with
IXP Strategy • If interconnect demand grows substantially with smaller players in the DC, start TT’s own
IXP, which may be either self-managed or managed by an international IXP operator
• TT sells domestic services to ISPs and Corporates exclusively with its published, regulated
pricing, while TTI serves the ecosystems of global players in the hub
Regulatory Strategy
• Show the regulator consistency in treating all tenants, and negate the need to peer with
ISPs on a third-party IXP by ensuring the ecosystem exists in TT’s own data center
Sources: Strategy& analysis
Strategy& Prepared for TTG 47
“Hybrid” improves tenant value prop by adding carrier
neutrality but limits DC accessibility to telcos and CPs
TTG Option Definition – “Hybrid”
• Attract global carriers and content players to Turkey to serve the region and increase the
Strategic Intent country’s IP Gravity, while ensuring that the presence of these global players does not
substantially affect the local market
• Set up a data center owned and operated by a non-regulated third party, as a dedicated
data hub ; a carrier hotel only
Data Center Strategy
• Carrier neutral facility: besides international players, it will accept only domestic
infrastructure operators (Vodafone, Turkcell, Turk Telekom). Not ISPs, etc.
• The data center owner invites any independent IXP to join the data center.
IXP Strategy
• TTG may also look at partnering with a global IXP to introduce them into this data center
• TT sells domestic connectivity services from/to the data center using its published,
regulated pricing, and competing with Superonline and Vodafone on that front
Regulatory Strategy
• TTI serves the ecosystems of global players in the hub internationally
• TT requests special subsidized prices from the regulator to interconnect “data hub” facilities
Sources: Strategy& analysis
Strategy& Prepared for TTG 48
“Liberal” aims to maximize the attractiveness of TTG’s
value prop to ICPs and hence int’l connectivity revenues
TTG Option Definition – “Setup Liberal”
• Attract a maximum of international players to create data Gravity in TT’s data hub, in order
Strategic Intent to substitute cannibalized domestic Corporate and Wholesale revenues with new
wholesale revenue streams, international and domestic
• Tier-1 carriers looking to sell IP Transit to regional ISPs and MNOs directly
• Global Corporate Connectivity providers looking to serve regional Corporates
Target Tenants
• International Content Providers: OTTs, CDNs, Cloud Providers, etc. looking to peer with
regional ISPs and MNOs directly
• Partner with a global name to project world-class neutral management, while retaining
equity in the data center and a substantial share of its revenues
Data Center Strategy
• Purchase / repurpose / build two locations in Istanbul to cater for the demand of space that
is close to submarine global routes and to the country’s largest city at once
• Encourage an independent international IXP to run in your data center in order to attract a
maximum number of content providers and regional operators (incumbents as well as
IXP Strategy
alternatives) into the location
• Create a virtuous cycle of ecosystem tenants to further increase the location’s IP Gravity
TTG’s Role
• As Turkey’s main telco infrastructure provider, TTG can
have an important stake in this market
2d T1 – RO • However, current regulated TTG prices are high - the
Data exchange through direct peering establishment of the hub could exert downward pressure
• This is an important risk that is examined separately from
Data exchange through the IXP the upside that is the focus of this section
Strategy& Confidential May 2015 Prepared for TTG 51
Summary slide on the mechanics
METHODOLOGY DETAILS IN APPENDIX
1.1
Capacity dimensioning: Calculation methodology summary
CP-RO direct • Model splits CP traffic from regional countries1 by CP and RO. If the traffic
• % share of traffic by CP
for a given CP-RO pair exceeds a certain threshold, then the CP and RO
peering • # ROs by target country in hub
will peer directly as opposed to exchanging traffic through the IXP. Total
traffic • Direct peering threshold
traffic is the sum of all direct peering traffic across all CP-RO pairs
• If the traffic for a given CP-RO pair is below threshold (see above), then the
• Same as above, plus: IXP
IXP traffic CP and RO will exchange traffic through the IXP. IXP also accommodates
peering as % content traffic
peering traffic, which is calculated as a % of total content traffic
• CPs buy connectivity from T1s to ‘fetch’ content from other data hubs when • Miss ratio
this is not available locally. This occurs with a low probability, the so called • Probability of CP being
CP-T1 traffic ‘miss’ ratio. Assuming that the CP and T1 are located in different DCs with located in a different DC than
probability A, it is CP-T1 traffic = CP-RO traffic x miss ratio x A its T1 vendor
• Intra–CP links are needed for CPs to ‘mirror’ the content uploaded in one
Intra-CP server to another geo-redundant server. Uploading of content represent a • Upload traffic as % of total CP
traffic small portion of the total traffic U (%) that depends on the upload : traffic
download ratio. Hence, intra-CP traffic = CP-RO traffic x U (%)
• The model calculates average IP transit traffic by T1-RO pair = IP transit • # ROs and # T1s
traffic1 / average # ROs per T1, where the latter is = ROs * T1 vendors per • T1 vendors per RO
T1-RO traffic RO / T1s. The model then multiplies the result with the probability of T1 • Prob. of RO being located in
being in different DC than an RO to get average T1-RO traffic different DC that its T1 vendor
1) Input from international connectivity module
Sources: Strategy& analysis
Strategy& Confidential May 2015 Prepared for TTG 52
Summary slide on the mechanics
1.1
Technical parameters
• 37.1%, 32.3% and 19.2% for Google, Akamai and Facebook respectively –
% share of traffic by CP
partially based on TTG data
• Varies from 1-3 by target country depending on Turkey’s share the
CP-RO direct # ROs by target country in hub
considered country; total # of operators grows accordingly to 46 over time
peering and
• 80 Gbps of traffic – if more than one 100G port is needed, the CP and RO will
IXP traffic Direct peering threshold
typically peer directly
IXP peering as % content traffic • 20% – typical ratio based on project experience
Upload as % of total CP traffic • 9.1% – based on a typical upload : download ratio of 1:10
Intra-CP
traffic • Grows to 19 over time – fed from the DC colocation module which assesses
# T1s
the entry of different T1s based on # addressable eyeballs considerations
• 2 – ROs typically maintain IP transit relationships with 3-4 T1s, of which 2 can
# T1 vendors per RO
be assumed to be in the Turkish data hub
T1-RO traffic
Prob. of RO being located in a • 40% – based on that T1s will most often collocate in 2 out of 3 nodes of the
different DCs than its T1 vendor hub
1) Input from international connectivity module
Sources: Strategy& analysis
Strategy& Confidential May 2015 Prepared for TTG 53
The establishment of a data hub creates a market for
connectivity between different hub players
DC metro connectivity module methodology – CP-RO direct peering traffic
Methodology
2a CP – RO
Methodology
Methodology
2b
IXP • CP–T1 links needed for CPs to ‘fetch’ data
Node CP – T1 from their servers other data hubs
A
• This occurs with a probability 15%-20% (so
called ‘miss’ ratio)
• Traffic will flow through a metro link only if
the CP / T1 are collocated in different DCs
• The calculation methodology is accordingly
as described below
IXP IXP
Node Node • Input: Total CP-RO traffic as calculated
B C previously
• Step 1: CP-T1 traffic = CP-RO traffic x
Internet miss ratio x probability T1 is in a different
location than the CP
Methodology
Methodology