Xem mẫu
- International Journal of Management (IJM)
Volume 11, Issue 3, March 2020, pp. 346–353, Article ID: IJM_11_03_037
Available online at http://www.iaeme.com/ijm/issues.asp?JType=IJM&VType=11&IType=3
Journal Impact Factor (2020): 10.1471 (Calculated by GISI) www.jifactor.com
ISSN Print: 0976-6502 and ISSN Online: 0976-6510
© IAEME Publication Scopus Indexed
DEMYSTIFYING DATA CENTER COSTS AND
PRICES
Sarvesh Kumar Tripathi
Mother Dairy Fruits & Vegetable Pvt Ltd, India
Avjeet Kaur
KR Mangalam University Gurugram, India
ABSTRACT
Management of costs on data is assuming significance with data growth. As
information technology (IT) costs are spread in firms, thin and complex, perceptions of
costs on data are hazy. Such haze become mist when cost basis for the data costs are
different for service provider and service consumer. Service consumers are offered data
services on the basis of per GB/ TB data stored, moved, retrieved or processed. While
data center costing is commonly based on the amount of electricity requirement like 10
MW data center. Range of costs estimates for data center in terms of $ per TB per month
is observed irrationally high, while range of data center services prices are observed in
rationally narrow range of TB per month.
Key Words: Data Centre Cost, Data Centre, On-cloud service, TCO, Costing for IT,
Data cost, Costing of data.
Cite this Article: Sarvesh Kumar Tripathi and Avjeet Kaur, Demystifying Data Center
Costs and Prices, International Journal of Management (IJM), 11 (3), 2020, pp. 346–
353.
http://www.iaeme.com/IJM/issues.asp?JType=IJM&VType=11&IType=3
1. INTRODUCTION
We are in transition from data age to digital intelligence age. Artificial intelligence requires
data as much one can accumulate. Digital data generation is involuntary and storage a
compulsion. People are indifferent to usefulness of data stored. Data growth is exponential and
same is the case with associated costs. It is feared approaching, a data tsunami in upcoming few
years (Tripathi Sarvesh Kumar). Costs on data is a concern for every individual or enterprise
of our times. Such concerns and compulsions are getting manifested into booming data center
business. Business wire has projected CAGR of 17 % for global data center market for the
period 2017- 2023 (Maida).
Direct costs to the individuals are camouflaged under hype of fast launched new mobile
phones, new versions appearing in market every other day.
http://www.iaeme.com/IJM/index.asp 346 editor@iaeme.com
- Sarvesh Kumar Tripathi and Avjeet Kaur
Indirect costs flowing through availability of free data storage offered by firms owning large
data centers of large information technology like firms Google and Amazon are keeping cost
concerns away from individuals, at least now. All data centers consume huge amounts of
electricity, multiplying carbon footprints thereby indirect costs to individual on the planet earth.
Though data storage facilities offer storage quota for free, in reality these are not free, other
users of data center share the costs. More over environmental costs are not considered as of
now. Phrase “There is no free lunch” applies to data storage too. World is waking up for the
concerns on the carbon footprints and environmental costs.
According to Pranav Prakash, the presales consultant at Photon, the data canter carbon
footprint is palpable:
“Seventeen percent of the total carbon footprint caused by technology is due to data centers.
The electricity that is needed to run these data centers is nearly 30 billion watts. These servers
waste 90 percent of the energy they use because they run on full capacity all day long.”
(Michel)[2].
For firms, costs are directly associated with size of data owned or held by firm. As data
sizes are rising costs are also rising, in spite of adoption of the energy efficient technology in
data processing and designing a data center. (Shehabi). Shift from the paradigm “Digitize all,
store all, store forever” (Tripathi Sarvesh Kumar) to “Store trash free Data” is expected to cost
reductions for firms and individuals both.
As information technology (IT) costs are spread in firms thin and complex, perceptions of
costs on data are hazy. Such haze become mist when cost drivers/ cost basis for the data costs
are different for service provide and service consumer. Service consumers are offered data
services on the basis of per GB/ TB data stored, moved, retrieved or processed. While data
center costing is commonly based on the amount of electricity requirement like 10 MW data
center or 100 MW data center.
In absence of the electricity efficiency standards for data centers, data size handled per unit
of electricity is expected to vary from operator to operator (Shehabi).
In order to rationalize costs on data, firms choose to avail data center services with
compromises on security and privacy. Current trend of mass migration to cloud deserves
examination of rationality behind such decisions. Option of migration to cloud may appear
cheap but first impact is loss of ownership of data and data holding infrastructure. Maurizio
Naldi and Loretta Mastroeni observed that –
Since migration involves a significant one-off effort to move all the data on the new
platform, and that can be exploited by cloud providers to cage the company in a lock-in
condition, the decision to migrate has to be examined thoroughly (Maurizio Naldi) .
Comparative prices offered from cloud computing service providers is natural process, but
comparison of costs could bring in some in-sights for the customer to take decision about what
proportion for data to move on cloud and what to be kept captive from importance of data and
its ownership considerations.
This paper aims at review of the secondary literature on the costs on the data in data centers
and prices of data center services. Typical comparison is presented on the data center costing
and data service pricing, prevailing in the market place. Attempt is made on arriving at similar
costs and prices basis i.e. per TB of data held or owned by a firm for a period (Monthly), popular
as total cost of ownership (TCO). Since IT costs of the firms are fast aligning towards size of
data held by it, aligning basis for costs and prices would enhance visibility to the bearer of costs.
Study is based on the secondary source of rare information. Considered assumptions are
applied for critical input needed for the derived calculations.
http://www.iaeme.com/IJM/index.asp 347 editor@iaeme.com
- Demystifying Data Center Costs and Prices
2. OBJECTIVE
To review cost of data held with individuals and firms. Review of costing of the data centers,
derivation of the costs at common costs driver i.e. Costs in $ per TB of data stored in data center
and used from data center for a month. Finally, costs comparison is made with data center prices
offered by popular data center service providers in marketplace.
3. METHODOLOGY
Latest literature is reviewed for costing models for data centers from secondary source of
information. Available data on costs at different cost driver’s basis converted to common cost
driver of cost per TB data per month by selecting relevant sources. Offer prices for the data
Centre services are collected and compared with the costs on common costs driver.
3.1. Observations
Perception of costs incurred on data transfer to data storage media, storage and retrieval is vague
for individuals. Individuals do not care for it separately. For individuals costs on data are inbuilt
in the costs of digital device. IT Firms are keeping data cost information to themselves as trade
secret. Firms also have problem in costing for data, due to absence of the costing model
agreeable to all and invisibility of IT costs spread everywhere in the organization (Pramanik) .
IT function operates on the budgets as percent of the sales turn over. IT product and service
vendor enjoy value based and skimming pricing strategies. Competitor based pricing strategy
is slowly emerging with competition in the on-cloud service market.
There is unanimity on the major cost components of the data Centre but costs on per unit of
the data differs due to grey area of the data storage capacity per server.
3.1.1. Costs on data storage for individuals
Though small but, individual’s store of data may be termed as a data Centre. Individuals have
built quite good size of the personal data bank since advent of desk top computer and transition
to laptop, palm top, tablet, mobile phones, and digital camera like electronic gadgets. Data size
is increasing and managed loosely. Costs on data held by individuals is inbuilt into costs of fast
replacing electronic gadgets. Hence, individual is not separately concerned with cost of storage
of his data. Data of Individuals is stored in compact discs, flash drives, hard drives, computer
discs, laptop or mobile phones. Many have begun storing data on cloud data storage facilities
like the one offered by Google. One can store data up to 15 GB at Google on-cloud facility for
free.
Typical data size stored for individual is in the tune of 1 TB, 2TB at the maximum. Since
individuals are getting increased data storage capacity with each time he upgrades the electronic
gadgets, (S) he is not concerned on data costs separately. As a consequence individual is
indifferent to the costs on the data held by him. Ever increasing size of data, multiple back-up
of same data poor data inventory management practices and tools will soon force individuals to
take cognizance of data costs separately.
Individuals store personal, live data (always ready to be accessed) in mobile phones or
computer hard discs. Off-line or cold data which is accessed occasionally is stored in CD, Flash
drives or standalone hard drives. Only prevailing rule to store and manage data is to store all
and let us keep one more copy of data, in case it is lost. As a result there is sizable portion of
redundant and useless data kept till medium holding data is functional and not lost. Whatever
insignificant, cost on data are rising for the individual owner of the data.
Following section provides feel of costs on data based on the prevailing prices in the
marketplace.
http://www.iaeme.com/IJM/index.asp 348 editor@iaeme.com
- Sarvesh Kumar Tripathi and Avjeet Kaur
3.1.2. Cost on Data held by individual person
In order to have assessment of the cost on data held by individual at prevailing market price, 03
gadgets of popular brands are chosen. Average of price differential is calculated for 02 models
of the gadgets. 30% price is attributed to the additional memory offered by each brand. Table
01 has estimated data cost per GB. Nominal incremental price corresponds to cost on one GB
additional data stored in mobile phones. Thus, estimated cost of data stored in mobile phone is
approximately $ 19.00 per TB per month.
Offline data stored in the flash drives works out to $ 12.50 per TB per month as in table 02.
And cost of data stored in external hard drives is approximately $ 0.4 per TB per month as in
table 03.
Above costs do not include invisible environmental costs. Though costs are insignificant
for an individual but collectively make them to be huge, especially environmental costs. A
rough estimate of plastic trash generated by plastic memory cards (weighing 2 gm) and flash
drives (weighing 25 gm) turns out to be 4000 MT and 47000 MT respectively, in the world.
This plastic waste is estimated with the assumption that quarter of world population has
discarded at least one memory device each in his life time. In reality individuals discard tens of
them in just 05 years’ time frame. Drops of digital plastic waste are filling up oceans of plastic
garbage. Situation is not better than recognized bad environmental impact by one time use
plastic waste.
3.1.3. Costs on data in the firms
Data storage costs are subset of other IT costs, but all other IT costs depend upon size of data
stored by the firm in-house or outside firm, on-cloud. Still per GB/ TB data stored is not
common cost driver in firms. In most of the cases, aggregate IT costs are taken into allocation
of indirect costs of product or service costing of firms. Prevailing practice of aggregate IT
costing keeps many data storage costs invisible within firm itself.
3.1.4. Information Technology (IT) Costs for Firms
Published material on actual costs incurred by firm on managing its digital data is irregular, in
absence of costing standards for IT costs. Such costs are spread among different departments
and pooled together for cost allocations as over heads. There seems incentive to firms for
keeping such costs invisible outside the firm.
Data storage has many hidden costs. What seems to be obvious becomes fuzzy when you
try to see it closer.
It is known to all, cost of hard drives is falling and their capacity is rising faster. Moore’s
law applies in this situation, and this often leads us to conclude that data storage is cheap. In
reality IT budgets of the firms are trending upward.
Like everything in IT, the true cost of a component is often disguised with lots of hidden
costs. These costs are often ignored causing budgets to blow out and, worse, system reliability
can be compromised. Any business owner knows or should know the cheapest part of any
computer system is hardware. But the operation, power and support costs will far outweigh the
hardware costs.
One of the drivers for upgrading, is running out of storage space because, like just about
every other small business, data is never deleted. 160 GB disks becomes 4 TB each very soon.
Addition of the disc is only solution deployed to deal with problem of growth of data.
http://www.iaeme.com/IJM/index.asp 349 editor@iaeme.com
- Demystifying Data Center Costs and Prices
Customer went from 40GB disks to 160GB. Customers have a habit of filling disk space
with data that is no longer required, multiple copies of the same files, and so on. Unless we can
find a way to re-educate the customer, in five years’ time we’ll be going through the same
upgrade cycle with them again.
There are various hidden costs associated with network storage beyond just buying a larger
hard drive. The first hidden cost is need for data to be backed up.
Other hidden costs need to be factored into the storage cost. Disaster recovery times blow
out when small businesses keep data that’s no longer needed by them. This is the one metric
you never want to be worried about, but if you have a guarantee that you’ll get the client up and
running 24 hours, time taken to copy the data onto replacement hardware eats into that time.
3.1.5. Shadow IT costs
One study suggests (Pramanik) that head of IT department of the firm (CIO) has only 60%
control on spending on IT, rest is spread in other departments or functions of same firm as
shadow costs. IT spending outside the IT department escape classification under IT costs, often
aggregates under other overheads or under cash spending. Following observation from the study
is common scene in the non-IT firms.
“Individual employees up to the head of sales are spending money on technology because
they see interesting or exciting opportunities to improve the business,” Horne explained. “They
want to experiment with technology. To the extent that is going on, we think it is healthy. But
it is unhealthy if it is just duplicating what the company is already doing, or if the smaller teams
are buying hardware that the company could get a better deal on through central purchasing.”
(Pramanik).
Some portion of shadow IT costs are incurred by employees too from their personal budget,
like knowledge upgrade and safety back up of data. As these costs are not significant, kept out
of scope of this paper.
3.1.6. Distribution of IT Costs with captive data Centre
Distribution of IT Costs with captive data Centre are presented in Figure- 01.
3.1.7. Distribution of IT Costs with on-cloud data Centre service
With the assumption of 50 % manpower relocation, distribution of IT costs with on-cloud data
Centre is shown in figure – 02.
Small or big IT department as separate functional group function is mandatory for any firm
today. They all have to continuously comparing the IT services bought or on lease. Buy or lease
decision depends on the cost per unit of data held and used by firm.
3.2. Data Centre
Data centers are centralized locations where computing and networking equipment are
concentrated for the collecting, storing, processing, distributing or allowing access to large
amounts of data. They have existed in one form or another since the advent of computers. Term
data Centre evolved through computer room in year 1940, Internet data Centre in the year 2000,
cloud data Centre in 2002.
IT department as separate function of the non-IT firms, generally, housed all the computing
equipment and resources till initial years of second millennium, say 2005. IT department
functioned as separate cost Centre and maintained all data of the organization as well.
Resources needed for storage of data also remained sub section of the IT department.
http://www.iaeme.com/IJM/index.asp 350 editor@iaeme.com
- Sarvesh Kumar Tripathi and Avjeet Kaur
As size of data swelled, cost size of the data holding infrastructure also swelled. Costs to
power the data holding infrastructure, costs on back up and costs on the cooling system for the
servers popped out of the IT department expenditure and budgets. In-house (on-premise) data
centers thus assumed separate identity. IT costs then, got clear bifurcation between data Centre
and rest of the IT services / functions.
Data centers built and maintained by very large firms like Amazon and Google started
offering data storage service by 2002 at prices lower than the costs on captive data centers. Such
services are popular now as on-cloud data Centre services (Maurizio Naldi).
Data Centre costs are subset to the IT costs of the firm, irrespective of the nature of the data
Centre services in-house or on-cloud.
3.2.1. Data Centre costs
Secondary information on the subject is rare. Probably, firms do not want to disclose in public
domain. Interesting area for a separate research. This work is an initiation to bring forth closely
guarded information, mostly available with large scale data Centre service operators. Limited
work could be spotted in secondary information sphere, that to, on different bases. Some have
presented data Centre costs on overall investment basis, like a promotional paper from
Government of Gujarat , India (Government Of Gujarat), others have presented data Centre
costs on a per rack costs basis and some have been worked on the power consumption basis
like costs on 10MW data Centre (Chandrakant D. Patel). None could be located for costs on per
GB or TB of data handled.
3.2.2. Constituent components of the data Centre costs
Major cost components for the data Centre same as any other commercial enterprise like land,
labor, machinery, electricity and technology. As data centers are characterized by high
consumption of electricity for computing and cooling, electricity costs are major cost
component. Other significant cost components are costs on software and wages to technical
staff. Common Cost components of the data Centre with prevailing standards are compiled in
table 04.
3.2.3. Estimation of Data Centre Costs
Three sources are selected for deriving and comparing the data Centre costs in $ per TB per
month. One for Indian Territory and other two from USA. Comparison details are presented in
TabIe-05.
3.3. Estimated costs on data Centre
Source -1: Promotional document form Government of Gujarat has pegged an investment of
INR 190 Cores ($268 Million) on 1080 racks data Centre (Government Of Gujarat). Cost
working is not extended to per GB data handled by such data Centre. Based on electricity
consumption of 250 W, this data Centre will require approx. 10MW electrical consumption.
Estimated costs per GB per month is estimated at $ 0.01 per TB per Month.
Source-2: Another source has estimated cost of $20,000 per month for 10 KW data Centre
in American environment (Chandrakant D. Patel). Cost estimation from this source turn out to
be $ 29.76 per TB per Month.
Source – 3: White paper from uptime institute has pegged at $ 2083,333 per month for 2176
KW data Centre in American environment. Over all data Centre costs (Taylor)]. Estimated costs
$ 1.14 per TB per Month.
There is wide variation in the costs on data Centre in terms of per unit data handled. This
variation need to be validated through future research. On reason for the variation in the data
http://www.iaeme.com/IJM/index.asp 351 editor@iaeme.com
- Demystifying Data Center Costs and Prices
Centre costs could be size of the data enter impacted by economy of scale. Bigger the size of
the data Centre, lower is the cost.
3.4. Impact of size on data Centre costs
Following economy of scale, costs on the data Centre fall with rise in the size of the data Centre.
Figure -03 shows the relation between data Centre costs and size of the data Centre.
3.5. Price comparison for on-cloud services
Prices for the on-cloud data Centre services from 05 nos. popular service providers in global
market are compared. Prices components of data transfer in , data transfer out , put requests and
get requests are not to be paid separately in case of captive data Centre. Hence over all prices
are considered for comparing price with the costs. These prices range between $ 24 to $ 34 per
TB per Month as in Table - 06.
4. CONCLUSION
Costs on data with individual is observed in the range of $ 0.4 to $ 19 per TB per month of the
data. Very high variation from $ 0.01 to $ 29.76 per TB per month has been observed in Data
Centre costs, estimated through different sources of the information. Variation in the data
Centre service prices is observed in narrow range from $ 24.08 to $34.67 per TB of data handled
per month. Vide variation in the data Centre costs is subject for the future research.
REFERENCE
[1] Chandrakant D. Patel, Amip J. Shah. Cost Model for Planning, Development and Operation of
a Data Center. Palo Alto: HP Laboratories, 2005.
[2] Government of Gujarat, India. “Establishment of Data Centre.” 8Th Global Summit.
Gandhinagar: Industrial Extension Bureau, 01 2017. Report.
[3] Harvey, Cynthia. “Cloud Storage Pricing: Top Vendors Price Comparison.” 26 04 2018.
https://www.enterprisestorageforum.com/cloud-storage/cloud-storage-pricing.html. Report. 02
12 2019.
[4] Luxembourg, Dr. Yvan Philippe. “IT Costs – The Costs, Growth and Financial Risk of Software
Assets.” 05 2013. http://omtco.eu/wp-content/uploads/OMTCO-IT-Costs-The-Costs-Growth-
And-Financial-Risk-Of-Software-Assets.pdf. Document. 02 12 2019.
[5] Maida, Jesse. “Global Data Center Market Outlook 2019-2023 | 17% CAGR Projection Over
the Next Five Years.” Technavio Research 23 08 2019: 1.
[6] Maurizio Naldi, Loretta Mastroeni. “Economic decision criteria for the migration to cloud
storage.” European Journal of Information Systems, 25:1, (2016): 16-28.
[7] Michel, Isberto. https://www.colocationamerica.com/blog/data-center-environmental-impacts.
03 05 2018. Blog. 20 11 2019.
[8] PonemonInstitute. Cost to Support Compute Capacity. Traverse City: Ponemon Institute, 2016.
Research Report.
[9] Pramanik, Debasish. “What is Shadow IT? Necessity & Its Impact on Enterprise Security.” 25
10 2017. https://www.cloudcodes.com/blog/what-is-shadow-it-and-its-impacts.html. Blog. 02
12 2019.
http://www.iaeme.com/IJM/index.asp 352 editor@iaeme.com
- Sarvesh Kumar Tripathi and Avjeet Kaur
[10] Shehabi, A., Smith, S.J., Horner, N., Azevedo, I., Brown, R., Koomey, J., Masanet, E., Sartor,
D., Herrlin, M., Lintner, W. United States data Center Energy Usage Report. Berkeley:
Lawrence Berkeley National Laboratory, 2016. Report.
[11] Taylor, Jonathan G KoomeyPitt TurnerJohn StanleyBruce. “A Simple Model for Determining
True Total Cost of Ownership for Data Centers.” Santa Fe: Uptime Institute, 01 2007. White
Paper.
[12] Tripathi Sarvesh Kumar, Chhabra M, Pandey RK. “Taming Tsunami of Data by Principles of
Inventory Management.” Journal of Business and Management (IOSR-JBM) (2018): 01-12.
[13] Dnyanesh Sarang, Srikanth Pingali and Giri Ganesh. Total Cost Modelling and Strategic
Sourcing Strategies for Data Centres in Upcoming ERA. International Journal of Management,
6(8), 2015, pp. 57-70.
[14] K. Selvakumar. Strengthening User’s Control of Data in the Cloud Service- an Innovative
Auditing and Logging Mechanism. International Journal of Computer Engineering and
Technology, 6(12), 2015, pp. 44-59.
http://www.iaeme.com/IJM/index.asp 353 editor@iaeme.com
nguon tai.lieu . vn
