Cloud Computing is being projected as next big thing in IT. Each company is coming up with their own set of offering and lapping up the customer with pay as you use model.
Their are multiple models in work for cloud. But let us start by first understanding the meaning of cloud computing and basic models at work for the same.
Cloud Computing : This is a model wherein a service provider offers pay per use model to its clients. Clients can lease storage space or processing power or applications from the service provider and the price being charged is only for the resources being used by the client. This is like extending the T&M rate card for resources to infrastructure and Software service.
Thus multiple benefits flows to end customer like :
1. No upfront investment in IT infrastructural cost
2. Competitive cost - Since vendor is enjoying economies of scale in cost like hardware and hosting, thus the benefits trickle down to customer
3. Sharing of Resources : The resources will be shared across multiple clients, thus fraction of cost only to get charged.4. Better Up time & SLA : Vendor will be able to give better SLAs in comparison to owned infrastructure.
5. Elasticity : The customer can release the additional units not being used and at the same time can increase the same without any hassle.
6. Easy Entry & Exit : Minimum or nil exit cost.
Owing to above benefits the cloud is being considered to be a win - win situation for both vendor and customer.
The cloud is being divided in two parts :
1. Public Cloud : This is a cloud where sharing of Infrastructure is being done across customers. Example of it can site like Big Daddy, Google, Sales Force and other such sites which offers hosting space to users and the features
2. Private Cloud : This is same as public cloud, with an additional feature of customization being done to suit the needs of the customer, like extra layer of security or data privacy being added.
Conceptually private cloud seems to be a good business and technical model. Which have the economies of scale and security of hosting model where customer owns and manage the entire infra.
But the devil is in detail. Private cloud if we look closely is more of a hosting model. The client though will not have the infrastructure in its name, but will effectively pay for the same. Let us try to look at the pricing model of a private cloud for a full scale Integrated Service Deal, involving hardware, software, Implementation, and maintenance. Each of this element has its own pricing method, lets deal with each one :
1. Infrastructure :
This will include components like Servers, racks, routers etc. In this case the hardware to be deployed will be customer specific, since hardware is derivative of application requirements, response time, Up time and security being required, the combination of this is unique to each organization with minimal capability of being shared with any other organization. In such scenario, wherein hardware is being set up for one particular customer, the cost will have to be recovered from such customer only.
Recovery to be made on pay per use model on monthly basis. Unit rate are to be agreed, with necessary checks like minimum commitment, ARC/RRC etc. The vendor will thus has to ensure recovery of following at minimum commitment :
a. Cost of Hardware - Or cost plus minimal profits
b. Finance Cost : Since the money is invested upfront and recovery is made over time, thus finance cost also needs to be recovered.
c. SLA/Contingency : This will depend on how the deal has been negotiated. If possible, vendor will try to recover even the possible SLA/contingency cost at minimum commitment level.
2. Applications :
This is a tricky area, since their are multiple vendors and varying terms. Most of the licenses are named licenses, especially the end used licenses, typical case in point being ERP licenses. Thus, this does away with any benefit of cloud. However, their can be Server licenses which can be shared across multiple users and bought in as pool. However, such licenses are on rental model and again pay per use. The cost of licenses in such case is considerably higher than the named licenses. Again, the vendor has to recover the following cost from the customer at minimum commitment:
a. Cost of Application/License - Or cost plus minimal profits
b. Finance Cost : Since the money is invested upfront and recovery is made over time, thus finance cost also needs to be recovered.
c. SLA/Contingency : This will depend on how the deal has been negotiated. If possible, vendor will try to recover even the possible SLA/contingency cost at minimum commitment level.
3. End User Computing
This will cover the end user devices like desktop, laptop, routers, switches etc. These assets will be handed over to customer and loses their value quickly. A used laptop will have a little rental value especially when it has to be used over a larger period. And these assets also have higher wear and tear. Again, the vendor has to recover the following cost from the customer at minimum commitment :
a. Cost of Equipment - Or cost plus minimal profits
b. Finance Cost : Since the money is invested upfront and recovery is made over time, thus finance cost also needs to be recovered.
c. SLA/Contingency : This will depend on how the deal has been negotiated. If possible, vendor will try to recover even the possible SLA/contingency cost at minimum commitment level..
4. Customization/Implementation/Migration Cost
In large deal, vendor offer migration as freebie, and pay out of their margins. The same can happen in cloud computing. But in case of customization and Implementation, specialized skills are being used, which are expensive and since higher time duration is involved thus amount at stake is also higher. The cost of implementation is a account specify cost thus has to be recovered from the customer at minimum commitment:
a. Cost of Implementation - Or cost plus minimal profits
b. Finance Cost : Since the money is invested upfront and recovery is made over time, thus finance cost also needs to be recovered.
c. SLA/Contingency : This will depend on how the deal has been negotiated. If possible, vendor will try to recover even the possible SLA/contingency cost at minimum commitment level..
5. Run & Maintenance Cost:
Their are multiple kind of cost involved under this head. This will include software ATS, hardware AMC, and also the cost of infra and apps FTEs managing the infrastructure and applications. This cost will be yearly cost and will not be incurred in case of premature termination. However, this is fixed for a year i.e. AMC/ATS paid at the beginning of the year to be recovered. Thus following cost to be recovered in from customer at minimum commitment:
a. Cost of AMC/ATS paid - Or cost plus minimal profits
b. Finance Cost : Since the money is invested upfront and recovery is made over time, thus finance cost also needs to be recovered.
c. SLA/Contingency : This will depend on how the deal has been negotiated. If possible, vendor will try to recover even the possible SLA/contingency cost at minimum commitment level..
6. SLA penalty :
The SLAs in case of private cloud will be computed on multiple parameters, which will be same as in case of any other project. However, the penalty will be levied on the monthly rate being charged, since that is only finance number being present in entire deal. This increase the risk and penalty cost multiple times. Since a network switch failure will lead to penalty being levied on entire monthly payment, which also includes recovery towards hardware and software. In a way penalty recovery for small components can be much higher than the business impact and cost of component/service being impacted. The vendor will have to necessarily have to cover his risk by ensuring penalty ceiling.
Owing to above factor, their is very limited window of benefit to customer. This model will also have very little upward or downward flexibility, since the entire infra is for a single customer, any increment beyond under utilized capacity will be charged extra and nothing and any downward movement will be prevented by minimum commitment, at which vendor is recovering its cost.
This model will also not have a free exit option. Since, vendor has incurred cost for specific to customer, he will have to recover the same, either at the duration of contract or at the termination of contract.
To summarize, though public cloud is a good concept and works for both vendor and customer, but private cloud is only a fancy financing arrangement, which works for none on following basis :
1. Customer has to pay all vendor cost as minimum commit
2. Anything paid over and above minimum commit is a pure margin to vendor.
3. Both the party gains or loses bases on other party,
a. if contract operates at minimum commit, the vendor do not make any margin,
b. if contract operates at Maximum capacity, the customer pays huge margins to vendor
c. if contract operates at projected capacity, vendor makes neutral margin and customer also do not overpay, however, then their is no benefit of the model.
4. In case of premature termination, customer pays entire cost of hardware, without any lien on the same.
Thus on above account, the private cloud model is more of a financial engineering than a technological engineering.
No comments:
Post a Comment