If you read Will’s blog on his experience in moving his firm’s hardware platform to the cloud for Disaster Recovery (DR), you will understand there are a lot of considerations to make as well as determining if the cost is justifiable vs. in-house and the difficulty in just flat flushing out all the players in this space is daunting at best. Hopefully we can give you a bit of a heads up on what I consider to be the ‘gotchas’ of signing on the dotted line when you are looking to go down the road of Infrastructure as a Service (IaaS).
So how much is this going to cost?
Now that being said it is obvious to anyone that if you are paying for a bundled service, only when you need it, more likely than not, will save you money. A prime example is using IaaS as a DR component in your Business Continuity (BC) or DR plan. This is the direction that Smith Debman took as Will described in his move to an IaaS platform. Basically we are talking about infrastructure only needed during a disaster and can be available when needed, therefore not paying that high cost of compute for infrastructure when it is sitting idle. The average cost to run a VM in the cloud is approximately 2 cents per hour. However this does not include backups and redundancy or managed services costs, if needed.
Most applications are tied to some sort of database and this is when the cost of running that application could mean big bucks. It may be that database is going to require a large amount of redevelopment to run efficiently in the cloud where it would be best to run on-prem. Understanding what it is going to cost, you need to consider application costs in an on-prem or cloud based system.
Let’s face it some applications are just built less efficient and are going to use an indecent amount of resources and that means your costs will be high. We have all spent a lot of money ensuring critical workloads remain functioning in the event of a failure. We invest in redundant systems to meet peek loads. Not surprising workloads designed for one Infrastructure paradigm often do not translate easily to the other. What this means is when we design an application to be as resilient as possible and you apply that same concept to a cloud based solution, you will indeed pay a lot more for that cloud offering.
There are things to look at when comparing the cost of on-prem vs. cloud. One overlooked cost is facilities and power. These play a factor when comparing cost based on the scale of your infrastructure. Other factors that assuredly should be weighed is the cost of IT Staff and the ability to take advantage of next-generation hardware which typically IaaS providers are built upon.
Service Level Agreements (SLA)
The bottom line on Service level agreements is to typically protect the provider. Practical Guide to Cloud Service Agreements Version 2.0 I say that because most claim a 99.9999% uptime when discussing with the provider about putting your hardware in the cloud, but let’s face it they typically come with a slew of exceptions and very limited guaranties of monetary backed compensation when these SLA’s are not met, this is usually due to the possibility of high penalties and major losses to the provider.
Cloud providers vary as to the responsibility of downtime. An example of this, Microsoft believes that if you are using Azure for compute , the major responsibility of maximum performance and uptime belongs to Microsoft where as other providers clearly protect themselves, with, in my opinion, unreasonable SLA’s that protect the provider from large monetary penalties. These penalties are usually in the form of credits and not in the form of credit in the bill you pay every month.
Be cognizant of encryption keys if they are used in your cloud environment. Make sure you own the keys and have them stored in a key vault to protect your data.
So be careful before you sign that contract and make sure the firm is protected as best as you can based on the agreement you are given by the provider as these can be negotiated , especially with the smaller providers. Also smaller providers seem to have better uptime than the larger ones. It is kind logical thinking when you say that; “the bigger the structure, the more centralized control, the bigger the failure.”
Final Thought
The cost of moving your infrastructure to the cloud should be determined based on your application workload. Will discovered that the cost for a DR solution was a savings to the firm vs. getting in a Co-location facility and buying additional hardware and services. What You Should Know Before Buying IaaS Try to calculate your costs on a per VM basis and maybe that will guide you on the individual components of an application that will add costs over an on-prem solution. There are also programs offered by cloud providers where they sale you spare compute capacity at a discounted price. Be sure to have as good an SLA as you can to protect your Firm’s important assets and ensure all your questions are answered before you sign on the bottom line.
Will Pulsifer's Blog - How we did it. One firms jump to iaaS.
#ITOperations