Is moving data storage from on-prem to public cloud always cost effective?

2.2k views18 Comments

Director of IT in Software, 201 - 500 employees
Only if appropriately designed and the right storage is used for the right data type.

Likely you don't need all your data all the time, so it's important to put archival data on a different storage type than daily accessed data. Aside from how often you access the data, it's important how many transactions on the data you perform. If you are copying/moving the data periodically, it's important to have it within the same region for the egress cost.
I've put terabytes of data into a storage account in the cloud and have chosen the cheaper type per GB, only to find that thousands of transactions are done on that data daily. The total monthly cost of all the transactions far exceeded the price per GB, so by switching it to a storage account that is more expensive per GB but cheaper per transaction, the overall cost significantly decreased.

For example, recently, I have reduced our data cost for backups by 85% by moving it from one object storage vendor to another. It's all about what makes sense for what type of data. Don't restrict yourself to using one type of storage or one vendor; explore and do the math on what makes sense for your situation.
Director in Manufacturing, 1,001 - 5,000 employees
Not if you don’t review performance requirements. In my experience if left to engineers and architects without guidance they over provision which is more expensive. I would do the same if I was only accountable for performance

It’s critical you tie performance and cost together with the decision process

Initially we let engineers just size environments and they consumed the entire years budget by end of February.

Key is to have mangers involved in provisioning that have both a technical background and understanding of budgeting and costs
CTO in Healthcare and Biotech, 2 - 10 employees
Things to think about: scale, how much data ingress and egress, accessibility, security, and management. Once you evaluate all of those and cloud offerings to support your data, you should be able to make the call on whether or not it’s cost effective. For example if you have a ton of data ingress and egress, you might not want to store all of your data in S3 because it will be spendy.
VP of Engineering in Software, 11 - 50 employees
It has to be correctly planned and executed. It also depends how much data needs to be moved and the cost of real internal IT.
CTO in Software, 51 - 200 employees
The answer is always it depends.

We found that it's very effective for use cases such as:
- Long-term Archival storage. The cost-saving over offsite backup and operational costs are significant. This is especially if you are committing to a long-term pricing model.
- Large-scale analytical data processing. This is close to, if not impossible to do so with an on-premise setup.

Another tip I always keep in mind is that: Storage is cheap; it's actually the compute expense, and labour those are expensive.
Solutions Architect in Software, 501 - 1,000 employees
In general - yes it is. There is a chance that on-prem will be more efficient, but that's usually an edge case. Just considering DevOps cost to support and troubleshoot cloud is a preferable option. However, the appropriate cloud solution should be picked.
Chief Technology Officer in Healthcare and Biotech, 1,001 - 5,000 employees
As others have said “it depends.”

The answer isn’t an unqualified yes or no, and while the cloud offers tremendous performance and scalability it can also easily see you blow out costs if not managed and designed well.

You need to consider storage volume, what will access your storage and how often - are those apps in the same cloud and region as the data? How frequently will you access the data - does it all need to be “warm”? There are a lot of issues so the answer to your question is not a blanket one.
Sr. Director of Engineering in Software, 51 - 200 employees
its always cost-effective to put old data on data warehouse storage. This ensures its availability and recovery as well. The choice of data store is important though and will optimise its use thereafter in terms of cost and usage. 
Earlier CIO in Manufacturing, Self-employed
Whether cost effective or not depends on your environment.  Points to consider are
a.  Are all the users located only in one place or distributed across the country / world ?  The cost of connectivity along with redundant connections will have to be also taken into account.

b.  Is the data production critical ?  If there is cut in the connectivity to access the data will operations in your organisation stop ?  So the cost of stoppage of business operations will have to be considered
c.  Are there any regulatory requirements to keep the data on premise or within the country ?
d.  How sensitive is the data that any access by any other third party will have serious consequences ?
e.  Extra security precautions that have to be taken when kept on cloud.  Those costs have to be also added
Senior Vice President, Engineering in Software, 1,001 - 5,000 employees
It purely depends on the amount of data and the archival strategy applied. In the case of a small amount of data it probably isn't much beneficial, but if the size is large and you are willing to let go of immediate availability of data and put it to something like AWS Glacier, then yes there is definite cost-effectiveness applicable.

Content you might like

Scalable AI49%

Composable Data and Analytics40%


Data Fabric35%

Engineering Decision Intelligence25%

Augmented Consumers6%

Edge Computing26%


1.1k views1 Upvote

CTO in Software, 201 - 500 employees
Without a doubt - Technical Debt! It's a ball and chain that creates an ever increasing drag on any organization, stifles innovation, and prevents transformation.
Read More Comments
46.5k views133 Upvotes324 Comments

We have already deployed/implemented data-driven projects34%

We are in the process of implementing or pilot testing data-driven projects45%

We have plans to deploy or implement data-driven projects within the next 12 months16%

We have no plans to deploy or implement data-driven projects5%


3.2k views2 Upvotes