The Great Debate – Public Cloud vs. Private Cloud Storage

A recent Storage Switzerland webinar sparked an online discussion on the idea of using cloud storage services coupled with a cloud caching appliance. As with all online discussions, there were some naysayers and some supporters. I enjoy debating anything – including politics and religion – and the use of the cloud is an important thing to debate.

I do believe that public cloud services like those provided by Amazon, Microsoft, and Google, can be less expensive than doing it yourself. This won’t be true for everyone, of course. The key to making public cloud services inexpensive is to understand the service levels available and to map your requirements to the appropriate service level. The key to making your own storage systems cheaper than the cloud is to use the same technology the cloud uses and in order to get rid of inefficiencies.

Most major cloud vendors have different service levels with different prices. Specifically in cloud storage services there is online storage, nearline storage, and cold storage. Some services like Amazon Glacier offer reduced monthly charges in exchange for an increased access time during retrievals. In contrast, all Google cloud storage services have the same access time for all levels but cost differently to retrieve the data.

The point is that you need to think about how and when you are going to access the data, because placing the data on the wrong tier of storage could be a waste of money on the front-end or the backend. For example, if you are never going to retrieve a particular set of data, paying to store that on Google’s multiregional service is a significant waste of money every single month. However, if you have a data set that you are going to regularly retrieve, storing it on Google Coldline could end up costing you a significant amount of money on the backend with retrieval costs. You could also end up paying storage costs for data that you no longer need if you put it on the wrong tier, since some of the less expensive tiers have minimum retention requirements of 90 days or more.

Another key to making the cloud inexpensive is to minimize the level of effort needed to put data in the cloud. This is why cloud caching appliances are so popular. They provide a quick and easy way to store data on-site and off-site using the cloud and are much easier and less expensive to manage than a typical filer.

But what if you put the data on your own inexpensive object storage system? Would that be less expensive than the cloud? It could be, but the biggest challenges you will fight when doing storage yourself is inefficiency. In the cloud, you only pay for the gigabytes you use as you use them. In the data center, you are going to pay for every spinning disk drive or flash drive that you have whether there is data on it or not. With typical efficiency percentages being well below 50 percent, this is going to significantly drive up your effective per gigabyte cost.

Fred Pinkett from Nasuni and I discussed these issues and more in our recent webinar, and it’s now available on demand. Check it out and join the discussion!

Watch On Demand

W. Curtis Preston (aka Mr. Backup) is an expert in backup & recovery systems; a space he has been working in since 1993. He has written three books on the subject, Backup & Recovery, Using SANs and NAS, and Unix Backup & Recovery. Mr. Preston is a writer and has spoken at hundreds of seminars and conferences around the world. Preston’s mission is to arm today’s IT managers with truly unbiased information about today’s storage industry and its products.

Tagged with: , , , , , ,
Posted in Blog
2 comments on “The Great Debate – Public Cloud vs. Private Cloud Storage
  1. Tim Wessels says:

    Well, Mr. Preston made some good points about the inappropriate use of public cloud storage. He loses some credibility by claiming private cloud storage is so inefficient that it will “significantly drive up your effective per gigabyte cost.” If Mr. Preston is talking about inefficiency in disk drive storage utilization, he needs to provide the examples. If he is claiming that private storage clouds are inefficient because they use additional storage for replicating or erasure coding data, then public cloud storage providers are in the same boat.

    Private storage cloud software vendors base their pricing on usable storage in the cluster. For example, a customer who needs to protect 100TB of data can do the math to determine the initial number of storage servers needed in the cluster. Growth in data storage over time will require adding storage servers to the cluster. A private storage cloud doesn’t need to be significantly larger than the amount of data the customer wants to protect. Private storage cloud software continuously monitors how much storage capacity is remaining. Adding new storage servers to a cluster is easy to do and it won’t interrupt the availability of the cluster.

    Everyone likes to toss around cost numbers when it comes to cloud storage. The public providers tout their low storage cost as long as you don’t touch it or move it. Private storage vendors are more interested in showing you the TCO over some period of time like five years, and not just the acquisition cost. You really need to compare public vs. private cloud storage over time to get a better understanding of your costs.

  2. wcurtispreston says:

    You used a snippet of a sentence to make me say something I didn’t say. I never said that private cloud storage is “so inefficient that it will “significantly drive up your effective per gigabyte cost.”” I said that typical utilization is 50% or less (and I believe I’m being generous) and so you have to calculate all the storage you own, not just the storage you are using.

    I never said that public cloud storage was inefficient. The inefficiency comes from the purchasing process, and I believe my 23 years of experience in IT gives me some insight into that process. I have never met a company that buys just the amount of storage they need today. They buy the amount they will need for the next n months or years, based on how difficult it is to expand that storage when they run out of capacity. While it is true that it’s really easy to add incremental capacity to most private cloud products (maybe one that your company could design and build), few companies have a capital purchasing cycle that allows them to quickly buy anything. So they tend to buy 500 TB when they know they only need 200 right now. Then they buy another 250 when they get anywhere near 500, etc. The result is that when you own your own storage — even if it’s the best private cloud — you are probably going to have a lot more storage than you are actually using. This drives down your efficiency and drives up your effective cost per GB. In my experience in the real world, efficiencies are far less than 50%, so I thought I was being generous. Again, it’s not the fault of the private cloud; it’s the fault of the buying process. Contrast this to the public cloud where you only pay for the gigabytes you actually use.

    You also alluded to the cost of additional copies and suggested that these costs would be the same in the public cloud. That is not the case. With a private cloud I must include the cost of all storage at all sites (including any unused storage I bought due to what I mentioned above) and the cost of bandwidth between the sites. With a decent public cloud vendor, I get all that for one price. They will put my data in 2-3 geographically different locations as part of the per GB price — assuming I choose that service level. So we just need to make sure we’re comparing like to like. Standard S3 (that includes multiple geos) pricing + cost of bandwidth to get to S3 vs multiple private cloud systems and the cost of bandwidth to go beween locations. Private cloud will require more bandwidth. With public cloud, I send it out once. With private cloud, I need bandwidth to site 1, then from site 1 to site 2, then from site 2 to site 3. Three times more bandwidth required for geographically separate storage.

    I actually like the private cloud idea and think it has a lot of advantages. I also think that it’s possible that the per GB cost could actually less depending on the vendor and how they price things. I’m aware, for example, of some private cloud storage products that have utilization pricing, where you essentially pay for them the same as you would for Amazon or Google. I just want to make sure that when we’re comparing two very different pricing structures, we have to make sure we consider all of the cost factors, which includes efficiency and bandwidth.

    FWIW I also think it’s appropriate for me to say this. While I would never say your argument is invalid because you work for a company that makes private cloud storage, I do think it’s relevant. Typically when someone who works for a vendor comments on our blogs, they will say so in their response. It’s called transparency. Just like you can see above that Nasuni sponsored this blog. You might consider doing that in the future.

Comments are closed.

Enter your email address to follow this blog and receive notifications of new posts by email.

Join 22,035 other followers

Blog Stats
%d bloggers like this: