While public cloud storage has benefits, such as pay-as-you-go pricing and unlimited growth, it also has its downsides. One often overlooked area is the cost of switching cloud storage service providers. In most cases, the single biggest challenge is the amount of data to be migrated.
Data migration is a particular problem because a cloud storage service is often seen as the ultimate data archive and the capacities an organization will try to transfer can be very large -- petabytes instead of terabytes. Migrating large data sets is always a challenge, but public clouds have the additional issues of limited bandwidth and potential download fees.
Why switch cloud storage service providers?
Businesses tend to switch cloud providers for one of two reasons. First, there may be a pricing advantage to moving to an alternative provider. However, the cost savings need to be substantial enough to justify the work it takes to make the switch. During the transition period, an organization's cloud investment costs will actually go up, since both services will need to be active while data is migrated. In addition, most organizations will keep the old services online for a period of time until they can confirm all the data has been migrated and is accessible.
The second motivation to switch can be more dramatic. It's when a cloud provider suddenly announces it's going out of business or will no longer provide cloud storage. We have, on record, one case of a provider that went out of business and didn’t charge companies a download fee to get their data back, but there was the added pressure of completing the migration in as little as two weeks.
The public cloud problem
Most public cloud storage service providers are open, meaning it can be just as easy to copy data out of their storage infrastructures as it can be to migrate data into their environments. The main challenge is the bandwidth available to complete that migration. While it may have taken years to build the data archive, the cost or potential loss of data makes migrating rapidly a top priority. There's also the added cost of potential fees, as many providers charge extra for downloads.
Organizations typically move data to the cloud via extensions to their applications or a gateway appliance that makes the cloud provider's storage look more like traditional storage (NFS, SMB or iSCSI). This means the process needs to be reversed when moving to another provider. To do this, the application or appliance must first recall all the data from the old provider into the organization's data center, and then copy it to the new provider's storage. In this instance, the available bandwidth impacts the migration process two times.
In addition, the organization needs to have enough on-premises storage to cache the data while waiting for the upload to the new provider to be completed. This storage area may need to be quite large since download speeds are typically much faster than upload speeds, and reads are much faster than writes.
Potential roadblock workarounds
If the data the organization is storing in the cloud is in the low terabyte range, the above process may work just fine. Some bandwidth contracts even allow for relatively inexpensive short-term bursts to a faster speed. But as data capacities begin to reach the multi-petabyte level, the time it takes to do the multi-hop transfer will exceed the available time allowed to complete the transfer.
The first workaround is to see if your original cloud provider and the new cloud provider have a direct-connect type of relationship. This brings two benefits:
- The bandwidth of a provider's data center is typically much higher and eliminates the customer's data center as a go-between. A direct-connect relationship allows for a much faster transfer and won’t require the purchase of "cache" storage in the on-premises data center.
- If both providers have the ability to transport data via disk or tape, the original provider can copy everything to portable storage and then ship it to the new provider via a carrier. Cloud migration is an ideal use case for tape because it has a much higher capacity per piece of media and is better designed for transport. It also better integrates encryption technology so the data transported is secure. Unfortunately, most cloud providers have an aversion to tape, so it's not a common option.
Cloud migrations will become more common as the market consolidates and pricing advantages become more appealing. Concerns about a provider potentially going out of business can be eliminated by leveraging two providers and ensuring data is copied to both of them. This increases costs, but provides the ultimate protection against vendor failure -- and it still may be cheaper than maintaining the data on-premises.
Migrating to a new provider to save money requires spending additional money first. While the urgency to complete the migration is reduced, the longer it takes to complete the transfer, the longer the organization will pay double the cloud provider fees. In both cases, organizations should look to alternatives to the brute-force method of copying from the original provider to on-premises storage and then from on-premises storage to the new provider. Direct connections or transportable storage are ideal ways to avoid this.
Cloud providers continue to drop prices
Why cloud storage data access is so complicated