Object storage is quickly becoming the standard for storing large amounts of unstructured data. One of the most appealing aspects of object storage is its flexibility that spans a variety of storage infrastructure configurations and storage locations. This flexibility is leading vendors to provide a wide variety of consumption models including on-premises turnkey solutions, do-it-yourself solutions where the organization provides the hardware and software, a variety of off-premises consumption models and hybrid solutions that leverage both off-premises and on-premises data centers.
Do-it-Yourself On-Premises Solutions
Most if not all object storage solutions are software defined, which means the bulk of its intellectual property is in software and it is designed to run on commodity hardware. A do-it-yourself solution means the organization will leverage the shelf software and have that software run on hardware it provides. This hardware can either be re-purposed existing hardware or can be bought new specifically for the task. The customer is then responsible for installing the software onto the hardware. There is also a networking concern since most object storage systems are scale-out, where all of the nodes must be clustered together in order to operate efficiently.
The challenge with the do-it-yourself approach is that the customer must play the role of integrator, which means not only installing the hardware and software but also making sure that the different components of the solution like storage media are compatible. While not out of reach of most IT professionals skill set, it may be out of the reach of their available time.
Turnkey On-Premises Solutions
An alternative to the do-it-yourself model is a turnkey on-premises solution. In this consumption model, the organization elects to get all the hardware and software from a single vendor. In most cases the vendor pre-installs the software on that hardware. The result should be much faster implementation times and less concern about hardware/software compatibility. In many cases the vendor will also perform the initial installation and configuration.
The concern with the turnkey approach is reduced flexibility and the potential for “vendor lock-in.” In some cases the vendor will not allow the integration of nodes the vendor does not supply. The limitation of using only-vendor supplied hardware means the customer may have to wait for certification of new technology and may have to pay more for the hardware. In most cases though vendors are very quick to certify new hardware and while the turnkey hardware is more expensive than do-it-yourself configurations, once the software and time to install are factored in the cost delta shrinks dramatically.
There are also off-premises solutions, or pure cloud solutions to consider, that also can be do-it-yourself or turnkey. The organization can either install the software onto compute servers and storage they are renting from the provider or they can get a turnkey service. The advantages of a pure cloud-based solution are the pay as you go consumption model and the savings associated with not requiring on-premises floor space.
The concerns about pure cloud-based storage are similar to almost any other cloud-based initiative, security concerns and long term costs. With object storage though these concerns are compounded since we may be dealing with petabytes of data that needs to be stored for a very long time.
The final consumption model is a hybrid model, where a portion of the data stays on-premises and a cloud provider stores another portion. The deployment considerations are the same depending on what type of consumption model the organization selects in each location. The other decision is how to use each location. The organization could design an architecture that stores the most active data on-premises and keeps the inactive data (or cold data) off-premises. Again, the challenge with this approach is long-terms cost. Alternatively, the organization could design the opposite configuration, where active data is stored off-premises and inactive data is stored on-premises. The provider would then use cloud compute resources to process data. A final option is to keep most of the data on-premises and use the pure cloud storage as a temporary burst location when the organization runs out of capacity unexpectedly.
As the flexibility of deployment options becomes a key consideration from customers, many object storage vendors offer multiple consumption models. A solution like Dell EMC Elastic Cloud Storage (ECS) offers 4 deployment options – as a turnkey storage appliance, as a software-only solution, through public cloud solutions via Virtustream, or as a dedicated cloud hosted solution, allowing customers to deploy this solution on their own terms.
While compute and security are often a consideration, which of these consumption models is best for your organization depends largely on the amount of time and budget available to it. Compute proximity is important for many object storage use cases but unless the processing application can’t be cloud hosted, compute can be moved to wherever the data is. Security is always a challenge and there is no real evidence that the average data center has better security controls than the top cloud providers.
Sponsored by Dell EMC